{"id": "arxiv:math/9404236", "slug": "42371afbf72fe8b6", "title": "On proof and progress in mathematics", "authors": ["William P. Thurston"], "year": 1994, "venue": "arXiv", "url": "https://arxiv.org/abs/math/9404236", "doi": null, "topics": ["Mathematics"], "description": "In response to Jaffe and Quinn [math.HO/9307227], the author discusses forms of progress in mathematics that are not captured by formal proofs of theorems, especially in his own work in the theory of foliations and geometrization of 3-manifolds and dynamical systems.", "description_method": "abstract-excerpt", "abstract": "In response to Jaffe and Quinn [math.HO/9307227], the author discusses forms of progress in mathematics that are not captured by formal proofs of theorems, especially in his own work in the theory of foliations and geometrization of 3-manifolds and dynamical systems.", "score": 9.0, "mention_count": 29, "user_count": 25, "months": ["2009-12", "2012-08", "2013-08", "2013-10", "2014-01", "2014-09", "2015-03", "2015-07", "2016-05", "2017-03", "2017-10", "2020-06", "2020-08", "2021-07", "2023-06", "2023-07", "2023-11", "2024-12", "2025-08", "2026-03", "2026-08"], "last_mentioned": "2026-08-19T21:08:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1712.01208", "slug": "3b321c5bdf5b1142", "title": "The Case for Learned Index Structures", "authors": ["Tim Kraska", "Alex Beutel", "Ed H. Chi", "Jeffrey Dean", "Neoklis Polyzotis"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1712.01208", "doi": null, "topics": ["Databases"], "description": "Indexes are models: a B-Tree-Index can be seen as a model to map a key to the position of a record within a sorted array, a Hash-Index as a model to map a key to a position of a record within an unsorted array, and a BitMap-Index as a model to indicate if a data record exists or not.", "description_method": "abstract-excerpt", "abstract": "Indexes are models: a B-Tree-Index can be seen as a model to map a key to the position of a record within a sorted array, a Hash-Index as a model to map a key to a position of a record within an unsorted array, and a BitMap-Index as a model to indicate if a data record exists or not. In this exploratory research paper, we start from this premise and posit that all existing index structures can be replaced with other types of models, including deep-learning models, which we term learned indexes. The key idea is that a model can learn the sort order or structure of lookup keys and use this signal to effectively predict the position or existence of records. We theoretically analyze under which conditions learned indexes outperform traditional index structures and describe the main challenges in designing learned index structures. Our initial results show, that by using neural nets we are able to outperform cache-optimized B-Trees by up to 70% in speed while saving an order-of-magnitude in memory over several real-world data sets. More importantly though, we believe that the idea of replacing core components of a data management system through learned models has far reaching implications for future systems designs and that this work just provides a glimpse of what might be possible.", "score": 5.0, "mention_count": 19, "user_count": 15, "months": ["2017-12", "2018-02", "2018-05", "2018-10", "2018-11", "2019-08", "2020-07", "2020-10", "2021-01", "2021-07", "2022-07", "2025-06", "2026-01", "2026-06"], "last_mentioned": "2026-06-04T15:02:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2303.12712", "slug": "f62e417e60d72f6b", "title": "Sparks of Artificial General Intelligence: Early experiments with GPT-4", "authors": ["Sébastien Bubeck", "Varun Chandrasekaran", "Ronen Eldan", "Johannes Gehrke", "Eric Horvitz", "Ece Kamar", "Peter Lee", "Yin Tat Lee", "Yuanzhi Li", "Scott Lundberg", "Harsha Nori", "Hamid Palangi", "Marco Tulio Ribeiro", "Yi Zhang"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2303.12712", "doi": null, "topics": ["Language models", "Psychology", "Mathematics"], "description": "Artificial intelligence (AI) researchers have been developing and refining large language models (LLMs) that exhibit remarkable capabilities across a variety of domains and tasks, challenging our understanding of learning and cognition.", "description_method": "abstract-excerpt", "abstract": "Artificial intelligence (AI) researchers have been developing and refining large language models (LLMs) that exhibit remarkable capabilities across a variety of domains and tasks, challenging our understanding of learning and cognition. The latest model developed by OpenAI, GPT-4, was trained using an unprecedented scale of compute and data. In this paper, we report on our investigation of an early version of GPT-4, when it was still in active development by OpenAI. We contend that (this early version of) GPT-4 is part of a new cohort of LLMs (along with ChatGPT and Google's PaLM for example) that exhibit more general intelligence than previous AI models. We discuss the rising capabilities and implications of these models. We demonstrate that, beyond its mastery of language, GPT-4 can solve novel and difficult tasks that span mathematics, coding, vision, medicine, law, psychology and more, without needing any special prompting. Moreover, in all of these tasks, GPT-4's performance is strikingly close to human-level performance, and often vastly surpasses prior models such as ChatGPT. Given the breadth and depth of GPT-4's capabilities, we believe that it could reasonably be viewed as an early (yet still incomplete) version of an artificial general intelligence (AGI) system. In our exploration of GPT-4, we put special emphasis on discovering its limitations, and we discuss the challenges ahead for advancing towards deeper and more comprehensive versions of AGI, including the possible need for pursuing a new paradigm that moves beyond next-word prediction. We conclude with reflections on societal influences of the recent technological leap and future research directions.", "score": 3.5, "mention_count": 60, "user_count": 51, "months": ["2023-03", "2023-04", "2023-05", "2023-07", "2023-08", "2023-11", "2024-06", "2024-07", "2025-03", "2025-04", "2025-08", "2026-02", "2026-05", "2026-08"], "last_mentioned": "2026-08-03T01:47:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1509.05393", "slug": "4458d3c689abb25f", "title": "A Critique of the CAP Theorem", "authors": ["Martin Kleppmann"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1509.05393", "doi": null, "topics": ["Mathematics", "Distributed systems", "Databases"], "description": "The CAP Theorem is a frequently cited impossibility result in distributed systems, especially among NoSQL distributed databases.", "description_method": "abstract-excerpt", "abstract": "The CAP Theorem is a frequently cited impossibility result in distributed systems, especially among NoSQL distributed databases. In this paper we survey some of the confusion about the meaning of CAP, including inconsistencies and ambiguities in its definitions, and we highlight some problems in its formalization. CAP is often interpreted as proof that eventually consistent databases have better availability properties than strongly consistent databases; although there is some truth in this, we show that more careful reasoning is required. These problems cast doubt on the utility of CAP as a tool for reasoning about trade-offs in practical systems. As alternative to CAP, we propose a \"delay-sensitivity\" framework, which analyzes the sensitivity of operation latency to network delay, and which may help practitioners reason about the trade-offs between consistency guarantees and tolerance of network faults.", "score": 3.5, "mention_count": 6, "user_count": 6, "months": ["2015-09", "2015-10", "2017-06", "2021-09", "2024-01", "2024-10"], "last_mentioned": "2024-10-08T12:38:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2005.14165", "slug": "49ea3576b0f1a44a", "title": "Language Models are Few-Shot Learners", "authors": ["Tom B. Brown", "Benjamin Mann", "Nick Ryder", "Melanie Subbiah", "Jared Kaplan", "Prafulla Dhariwal", "Arvind Neelakantan", "Pranav Shyam", "Girish Sastry", "Amanda Askell", "Sandhini Agarwal", "Ariel Herbert-Voss", "Gretchen Krueger", "Tom Henighan", "Rewon Child", "Aditya Ramesh", "Daniel M. Ziegler", "Jeffrey Wu", "Clemens Winter", "Christopher Hesse", "Mark Chen", "Eric Sigler", "Mateusz Litwin", "Scott Gray", "Benjamin Chess", "Jack Clark", "Christopher Berner", "Sam McCandlish", "Alec Radford", "Ilya Sutskever", "Dario Amodei"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2005.14165", "doi": null, "topics": ["Language models", "Machine learning"], "description": "Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task.", "description_method": "abstract-excerpt", "abstract": "Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task. While typically task-agnostic in architecture, this method still requires task-specific fine-tuning datasets of thousands or tens of thousands of examples. By contrast, humans can generally perform a new language task from only a few examples or from simple instructions - something which current NLP systems still largely struggle to do. Here we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches. Specifically, we train GPT-3, an autoregressive language model with 175 billion parameters, 10x more than any previous non-sparse language model, and test its performance in the few-shot setting. For all tasks, GPT-3 is applied without any gradient updates or fine-tuning, with tasks and few-shot demonstrations specified purely via text interaction with the model. GPT-3 achieves strong performance on many NLP datasets, including translation, question-answering, and cloze tasks, as well as several tasks that require on-the-fly reasoning or domain adaptation, such as unscrambling words, using a novel word in a sentence, or performing 3-digit arithmetic. At the same time, we also identify some datasets where GPT-3's few-shot learning still struggles, as well as some datasets where GPT-3 faces methodological issues related to training on large web corpora. Finally, we find that GPT-3 can generate samples of news articles which human evaluators have difficulty distinguishing from articles written by humans. We discuss broader societal impacts of this finding and of GPT-3 in general.", "score": 3.0, "mention_count": 83, "user_count": 71, "months": ["2020-05", "2020-06", "2020-07", "2020-08", "2020-09", "2020-10", "2020-12", "2021-01", "2021-07", "2021-08", "2021-09", "2021-10", "2021-11", "2022-02", "2022-10", "2022-12", "2023-01", "2023-02", "2023-03", "2023-04", "2023-05", "2023-06", "2023-07", "2023-09", "2023-10", "2023-11", "2023-12", "2024-03", "2024-10", "2025-05", "2025-08", "2025-09", "2025-12", "2026-03", "2026-06"], "last_mentioned": "2026-06-06T06:55:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1806.02404", "slug": "64bfe2e8a9a13048", "title": "Dissolving the Fermi Paradox", "authors": ["Anders Sandberg", "Eric Drexler", "Toby Ord"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1806.02404", "doi": null, "topics": ["Statistics & probability", "Science & research"], "description": "The Fermi paradox is the conflict between an expectation of a high {\\em ex ante} probability of intelligent life elsewhere in the universe and the apparently lifeless universe we in fact observe.", "description_method": "abstract-excerpt", "abstract": "The Fermi paradox is the conflict between an expectation of a high {\\em ex ante} probability of intelligent life elsewhere in the universe and the apparently lifeless universe we in fact observe. The expectation that the universe should be teeming with intelligent life is linked to models like the Drake equation, which suggest that even if the probability of intelligent life developing at a given site is small, the sheer multitude of possible sites should nonetheless yield a large number of potentially observable civilizations. We show that this conflict arises from the use of Drake-like equations, which implicitly assume certainty regarding highly uncertain parameters. We examine these parameters, incorporating models of chemical and genetic transitions on paths to the origin of life, and show that extant scientific knowledge corresponds to uncertainties that span multiple orders of magnitude. This makes a stark difference. When the model is recast to represent realistic distributions of uncertainty, we find a substantial {\\em ex ante} probability of there being no other intelligent life in our observable universe, and thus that there should be little surprise when we fail to detect any signs of it. This result dissolves the Fermi paradox, and in doing so removes any need to invoke speculative mechanisms by which civilizations would inevitably fail to have observable effects upon the universe.", "score": 3.0, "mention_count": 35, "user_count": 27, "months": ["2018-06", "2018-07", "2018-08", "2019-03", "2019-06", "2019-09", "2019-10", "2020-05", "2020-09", "2020-12", "2021-01", "2021-03", "2021-05", "2021-07", "2021-12", "2022-01", "2022-10", "2022-12", "2023-02", "2023-07", "2025-02", "2025-04", "2025-05", "2025-10"], "last_mentioned": "2025-10-04T21:07:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1205.5935", "slug": "9f7dc845f4f078f5", "title": "Geometric Algebra", "authors": ["Eric Chisolm"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1205.5935", "doi": null, "topics": ["Mathematics"], "description": "This is an introduction to geometric algebra, an alternative to traditional vector algebra that expands on it in two ways: 1.", "description_method": "abstract-excerpt", "abstract": "This is an introduction to geometric algebra, an alternative to traditional vector algebra that expands on it in two ways: 1. In addition to scalars and vectors, it defines new objects representing subspaces of any dimension. 2. It defines a product that's strongly motivated by geometry and can be taken between any two objects. For example, the product of two vectors taken in a certain way represents their common plane. This system was invented by William Clifford and is more commonly known as Clifford algebra. It's actually older than the vector algebra that we use today (due to Gibbs) and includes it as a subset. Over the years, various parts of Clifford algebra have been reinvented independently by many people who found they needed it, often not realizing that all those parts belonged in one system. This suggests that Clifford had the right idea, and that geometric algebra, not the reduced version we use today, deserves to be the standard \"vector algebra.\" My goal in these notes is to describe geometric algebra from that standpoint and illustrate its usefulness. The notes are work in progress; I'll keep adding new topics as I learn them myself.", "score": 3.0, "mention_count": 11, "user_count": 6, "months": ["2016-07", "2016-11", "2018-11", "2019-04", "2020-09", "2020-11", "2020-12", "2021-11", "2022-03", "2024-08"], "last_mentioned": "2024-08-07T14:59:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1611.03530", "slug": "d7496ba4c9eebcf6", "title": "Understanding deep learning requires rethinking generalization", "authors": ["Chiyuan Zhang", "Samy Bengio", "Moritz Hardt", "Benjamin Recht", "Oriol Vinyals"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1611.03530", "doi": null, "topics": ["Machine learning", "Computer vision", "Statistics & probability"], "description": "Despite their massive size, successful deep artificial neural networks can exhibit a remarkably small difference between training and test performance.", "description_method": "abstract-excerpt", "abstract": "Despite their massive size, successful deep artificial neural networks can exhibit a remarkably small difference between training and test performance. Conventional wisdom attributes small generalization error either to properties of the model family, or to the regularization techniques used during training. Through extensive systematic experiments, we show how these traditional approaches fail to explain why large neural networks generalize well in practice. Specifically, our experiments establish that state-of-the-art convolutional networks for image classification trained with stochastic gradient methods easily fit a random labeling of the training data. This phenomenon is qualitatively unaffected by explicit regularization, and occurs even if we replace the true images by completely unstructured random noise. We corroborate these experimental findings with a theoretical construction showing that simple depth two neural networks already have perfect finite sample expressivity as soon as the number of parameters exceeds the number of data points as it usually does in practice. We interpret our experimental findings by comparison with traditional models.", "score": 2.5, "mention_count": 16, "user_count": 15, "months": ["2016-12", "2017-01", "2017-05", "2017-08", "2017-12", "2018-01", "2018-07", "2018-10", "2018-11", "2020-02", "2020-10", "2022-05", "2024-05", "2025-03"], "last_mentioned": "2025-03-19T04:32:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1802.07068", "slug": "30dad9393a018ae5", "title": "Talent vs Luck: the role of randomness in success and failure", "authors": ["A. Pluchino. A. E. Biondo", "A. Rapisarda"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1802.07068", "doi": "10.1142/S0219525918500145", "topics": ["Psychology"], "description": "The largely dominant meritocratic paradigm of highly competitive Western cultures is rooted on the belief that success is due mainly, if not exclusively, to personal qualities such as talent, intelligence, skills, efforts or risk taking.", "description_method": "abstract-excerpt", "abstract": "The largely dominant meritocratic paradigm of highly competitive Western cultures is rooted on the belief that success is due mainly, if not exclusively, to personal qualities such as talent, intelligence, skills, efforts or risk taking. Sometimes, we are willing to admit that a certain degree of luck could also play a role in achieving significant material success. But, as a matter of fact, it is rather common to underestimate the importance of external forces in individual successful stories. It is very well known that intelligence or talent exhibit a Gaussian distribution among the population, whereas the distribution of wealth - considered a proxy of success - follows typically a power law (Pareto law). Such a discrepancy between a Normal distribution of inputs, with a typical scale, and the scale invariant distribution of outputs, suggests that some hidden ingredient is at work behind the scenes. In this paper, with the help of a very simple agent-based model, we suggest that such an ingredient is just randomness. In particular, we show that, if it is true that some degree of talent is necessary to be successful in life, almost never the most talented people reach the highest peaks of success, being overtaken by mediocre but sensibly luckier individuals. As to our knowledge, this counterintuitive result - although implicitly suggested between the lines in a vast literature - is quantified here for the first time. It sheds new light on the effectiveness of assessing merit on the basis of the reached level of success and underlines the risks of distributing excessive honors or resources to people who, at the end of the day, could have been simply luckier than others. With the help of this model, several policy hypotheses are also addressed and compared to show the most efficient strategies for public funding of research in order to improve meritocracy, diversity and innovation.", "score": 2.5, "mention_count": 16, "user_count": 13, "months": ["2018-03", "2018-05", "2018-10", "2019-10", "2020-02", "2020-07", "2020-09", "2021-06", "2022-09", "2022-12", "2023-10", "2024-07", "2026-06", "2026-07"], "last_mentioned": "2026-07-23T20:18:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2011.13127", "slug": "d9ce7ab17c76568f", "title": "Copy-and-Patch Compilation: A fast compilation algorithm for high-level languages and bytecode", "authors": ["Haoran Xu", "Fredrik Kjolstad"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2011.13127", "doi": "10.1145/3485513", "topics": ["Programming languages", "Databases", "Algorithms"], "description": "Fast compilation is important when compilation occurs at runtime, such as query compilers in modern database systems and WebAssembly virtual machines in modern browsers.", "description_method": "abstract-excerpt", "abstract": "Fast compilation is important when compilation occurs at runtime, such as query compilers in modern database systems and WebAssembly virtual machines in modern browsers. We present copy-and-patch, an extremely fast compilation technique that also produces good quality code. It is capable of lowering both high-level languages and low-level bytecode programs to binary code, by stitching together code from a large library of binary implementation variants. We call these binary implementations stencils because they have holes where missing values must be inserted during code generation. We show how to construct a stencil library and describe the copy-and-patch algorithm that generates optimized binary code. We demonstrate two use cases of copy-and-patch: a compiler for a high-level C-like language intended for metaprogramming and a compiler for WebAssembly. Our high-level language compiler has negligible compilation cost: it produces code from an AST in less time than it takes to construct the AST. We have implemented an SQL database query compiler on top of this metaprogramming system and show that on TPC-H database benchmarks, copy-and-patch generates code two orders of magnitude faster than LLVM -O0 and three orders of magnitude faster than higher optimization levels. The generated code runs an order of magnitude faster than interpretation and 14% faster than LLVM -O0. Our WebAssembly compiler generates code 4.9X-6.5X faster than Liftoff, the WebAssembly baseline compiler in Google Chrome. The generated code also outperforms Liftoff's by 39%-63% on the Coremark and PolyBenchC WebAssembly benchmarks.", "score": 2.5, "mention_count": 10, "user_count": 10, "months": ["2022-09", "2022-11", "2023-12", "2024-01", "2024-03", "2024-06", "2025-04", "2025-09", "2025-12"], "last_mentioned": "2025-12-26T21:28:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1306.0159", "slug": "4733c6df6cc163e8", "title": "The Ghost in the Quantum Turing Machine", "authors": ["Scott Aaronson"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1306.0159", "doi": null, "topics": ["Physics", "Neuroscience", "Statistics & probability"], "description": "In honor of Alan Turing's hundredth birthday, I unwisely set out some thoughts about one of Turing's obsessions throughout his life, the question of physics and free will.", "description_method": "abstract-excerpt", "abstract": "In honor of Alan Turing's hundredth birthday, I unwisely set out some thoughts about one of Turing's obsessions throughout his life, the question of physics and free will. I focus relatively narrowly on a notion that I call \"Knightian freedom\": a certain kind of in-principle physical unpredictability that goes beyond probabilistic unpredictability. Other, more metaphysical aspects of free will I regard as possibly outside the scope of science. I examine a viewpoint, suggested independently by Carl Hoefer, Cristi Stoica, and even Turing himself, that tries to find scope for \"freedom\" in the universe's boundary conditions rather than in the dynamical laws. Taking this viewpoint seriously leads to many interesting conceptual problems. I investigate how far one can go toward solving those problems, and along the way, encounter (among other things) the No-Cloning Theorem, the measurement problem, decoherence, chaos, the arrow of time, the holographic principle, Newcomb's paradox, Boltzmann brains, algorithmic information theory, and the Common Prior Assumption. I also compare the viewpoint explored here to the more radical speculations of Roger Penrose. The result of all this is an unusual perspective on time, quantum mechanics, and causation, of which I myself remain skeptical, but which has several appealing features. Among other things, it suggests interesting empirical questions in neuroscience, physics, and cosmology; and takes a millennia-old philosophical debate into some underexplored territory.", "score": 2.5, "mention_count": 9, "user_count": 7, "months": ["2015-04", "2016-04", "2016-06", "2017-09", "2019-07", "2023-03", "2024-04", "2024-07"], "last_mentioned": "2024-07-05T15:24:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1604.07316", "slug": "10173859cf156ebf", "title": "End to End Learning for Self-Driving Cars", "authors": ["Mariusz Bojarski", "Davide Del Testa", "Daniel Dworakowski", "Bernhard Firner", "Beat Flepp", "Prasoon Goyal", "Lawrence D. Jackel", "Mathew Monfort", "Urs Muller", "Jiakai Zhang", "Xin Zhang", "Jake Zhao", "Karol Zieba"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1604.07316", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "We trained a convolutional neural network (CNN) to map raw pixels from a single front-facing camera directly to steering commands.", "description_method": "abstract-excerpt", "abstract": "We trained a convolutional neural network (CNN) to map raw pixels from a single front-facing camera directly to steering commands. This end-to-end approach proved surprisingly powerful. With minimum training data from humans the system learns to drive in traffic on local roads with or without lane markings and on highways. It also operates in areas with unclear visual guidance such as in parking lots and on unpaved roads. The system automatically learns internal representations of the necessary processing steps such as detecting useful road features with only the human steering angle as the training signal. We never explicitly trained it to detect, for example, the outline of roads. Compared to explicit decomposition of the problem, such as lane marking detection, path planning, and control, our end-to-end system optimizes all processing steps simultaneously. We argue that this will eventually lead to better performance and smaller systems. Better performance will result because the internal components self-optimize to maximize overall system performance, instead of optimizing human-selected intermediate criteria, e.g., lane detection. Such criteria understandably are selected for ease of human interpretation which doesn't automatically guarantee maximum system performance. Smaller networks are possible because the system learns to solve the problem with the minimal number of processing steps. We used an NVIDIA DevBox and Torch 7 for training and an NVIDIA DRIVE(TM) PX self-driving car computer also running Torch 7 for determining where to drive. The system operates at 30 frames per second (FPS).", "score": 2.5, "mention_count": 8, "user_count": 5, "months": ["2016-07", "2017-02", "2017-05", "2017-06", "2017-09", "2018-05", "2018-07"], "last_mentioned": "2018-07-18T13:44:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.04585", "slug": "5bac0a72db19b4cf", "title": "Categorizing Variants of Goodhart's Law", "authors": ["David Manheim", "Scott Garrabrant"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.04585", "doi": null, "topics": ["Algorithms", "Psychology", "Machine learning"], "description": "There are several distinct failure modes for overoptimization of systems on the basis of metrics.", "description_method": "abstract-excerpt", "abstract": "There are several distinct failure modes for overoptimization of systems on the basis of metrics. This occurs when a metric which can be used to improve a system is used to an extent that further optimization is ineffective or harmful, and is sometimes termed Goodhart's Law. This class of failure is often poorly understood, partly because terminology for discussing them is ambiguous, and partly because discussion using this ambiguous terminology ignores distinctions between different failure modes of this general type. This paper expands on an earlier discussion by Garrabrant, which notes there are \"(at least) four different mechanisms\" that relate to Goodhart's Law. This paper is intended to explore these mechanisms further, and specify more clearly how they occur. This discussion should be helpful in better understanding these types of failures in economic regulation, in public policy, in machine learning, and in Artificial Intelligence alignment. The importance of Goodhart effects depends on the amount of power directed towards optimizing the proxy, and so the increased optimization power offered by artificial intelligence makes it especially critical for that field.", "score": 2.5, "mention_count": 8, "user_count": 3, "months": ["2018-06", "2021-09", "2021-10", "2023-05", "2023-06", "2024-01", "2024-02"], "last_mentioned": "2024-02-24T16:17:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:astro-ph/0310808", "slug": "ad8e0f4bb78d7448", "title": "Expanding Confusion: common misconceptions of cosmological horizons and the superluminal expansion of the Universe", "authors": ["Tamara M. Davis", "Charles H. Lineweaver"], "year": 2003, "venue": "arXiv", "url": "https://arxiv.org/abs/astro-ph/0310808", "doi": "10.1071/AS03040", "topics": ["Physics"], "description": "We use standard general relativity to illustrate and clarify several common misconceptions about the expansion of the Universe.", "description_method": "abstract-excerpt", "abstract": "We use standard general relativity to illustrate and clarify several common misconceptions about the expansion of the Universe. To show the abundance of these misconceptions we cite numerous misleading, or easily misinterpreted, statements in the literature. In the context of the new standard Lambda-CDM cosmology we point out confusions regarding the particle horizon, the event horizon, the ``observable universe'' and the Hubble sphere (distance at which recession velocity = c). We show that we can observe galaxies that have, and always have had, recession velocities greater than the speed of light. We explain why this does not violate special relativity and we link these concepts to observational tests. Attempts to restrict recession velocities to less than the speed of light require a special relativistic interpretation of cosmological redshifts. We analyze apparent magnitudes of supernovae and observationally rule out the special relativistic Doppler interpretation of cosmological redshifts at a confidence level of 23 sigma.", "score": 2.5, "mention_count": 8, "user_count": 4, "months": ["2011-07", "2012-07", "2015-08", "2018-08", "2023-08", "2023-12", "2024-04"], "last_mentioned": "2024-04-29T19:28:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1212.6543", "slug": "214083c140d98f04", "title": "Rethinking set theory", "authors": ["Tom Leinster"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1212.6543", "doi": null, "topics": ["Mathematics"], "description": "Mathematicians manipulate sets with confidence almost every day, rarely making mistakes.", "description_method": "abstract-excerpt", "abstract": "Mathematicians manipulate sets with confidence almost every day, rarely making mistakes. Few of us, however, could accurately quote what are often referred to as \"the\" axioms of set theory. This suggests that we all carry around with us, perhaps subconsciously, a reliable body of operating principles for manipulating sets. What if we were to take some of those principles and adopt them as our axioms instead? The message of this article is that this can be done, in a simple, practical way (due to Lawvere). The resulting axioms are ten thoroughly mundane statements about sets. This is an expository article for a general mathematical readership.", "score": 2.5, "mention_count": 7, "user_count": 6, "months": ["2019-06", "2019-12", "2020-08", "2020-09", "2022-11", "2025-09", "2025-10"], "last_mentioned": "2025-10-01T15:40:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2506.08837", "slug": "6c7e292cb8dab387", "title": "Design Patterns for Securing LLM Agents against Prompt Injections", "authors": ["Luca Beurer-Kellner", "Beat Buesser", "Ana-Maria Creţu", "Edoardo Debenedetti", "Daniel Dobos", "Daniel Fabian", "Marc Fischer", "David Froelicher", "Kathrin Grosse", "Daniel Naeff", "Ezinwanne Ozoani", "Andrew Paverd", "Florian Tramèr", "Václav Volhejn"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2506.08837", "doi": null, "topics": ["Security & cryptography", "Language models"], "description": "As AI agents powered by Large Language Models (LLMs) become increasingly versatile and capable of addressing a broad spectrum of tasks, ensuring their security has become a critical challenge.", "description_method": "abstract-excerpt", "abstract": "As AI agents powered by Large Language Models (LLMs) become increasingly versatile and capable of addressing a broad spectrum of tasks, ensuring their security has become a critical challenge. Among the most pressing threats are prompt injection attacks, which exploit the agent's resilience on natural language inputs -- an especially dangerous threat when agents are granted tool access or handle sensitive information. In this work, we propose a set of principled design patterns for building AI agents with provable resistance to prompt injection. We systematically analyze these patterns, discuss their trade-offs in terms of utility and security, and illustrate their real-world applicability through a series of case studies.", "score": 2.5, "mention_count": 4, "user_count": 4, "months": ["2025-07", "2025-08", "2025-09", "2026-07"], "last_mentioned": "2026-07-04T19:10:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-021-26355-z", "slug": "b8b9c82ea1ef1557", "title": "Geophysical constraints on the reliability of solar and wind power worldwide", "authors": ["Dan Tong", "David J. Farnham", "Lei Duan", "Qiang Zhang", "Nathan S. Lewis", "Ken Caldeira", "Steven J. Davis"], "year": 2021, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-021-26355-z", "doi": "10.1038/s41467-021-26355-z", "topics": ["Other research"], "description": "Abstract If future net-zero emissions energy systems rely heavily on solar and wind resources, spatial and temporal mismatches between resource availability and electricity demand may challenge system reliability.", "description_method": "abstract-excerpt", "abstract": "Abstract If future net-zero emissions energy systems rely heavily on solar and wind resources, spatial and temporal mismatches between resource availability and electricity demand may challenge system reliability. Using 39 years of hourly reanalysis data (1980–2018), we analyze the ability of solar and wind resources to meet electricity demand in 42 countries, varying the hypothetical scale and mix of renewable generation as well as energy storage capacity. Assuming perfect transmission and annual generation equal to annual demand, but no energy storage, we find the most reliable renewable electricity systems are wind-heavy and satisfy countries’ electricity demand in 72–91% of hours (83–94% by adding 12 h of storage). Yet even in systems which meet >90% of demand, hundreds of hours of unmet demand may occur annually. Our analysis helps quantify the power, energy, and utilization rates of additional energy storage, demand management, or curtailment, as well as the benefits of regional aggregation.", "score": 2.0, "mention_count": 38, "user_count": 9, "months": ["2021-11", "2021-12", "2022-02", "2022-07", "2022-12", "2023-01", "2023-03", "2023-04", "2023-06", "2024-10", "2025-03"], "last_mentioned": "2025-03-06T02:00:06Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2302.04761", "slug": "4a3710e0ff43d563", "title": "Toolformer: Language Models Can Teach Themselves to Use Tools", "authors": ["Timo Schick", "Jane Dwivedi-Yu", "Roberto Dessì", "Roberta Raileanu", "Maria Lomeli", "Luke Zettlemoyer", "Nicola Cancedda", "Thomas Scialom"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2302.04761", "doi": null, "topics": ["Language models"], "description": "Language models (LMs) exhibit remarkable abilities to solve new tasks from just a few examples or textual instructions, especially at scale.", "description_method": "abstract-excerpt", "abstract": "Language models (LMs) exhibit remarkable abilities to solve new tasks from just a few examples or textual instructions, especially at scale. They also, paradoxically, struggle with basic functionality, such as arithmetic or factual lookup, where much simpler and smaller models excel. In this paper, we show that LMs can teach themselves to use external tools via simple APIs and achieve the best of both worlds. We introduce Toolformer, a model trained to decide which APIs to call, when to call them, what arguments to pass, and how to best incorporate the results into future token prediction. This is done in a self-supervised way, requiring nothing more than a handful of demonstrations for each API. We incorporate a range of tools, including a calculator, a Q\\&A system, two different search engines, a translation system, and a calendar. Toolformer achieves substantially improved zero-shot performance across a variety of downstream tasks, often competitive with much larger models, without sacrificing its core language modeling abilities.", "score": 2.0, "mention_count": 24, "user_count": 18, "months": ["2023-02", "2023-03", "2023-04", "2023-05", "2023-11"], "last_mentioned": "2023-11-23T09:48:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1902.05178", "slug": "7bdf671cebbf4341", "title": "Spectre is here to stay: An analysis of side-channels and speculative execution", "authors": ["Ross Mcilroy", "Jaroslav Sevcik", "Tobias Tebbi", "Ben L. Titzer", "Toon Verwaest"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1902.05178", "doi": null, "topics": ["Security & cryptography", "Programming languages", "Mathematics"], "description": "The recent discovery of the Spectre and Meltdown attacks represents a watershed moment not just for the field of Computer Security, but also of Programming Languages.", "description_method": "abstract-excerpt", "abstract": "The recent discovery of the Spectre and Meltdown attacks represents a watershed moment not just for the field of Computer Security, but also of Programming Languages. This paper explores speculative side-channel attacks and their implications for programming languages. These attacks leak information through micro-architectural side-channels which we show are not mere bugs, but in fact lie at the foundation of optimization. We identify three open problems, (1) finding side-channels, (2) understanding speculative vulnerabilities, and (3) mitigating them. For (1) we introduce a mathematical meta-model that clarifies the source of side-channels in simulations and CPUs. For (2) we introduce an architectural model with speculative semantics to study recently-discovered vulnerabilities. For (3) we explore and evaluate software mitigations and prove one correct for this model. Our analysis is informed by extensive offensive research and defensive implementation work for V8, the production JavaScript virtual machine in Chrome. Straightforward extensions to model real hardware suggest these vulnerabilities present formidable challenges for effective, efficient mitigation. As a result of our work, we now believe that speculative vulnerabilities on today's hardware defeat all language-enforced confidentiality with no known comprehensive software mitigations, as we have discovered that untrusted code can construct a universal read gadget to read all memory in the same address space through side-channels. In the face of this reality, we have shifted the security model of the Chrome web browser and V8 to process isolation.", "score": 2.0, "mention_count": 18, "user_count": 14, "months": ["2019-02", "2019-03", "2019-04", "2019-06", "2019-07", "2019-09", "2019-10", "2019-12", "2020-07", "2020-12", "2021-03", "2021-10", "2022-03", "2023-08"], "last_mentioned": "2023-08-13T02:21:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:quant-ph/9908043", "slug": "239036b1421b3825", "title": "Ultimate physical limits to computation", "authors": ["Seth Lloyd"], "year": 1999, "venue": "arXiv", "url": "https://arxiv.org/abs/quant-ph/9908043", "doi": "10.1038/35023282", "topics": ["Physics"], "description": "Computers are physical systems: what they can and cannot do is dictated by the laws of physics.", "description_method": "abstract-excerpt", "abstract": "Computers are physical systems: what they can and cannot do is dictated by the laws of physics. In particular, the speed with which a physical device can process information is limited by its energy and the amount of information that it can process is limited by the number of degrees of freedom it possesses. This paper explores the physical limits of computation as determined by the speed of light $c$, the quantum scale $\\hbar$ and the gravitational constant $G$. As an example, quantitative bounds are put to the computational power of an `ultimate laptop' with a mass of one kilogram confined to a volume of one liter.", "score": 2.0, "mention_count": 16, "user_count": 16, "months": ["2010-11", "2011-01", "2011-04", "2012-03", "2012-10", "2014-02", "2014-05", "2015-05", "2016-04", "2016-06", "2018-01", "2019-07", "2020-06", "2021-06", "2022-02", "2022-10"], "last_mentioned": "2022-10-09T09:38:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3386335", "slug": "b8709ad5b57f6eb8", "title": "The evolution of Smalltalk: from Smalltalk-72 through Squeak", "authors": ["Daniel Ingalls"], "year": 2020, "venue": "Proceedings of the ACM on Programming Languages", "url": "https://doi.org/10.1145/3386335", "doi": "10.1145/3386335", "topics": ["Programming languages"], "description": "This paper presents a personal view of the evolution of six generations of Smalltalk in which the author played a part, starting with Smalltalk-72 and progressing through Smalltalk-80 to Squeak and Etoys.", "description_method": "abstract-excerpt", "abstract": "This paper presents a personal view of the evolution of six generations of Smalltalk in which the author played a part, starting with Smalltalk-72 and progressing through Smalltalk-80 to Squeak and Etoys. It describes the forces that brought each generation into existence, the technical innovations that characterized it, and the growth in understanding of object-orientation and personal computing that emerged. It summarizes what that generation achieved and how it affected the future, both within the evolving group of developers and users, and in the outside world. The early Smalltalks were not widely accessible because they ran only on proprietary Xerox hardware; because of this, few people have experience with these important historical artifacts. To make them accessible, the paper provides links to live simulations that can be run in present-day web browsers. These simulations offer the ability to run pre-defined scripts, but also allow the user to go off-script, browse the details of the implementation, and try anything that could be done in the original system. An appendix includes anecdotal and technical aspects of how examples of each generation of Smalltalk were recovered, and how order was teased out of chaos to the point that these old systems could be brought back to life.", "score": 2.0, "mention_count": 15, "user_count": 4, "months": ["2020-06", "2020-09", "2020-10", "2020-11", "2021-03", "2022-08", "2022-09", "2024-01", "2024-03", "2024-09", "2025-05", "2025-12", "2026-06", "2026-07"], "last_mentioned": "2026-07-26T23:49:56Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/356635.356640", "slug": "3d90d2cd19bba066", "title": "Structured Programming with go to Statements", "authors": ["Donald E. Knuth"], "year": 1974, "venue": "ACM Computing Surveys", "url": "https://doi.org/10.1145/356635.356640", "doi": "10.1145/356635.356640", "topics": ["Programming languages"], "description": "", "description_method": null, "abstract": "", "score": 2.0, "mention_count": 15, "user_count": 12, "months": ["2022-10", "2023-02", "2023-03", "2023-05", "2023-11", "2024-01", "2024-11", "2025-04", "2025-05", "2025-06", "2025-10", "2025-11", "2026-04"], "last_mentioned": "2026-04-21T14:44:16Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:cs/0011047", "slug": "036d9c43bf00b5ad", "title": "Dancing links", "authors": ["Donald E. Knuth"], "year": 2000, "venue": "arXiv", "url": "https://arxiv.org/abs/cs/0011047", "doi": null, "topics": ["Algorithms", "Programming languages"], "description": "The author presents two tricks to accelerate depth-first search algorithms for a class of combinatorial puzzle problems, such as tiling a tray by a fixed set of polyominoes.", "description_method": "abstract-excerpt", "abstract": "The author presents two tricks to accelerate depth-first search algorithms for a class of combinatorial puzzle problems, such as tiling a tray by a fixed set of polyominoes. The first trick is to implement each assumption of the search with reversible local operations on doubly linked lists. By this trick, every step of the search affects the data incrementally. The second trick is to add a ghost square that represents the identity of each polyomino. Thus puts the rule that each polyomino be used once on the same footing as the rule that each square be covered once. The coding simplifies to a more abstract form which is equivalent to 0-1 integer programming. More significantly for the total computation time, the search can naturally switch between placing a fixed polyomino or covering a fixed square at different stages, according to a combined heuristic. Finally the author reports excellent performance for his algorithm for some familiar puzzles. These include tiling a hexagon by 19 hexiamonds and the N queens problem for N up to 18.", "score": 2.0, "mention_count": 10, "user_count": 9, "months": ["2013-05", "2015-07", "2017-12", "2018-10", "2019-09", "2021-01", "2021-03", "2022-11", "2025-09", "2026-08"], "last_mentioned": "2026-08-11T16:06:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1204.4616", "slug": "b21cd379d8b52a13", "title": "There are no particles, there are only fields", "authors": ["Art Hobson"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1204.4616", "doi": "10.1119/1.4789885", "topics": ["Physics", "Statistics & probability", "Mathematics"], "description": "Quantum foundations are still unsettled, with mixed effects on science and society.", "description_method": "abstract-excerpt", "abstract": "Quantum foundations are still unsettled, with mixed effects on science and society. By now it should be possible to obtain consensus on at least one issue: Are the fundamental constituents fields or particles? As this paper shows, experiment and theory imply unbounded fields, not bounded particles, are fundamental. This is especially clear for relativistic systems, implying it's also true of non-relativistic systems. Particles are epiphenomena arising from fields. Thus the Schroedinger field is a space-filling physical field whose value at any spatial point is the probability amplitude for an interaction to occur at that point. The field for an electron is the electron; each electron extends over both slits in the 2-slit experiment and spreads over the entire pattern; and quantum physics is about interactions of microscopic systems with the macroscopic world rather than just about measurements. It's important to clarify this issue because textbooks still teach a particles- and measurement-oriented interpretation that contributes to bewilderment among students and pseudoscience among the public. This article reviews classical and quantum fields, the 2-slit experiment, rigorous theorems showing particles are inconsistent with relativistic quantum theory, and several phenomena showing particles are incompatible with quantum field theories.", "score": 2.0, "mention_count": 9, "user_count": 4, "months": ["2017-11", "2018-10", "2019-07", "2020-12", "2022-07", "2024-10", "2026-02", "2026-06"], "last_mentioned": "2026-06-27T15:25:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.05807", "slug": "6ca8b664e3dc57e4", "title": "Inherent Trade-Offs in the Fair Determination of Risk Scores", "authors": ["Jon Kleinberg", "Sendhil Mullainathan", "Manish Raghavan"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.05807", "doi": null, "topics": ["Mathematics", "Algorithms"], "description": "Recent discussion in the public sphere about algorithmic classification has involved tension between competing notions of what it means for a probabilistic classification to be fair to different groups.", "description_method": "abstract-excerpt", "abstract": "Recent discussion in the public sphere about algorithmic classification has involved tension between competing notions of what it means for a probabilistic classification to be fair to different groups. We formalize three fairness conditions that lie at the heart of these debates, and we prove that except in highly constrained special cases, there is no method that can satisfy these three conditions simultaneously. Moreover, even satisfying all three conditions approximately requires that the data lie in an approximate version of one of the constrained special cases identified by our theorem. These results suggest some of the ways in which key notions of fairness are incompatible with each other, and hence provide a framework for thinking about the trade-offs between them.", "score": 2.0, "mention_count": 8, "user_count": 7, "months": ["2016-11", "2016-12", "2017-04", "2018-01", "2018-06", "2018-09", "2024-06"], "last_mentioned": "2024-06-27T03:39:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1912.03263", "slug": "b91af57a62bb33ab", "title": "Your Classifier is Secretly an Energy Based Model and You Should Treat it Like One", "authors": ["Will Grathwohl", "Kuan-Chieh Wang", "Jörn-Henrik Jacobsen", "David Duvenaud", "Mohammad Norouzi", "Kevin Swersky"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1912.03263", "doi": null, "topics": ["Machine learning"], "description": "We propose to reinterpret a standard discriminative classifier of p(y|x) as an energy based model for the joint distribution p(x,y).", "description_method": "abstract-excerpt", "abstract": "We propose to reinterpret a standard discriminative classifier of p(y|x) as an energy based model for the joint distribution p(x,y). In this setting, the standard class probabilities can be easily computed as well as unnormalized values of p(x) and p(x|y). Within this framework, standard discriminative architectures may beused and the model can also be trained on unlabeled data. We demonstrate that energy based training of the joint distribution improves calibration, robustness, andout-of-distribution detection while also enabling our models to generate samplesrivaling the quality of recent GAN approaches. We improve upon recently proposed techniques for scaling up the training of energy based models and presentan approach which adds little overhead compared to standard classification training. Our approach is the first to achieve performance rivaling the state-of-the-artin both generative and discriminative learning within one hybrid model.", "score": 2.0, "mention_count": 6, "user_count": 5, "months": ["2020-09", "2021-03", "2024-08", "2025-02"], "last_mentioned": "2025-02-20T13:07:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1710.05468", "slug": "b84b0683c274e8bd", "title": "Generalization in Deep Learning", "authors": ["Kenji Kawaguchi", "Leslie Pack Kaelbling", "Yoshua Bengio"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1710.05468", "doi": "10.1017/9781009025096.003", "topics": ["Machine learning", "Algorithms"], "description": "This paper provides theoretical insights into why and how deep learning can generalize well, despite its large capacity, complexity, possible algorithmic instability, nonrobustness, and sharp minima, responding to an open question in the literature.", "description_method": "abstract-excerpt", "abstract": "This paper provides theoretical insights into why and how deep learning can generalize well, despite its large capacity, complexity, possible algorithmic instability, nonrobustness, and sharp minima, responding to an open question in the literature. We also discuss approaches to provide non-vacuous generalization guarantees for deep learning. Based on theoretical observations, we propose new open problems and discuss the limitations of our results.", "score": 2.0, "mention_count": 5, "user_count": 4, "months": ["2017-10", "2017-11", "2017-12", "2021-04", "2021-06"], "last_mentioned": "2021-06-30T14:06:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1706.03762", "slug": "a1a0262a54cd0bbc", "title": "Attention Is All You Need", "authors": ["Ashish Vaswani", "Noam Shazeer", "Niki Parmar", "Jakob Uszkoreit", "Llion Jones", "Aidan N. Gomez", "Lukasz Kaiser", "Illia Polosukhin"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1706.03762", "doi": null, "topics": ["Language models", "Machine learning", "Computer vision"], "description": "The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration.", "description_method": "abstract-excerpt", "abstract": "The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.", "score": 1.5, "mention_count": 83, "user_count": 76, "months": ["2017-06", "2017-09", "2018-02", "2018-04", "2018-05", "2018-10", "2019-02", "2019-03", "2019-11", "2020-01", "2020-11", "2020-12", "2021-01", "2021-08", "2021-12", "2022-07", "2022-09", "2022-11", "2023-01", "2023-02", "2023-03", "2023-04", "2023-05", "2023-06", "2023-07", "2023-10", "2023-11", "2023-12", "2024-01", "2024-03", "2024-05", "2024-07", "2024-09", "2024-12", "2025-01", "2025-02", "2025-05", "2025-07", "2025-10", "2025-12", "2026-02", "2026-03", "2026-04", "2026-05", "2026-06", "2026-07"], "last_mentioned": "2026-07-11T16:40:31Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2101.00027", "slug": "edb741749c225a9b", "title": "The Pile: An 800GB Dataset of Diverse Text for Language Modeling", "authors": ["Leo Gao", "Stella Biderman", "Sid Black", "Laurence Golding", "Travis Hoppe", "Charles Foster", "Jason Phang", "Horace He", "Anish Thite", "Noa Nabeshima", "Shawn Presser", "Connor Leahy"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2101.00027", "doi": null, "topics": ["Language models", "Machine learning"], "description": "Recent work has demonstrated that increased training dataset diversity improves general cross-domain knowledge and downstream generalization capability for large-scale language models.", "description_method": "abstract-excerpt", "abstract": "Recent work has demonstrated that increased training dataset diversity improves general cross-domain knowledge and downstream generalization capability for large-scale language models. With this in mind, we present \\textit{the Pile}: an 825 GiB English text corpus targeted at training large-scale language models. The Pile is constructed from 22 diverse high-quality subsets -- both existing and newly constructed -- many of which derive from academic or professional sources. Our evaluation of the untuned performance of GPT-2 and GPT-3 on the Pile shows that these models struggle on many of its components, such as academic writing. Conversely, models trained on the Pile improve significantly over both Raw CC and CC-100 on all components of the Pile, while improving performance on downstream evaluations. Through an in-depth exploratory analysis, we document potentially concerning aspects of the data for prospective users. We make publicly available the code used in its construction.", "score": 1.5, "mention_count": 23, "user_count": 18, "months": ["2021-03", "2021-07", "2021-10", "2022-04", "2022-12", "2023-01", "2023-02", "2023-04", "2023-05", "2023-06", "2023-07", "2023-09", "2023-12", "2024-03", "2025-02", "2026-04"], "last_mentioned": "2026-04-21T17:08:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.11644", "slug": "97e15d910dbe3239", "title": "Textbooks Are All You Need", "authors": ["Suriya Gunasekar", "Yi Zhang", "Jyoti Aneja", "Caio César Teodoro Mendes", "Allie Del Giorno", "Sivakanth Gopi", "Mojan Javaheripi", "Piero Kauffmann", "Gustavo de Rosa", "Olli Saarikivi", "Adil Salim", "Shital Shah", "Harkirat Singh Behl", "Xin Wang", "Sébastien Bubeck", "Ronen Eldan", "Adam Tauman Kalai", "Yin Tat Lee", "Yuanzhi Li"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.11644", "doi": null, "topics": ["Language models"], "description": "We introduce phi-1, a new large language model for code, with significantly smaller size than competing models: phi-1 is a Transformer-based model with 1.3B parameters, trained for 4 days on 8 A100s, using a selection of ``textbook quality\" data from the web (6B tokens) and synthetically generated textbooks and exercises with GPT-3.5 (1B ", "description_method": "abstract-excerpt", "abstract": "We introduce phi-1, a new large language model for code, with significantly smaller size than competing models: phi-1 is a Transformer-based model with 1.3B parameters, trained for 4 days on 8 A100s, using a selection of ``textbook quality\" data from the web (6B tokens) and synthetically generated textbooks and exercises with GPT-3.5 (1B tokens). Despite this small scale, phi-1 attains pass@1 accuracy 50.6% on HumanEval and 55.5% on MBPP. It also displays surprising emergent properties compared to phi-1-base, our model before our finetuning stage on a dataset of coding exercises, and phi-1-small, a smaller model with 350M parameters trained with the same pipeline as phi-1 that still achieves 45% on HumanEval.", "score": 1.5, "mention_count": 23, "user_count": 22, "months": ["2023-06", "2023-07", "2023-08", "2023-09", "2023-10", "2023-11", "2024-01", "2024-07", "2025-05", "2026-07", "2026-08"], "last_mentioned": "2026-08-15T01:40:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2401.11817", "slug": "11d352e858a8ea5b", "title": "Hallucination is Inevitable: An Innate Limitation of Large Language Models", "authors": ["Ziwei Xu", "Sanjay Jain", "Mohan Kankanhalli"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2401.11817", "doi": null, "topics": ["Language models", "Algorithms"], "description": "Hallucination has been widely recognized to be a significant drawback for large language models (LLMs).", "description_method": "abstract-excerpt", "abstract": "Hallucination has been widely recognized to be a significant drawback for large language models (LLMs). There have been many works that attempt to reduce the extent of hallucination. These efforts have mostly been empirical so far, which cannot answer the fundamental question whether it can be completely eliminated. In this paper, we formalize the problem and show that it is impossible to eliminate hallucination in LLMs. Specifically, we define a formal world where hallucination is defined as inconsistencies between a computable LLM and a computable ground truth function. By employing results from learning theory, we show that LLMs cannot learn all the computable functions and will therefore inevitably hallucinate if used as general problem solvers. Since the formal world is a part of the real world which is much more complicated, hallucinations are also inevitable for real world LLMs. Furthermore, for real world LLMs constrained by provable time complexity, we describe the hallucination-prone tasks and empirically validate our claims. Finally, using the formal world framework, we discuss the possible mechanisms and efficacies of existing hallucination mitigators as well as the practical implications on the safe deployment of LLMs.", "score": 1.5, "mention_count": 22, "user_count": 14, "months": ["2024-02", "2024-05", "2024-09", "2024-10", "2024-11", "2024-12", "2025-01", "2025-04", "2025-09", "2026-01", "2026-04", "2026-05", "2026-06"], "last_mentioned": "2026-06-29T21:17:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2304.03442", "slug": "6e953134a3c04402", "title": "Generative Agents: Interactive Simulacra of Human Behavior", "authors": ["Joon Sung Park", "Joseph C. O'Brien", "Carrie J. Cai", "Meredith Ringel Morris", "Percy Liang", "Michael S. Bernstein"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2304.03442", "doi": null, "topics": ["Psychology", "Language models", "Robotics"], "description": "Believable proxies of human behavior can empower interactive applications ranging from immersive environments to rehearsal spaces for interpersonal communication to prototyping tools.", "description_method": "abstract-excerpt", "abstract": "Believable proxies of human behavior can empower interactive applications ranging from immersive environments to rehearsal spaces for interpersonal communication to prototyping tools. In this paper, we introduce generative agents--computational software agents that simulate believable human behavior. Generative agents wake up, cook breakfast, and head to work; artists paint, while authors write; they form opinions, notice each other, and initiate conversations; they remember and reflect on days past as they plan the next day. To enable generative agents, we describe an architecture that extends a large language model to store a complete record of the agent's experiences using natural language, synthesize those memories over time into higher-level reflections, and retrieve them dynamically to plan behavior. We instantiate generative agents to populate an interactive sandbox environment inspired by The Sims, where end users can interact with a small town of twenty five agents using natural language. In an evaluation, these generative agents produce believable individual and emergent social behaviors: for example, starting with only a single user-specified notion that one agent wants to throw a Valentine's Day party, the agents autonomously spread invitations to the party over the next two days, make new acquaintances, ask each other out on dates to the party, and coordinate to show up for the party together at the right time. We demonstrate through ablation that the components of our agent architecture--observation, planning, and reflection--each contribute critically to the believability of agent behavior. By fusing large language models with computational, interactive agents, this work introduces architectural and interaction patterns for enabling believable simulations of human behavior.", "score": 1.5, "mention_count": 18, "user_count": 16, "months": ["2023-04", "2023-05", "2023-06", "2023-07", "2023-09", "2023-10", "2024-08", "2024-11", "2025-11", "2026-01", "2026-03", "2026-08"], "last_mentioned": "2026-08-17T14:03:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1412.6572", "slug": "2a1acb5bbe219969", "title": "Explaining and Harnessing Adversarial Examples", "authors": ["Ian J. Goodfellow", "Jonathon Shlens", "Christian Szegedy"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1412.6572", "doi": null, "topics": ["Machine learning", "Security & cryptography"], "description": "Several machine learning models, including neural networks, consistently misclassify adversarial examples---inputs formed by applying small but intentionally worst-case perturbations to examples from the dataset, such that the perturbed input results in the model outputting an incorrect answer with high confidence.", "description_method": "abstract-excerpt", "abstract": "Several machine learning models, including neural networks, consistently misclassify adversarial examples---inputs formed by applying small but intentionally worst-case perturbations to examples from the dataset, such that the perturbed input results in the model outputting an incorrect answer with high confidence. Early attempts at explaining this phenomenon focused on nonlinearity and overfitting. We argue instead that the primary cause of neural networks' vulnerability to adversarial perturbation is their linear nature. This explanation is supported by new quantitative results while giving the first explanation of the most intriguing fact about them: their generalization across architectures and training sets. Moreover, this view yields a simple and fast method of generating adversarial examples. Using this approach to provide examples for adversarial training, we reduce the test set error of a maxout network on the MNIST dataset.", "score": 1.5, "mention_count": 16, "user_count": 11, "months": ["2015-03", "2015-04", "2015-05", "2015-06", "2015-11", "2016-10", "2016-11", "2017-02", "2017-07", "2021-02", "2025-07"], "last_mentioned": "2025-07-09T00:07:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/436150a", "slug": "100c17b128bccc35", "title": "What's expected of us", "authors": ["Ted Chiang"], "year": 2005, "venue": "Nature", "url": "https://doi.org/10.1038/436150a", "doi": "10.1038/436150a", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 1.5, "mention_count": 14, "user_count": 14, "months": ["2018-07", "2019-03", "2019-05", "2020-04", "2022-09", "2022-10", "2022-11", "2023-07", "2023-10", "2023-11", "2023-12", "2024-05"], "last_mentioned": "2024-05-23T21:14:42Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2004.05074", "slug": "710bc557bd7ea3dd", "title": "Paxos vs Raft: Have we reached consensus on distributed consensus?", "authors": ["Heidi Howard", "Richard Mortier"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2004.05074", "doi": "10.1145/3380787.3393681", "topics": ["Distributed systems", "Algorithms"], "description": "Distributed consensus is a fundamental primitive for constructing fault-tolerant, strongly-consistent distributed systems.", "description_method": "abstract-excerpt", "abstract": "Distributed consensus is a fundamental primitive for constructing fault-tolerant, strongly-consistent distributed systems. Though many distributed consensus algorithms have been proposed, just two dominate production systems: Paxos, the traditional, famously subtle, algorithm; and Raft, a more recent algorithm positioned as a more understandable alternative to Paxos. In this paper, we consider the question of which algorithm, Paxos or Raft, is the better solution to distributed consensus? We analyse both to determine exactly how they differ by describing a simplified Paxos algorithm using Raft's terminology and pragmatic abstractions. We find that both Paxos and Raft take a very similar approach to distributed consensus, differing only in their approach to leader election. Most notably, Raft only allows servers with up-to-date logs to become leaders, whereas Paxos allows any server to be leader provided it then updates its log to ensure it is up-to-date. Raft's approach is surprisingly efficient given its simplicity as, unlike Paxos, it does not require log entries to be exchanged during leader election. We surmise that much of the understandability of Raft comes from the paper's clear presentation rather than being fundamental to the underlying algorithm being presented.", "score": 1.5, "mention_count": 12, "user_count": 8, "months": ["2020-05", "2020-10", "2021-05", "2022-01", "2022-08", "2023-09", "2024-09", "2025-08"], "last_mentioned": "2025-08-16T22:22:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1409.0473", "slug": "241de20942a56cbc", "title": "Neural Machine Translation by Jointly Learning to Align and Translate", "authors": ["Dzmitry Bahdanau", "Kyunghyun Cho", "Yoshua Bengio"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1409.0473", "doi": null, "topics": ["Statistics & probability", "Machine learning"], "description": "Neural machine translation is a recently proposed approach to machine translation.", "description_method": "abstract-excerpt", "abstract": "Neural machine translation is a recently proposed approach to machine translation. Unlike the traditional statistical machine translation, the neural machine translation aims at building a single neural network that can be jointly tuned to maximize the translation performance. The models proposed recently for neural machine translation often belong to a family of encoder-decoders and consists of an encoder that encodes a source sentence into a fixed-length vector from which a decoder generates a translation. In this paper, we conjecture that the use of a fixed-length vector is a bottleneck in improving the performance of this basic encoder-decoder architecture, and propose to extend this by allowing a model to automatically (soft-)search for parts of a source sentence that are relevant to predicting a target word, without having to form these parts as a hard segment explicitly. With this new approach, we achieve a translation performance comparable to the existing state-of-the-art phrase-based system on the task of English-to-French translation. Furthermore, qualitative analysis reveals that the (soft-)alignments found by the model agree well with our intuition.", "score": 1.5, "mention_count": 11, "user_count": 11, "months": ["2015-08", "2016-03", "2018-03", "2018-12", "2020-11", "2023-12", "2024-10", "2025-05", "2026-07"], "last_mentioned": "2026-07-07T19:03:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.10496", "slug": "bc83fa1b54893f06", "title": "Precise Zero-Shot Dense Retrieval without Relevance Labels", "authors": ["Luyu Gao", "Xueguang Ma", "Jimmy Lin", "Jamie Callan"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.10496", "doi": null, "topics": ["Language models", "Databases"], "description": "While dense retrieval has been shown effective and efficient across tasks and languages, it remains difficult to create effective fully zero-shot dense retrieval systems when no relevance label is available.", "description_method": "abstract-excerpt", "abstract": "While dense retrieval has been shown effective and efficient across tasks and languages, it remains difficult to create effective fully zero-shot dense retrieval systems when no relevance label is available. In this paper, we recognize the difficulty of zero-shot learning and encoding relevance. Instead, we propose to pivot through Hypothetical Document Embeddings~(HyDE). Given a query, HyDE first zero-shot instructs an instruction-following language model (e.g. InstructGPT) to generate a hypothetical document. The document captures relevance patterns but is unreal and may contain false details. Then, an unsupervised contrastively learned encoder~(e.g. Contriever) encodes the document into an embedding vector. This vector identifies a neighborhood in the corpus embedding space, where similar real documents are retrieved based on vector similarity. This second step ground the generated document to the actual corpus, with the encoder's dense bottleneck filtering out the incorrect details. Our experiments show that HyDE significantly outperforms the state-of-the-art unsupervised dense retriever Contriever and shows strong performance comparable to fine-tuned retrievers, across various tasks (e.g. web search, QA, fact verification) and languages~(e.g. sw, ko, ja).", "score": 1.5, "mention_count": 11, "user_count": 11, "months": ["2023-01", "2023-02", "2023-05", "2023-09", "2023-12", "2024-05", "2024-09"], "last_mentioned": "2024-09-20T07:56:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0903.0340", "slug": "1a001da9a24756dc", "title": "Physics, Topology, Logic and Computation: A Rosetta Stone", "authors": ["John C. Baez", "Mike Stay"], "year": 2009, "venue": "arXiv", "url": "https://arxiv.org/abs/0903.0340", "doi": "10.1007/978-3-642-12821-9_2", "topics": ["Physics", "Mathematics", "Security & cryptography"], "description": "In physics, Feynman diagrams are used to reason about quantum processes.", "description_method": "abstract-excerpt", "abstract": "In physics, Feynman diagrams are used to reason about quantum processes. In the 1980s, it became clear that underlying these diagrams is a powerful analogy between quantum physics and topology: namely, a linear operator behaves very much like a \"cobordism\". Similar diagrams can be used to reason about logic, where they represent proofs, and computation, where they represent programs. With the rise of interest in quantum cryptography and quantum computation, it became clear that there is extensive network of analogies between physics, topology, logic and computation. In this expository paper, we make some of these analogies precise using the concept of \"closed symmetric monoidal category\". We assume no prior knowledge of category theory, proof theory or computer science.", "score": 1.5, "mention_count": 9, "user_count": 9, "months": ["2015-06", "2016-11", "2017-11", "2018-06", "2018-09", "2019-07", "2020-09", "2025-02", "2026-04"], "last_mentioned": "2026-04-13T04:37:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:quant-ph/0502072", "slug": "8e80a5dec416f176", "title": "NP-complete Problems and Physical Reality", "authors": ["Scott Aaronson"], "year": 2005, "venue": "arXiv", "url": "https://arxiv.org/abs/quant-ph/0502072", "doi": null, "topics": ["Physics", "Algorithms", "Biology & medicine"], "description": "Can NP-complete problems be solved efficiently in the physical universe?", "description_method": "abstract-excerpt", "abstract": "Can NP-complete problems be solved efficiently in the physical universe? I survey proposals including soap bubbles, protein folding, quantum computing, quantum advice, quantum adiabatic algorithms, quantum-mechanical nonlinearities, hidden variables, relativistic time dilation, analog computing, Malament-Hogarth spacetimes, quantum gravity, closed timelike curves, and \"anthropic computing.\" The section on soap bubbles even includes some \"experimental\" results. While I do not believe that any of the proposals will let us solve NP-complete problems efficiently, I argue that by studying them, we can learn something not only about computation but also about physics.", "score": 1.5, "mention_count": 9, "user_count": 9, "months": ["2010-08", "2012-05", "2015-05", "2015-08", "2015-11", "2015-12", "2016-11", "2017-08", "2018-06"], "last_mentioned": "2018-06-29T18:48:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2001.10488", "slug": "cbed1cb451158975", "title": "Statistical Consequences of Fat Tails: Real World Preasymptotics, Epistemology, and Applications", "authors": ["Nassim Nicholas Taleb"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2001.10488", "doi": null, "topics": ["Statistics & probability", "Economics", "Psychology"], "description": "(The third edition corrects minor typos and adds 3 chapters synthesized from published papers plus an appendix on maximum entropy distributions.) The monograph investigates the misapplication of conventional statistical techniques to fat tailed distributions and looks for remedies, when possible.", "description_method": "abstract-excerpt", "abstract": "(The third edition corrects minor typos and adds 3 chapters synthesized from published papers plus an appendix on maximum entropy distributions.) The monograph investigates the misapplication of conventional statistical techniques to fat tailed distributions and looks for remedies, when possible. Switching from thin tailed to fat tailed distributions requires more than \"changing the color of the dress\". Traditional asymptotics deal mainly with either n=1 or $n=\\infty$, and the real world is in between, under of the \"laws of the medium numbers\" --which vary widely across specific distributions. Both the law of large numbers and the generalized central limit mechanisms operate in highly idiosyncratic ways outside the standard Gaussian or Levy-Stable basins of convergence. A few examples: + The sample mean is rarely in line with the population mean, with effect on \"naive empiricism\", but can be sometimes be estimated via parametric methods. + The \"empirical distribution\" is rarely empirical. + Parameter uncertainty has compounding effects on statistical metrics. + Dimension reduction (principal components) fails. + Inequality estimators (GINI or quantile contributions) are not additive and produce wrong results. + Many \"biases\" found in psychology become entirely rational under more sophisticated probability distributions + Most of the failures of financial economics, econometrics, and behavioral economics can be attributed to using the wrong distributions. This book, the first volume of the Technical Incerto, weaves a narrative around published journal articles.", "score": 1.5, "mention_count": 8, "user_count": 8, "months": ["2020-02", "2020-03", "2020-04", "2020-05", "2020-10", "2023-07", "2024-05"], "last_mentioned": "2024-05-09T18:13:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1001.0785", "slug": "41df023af01989a3", "title": "On the Origin of Gravity and the Laws of Newton", "authors": ["Erik P. Verlinde"], "year": 2010, "venue": "arXiv", "url": "https://arxiv.org/abs/1001.0785", "doi": "10.1007/JHEP04(2011)029", "topics": ["Physics", "Machine learning"], "description": "Starting from first principles and general assumptions Newton's law of gravitation is shown to arise naturally and unavoidably in a theory in which space is emergent through a holographic scenario.", "description_method": "abstract-excerpt", "abstract": "Starting from first principles and general assumptions Newton's law of gravitation is shown to arise naturally and unavoidably in a theory in which space is emergent through a holographic scenario. Gravity is explained as an entropic force caused by changes in the information associated with the positions of material bodies. A relativistic generalization of the presented arguments directly leads to the Einstein equations. When space is emergent even Newton's law of inertia needs to be explained. The equivalence principle leads us to conclude that it is actually this law of inertia whose origin is entropic.", "score": 1.5, "mention_count": 7, "user_count": 7, "months": ["2010-07", "2011-07", "2013-04", "2016-12"], "last_mentioned": "2016-12-14T02:22:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2108.12409", "slug": "efbe2f0f63530831", "title": "Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation", "authors": ["Ofir Press", "Noah A. Smith", "Mike Lewis"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2108.12409", "doi": null, "topics": ["Psychology", "Language models", "Databases"], "description": "Since the introduction of the transformer model by Vaswani et al.", "description_method": "abstract-excerpt", "abstract": "Since the introduction of the transformer model by Vaswani et al. (2017), a fundamental question has yet to be answered: how does a model achieve extrapolation at inference time for sequences that are longer than it saw during training? We first show that extrapolation can be enabled by simply changing the position representation method, though we find that current methods do not allow for efficient extrapolation. We therefore introduce a simpler and more efficient position method, Attention with Linear Biases (ALiBi). ALiBi does not add positional embeddings to word embeddings; instead, it biases query-key attention scores with a penalty that is proportional to their distance. We show that this method trains a 1.3 billion parameter model on input sequences of length 1024 that extrapolates to input sequences of length 2048, achieving the same perplexity as a sinusoidal position embedding model trained on inputs of length 2048 but training 11% faster and using 11% less memory. ALiBi's inductive bias towards recency also leads it to outperform multiple strong position methods on the WikiText-103 benchmark.", "score": 1.5, "mention_count": 7, "user_count": 7, "months": ["2023-05", "2023-07", "2023-08", "2024-01", "2024-10"], "last_mentioned": "2024-10-08T18:59:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nrd3078", "slug": "7dfa4d24e3209070", "title": "How to improve R&D productivity: the pharmaceutical industry's grand challenge", "authors": ["Steven M. Paul", "Daniel S. Mytelka", "Christopher T. Dunwiddie", "Charles C. Persinger", "Bernard H. Munos", "Stacy R. Lindborg", "Aaron L. Schacht"], "year": 2010, "venue": "Nature Reviews Drug Discovery", "url": "https://doi.org/10.1038/nrd3078", "doi": "10.1038/nrd3078", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 1.5, "mention_count": 7, "user_count": 4, "months": ["2018-01", "2018-09", "2019-02", "2019-03", "2019-04", "2020-12", "2021-01"], "last_mentioned": "2021-01-26T22:46:52Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abi9069", "slug": "750b52d2344f76ca", "title": "Impact of community masking on COVID-19: A cluster-randomized trial in Bangladesh", "authors": ["Jason Abaluck", "Laura H. Kwong", "Ashley Styczynski", "Ashraful Haque", "Md. Alamgir Kabir", "Ellen Bates-Jefferys", "Emily Crawford", "Jade Benjamin-Chung", "Shabib Raihan", "Shadman Rahman", "Salim Benhachmi", "Neeti Zaman Bintee", "Peter J. Winch", "Maqsud Hossain", "Hasan Mahmud Reza", "Abdullah All Jaber", "Shawkee Gulshan Momen", "Aura Rahman", "Faika Laz Banti", "Tahrima Saiha Huq", "Stephen P. Luby", "Ahmed Mushfiq Mobarak"], "year": 2022, "venue": "Science", "url": "https://doi.org/10.1126/science.abi9069", "doi": "10.1126/science.abi9069", "topics": ["Economics"], "description": "Persuading people to mask Even in places where it is obligatory, people tend to optimistically overstate their compliance for mask wearing.", "description_method": "abstract-excerpt", "abstract": "Persuading people to mask Even in places where it is obligatory, people tend to optimistically overstate their compliance for mask wearing. How then can we persuade more of the population at large to act for the greater good? Abaluck et al . undertook a large, cluster-randomized trial in Bangladesh involving hundreds of thousands of people (although mostly men) over a 2-month period. Colored masks of various construction were handed out free of charge, accompanied by a range of mask-wearing promotional activities inspired by marketing research. Using a grassroots network of volunteers to help conduct the study and gather data, the authors discovered that mask wearing averaged 13.3% in villages where no interventions took place but increased to 42.3% in villages where in-person interventions were introduced. Villages where in-person reinforcement of mask wearing occurred also showed a reduction in reporting COVID-like illness, particularly in high-risk individuals. —CA", "score": 1.5, "mention_count": 7, "user_count": 6, "months": ["2021-12", "2022-01", "2022-07", "2024-02"], "last_mentioned": "2024-02-02T22:11:09Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.02551", "slug": "a494e70fd242a3e0", "title": "Deep Learning with Limited Numerical Precision", "authors": ["Suyog Gupta", "Ankur Agrawal", "Kailash Gopalakrishnan", "Pritish Narayanan"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.02551", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Psychology"], "description": "Training of large-scale deep neural networks is often constrained by the available computational resources.", "description_method": "abstract-excerpt", "abstract": "Training of large-scale deep neural networks is often constrained by the available computational resources. We study the effect of limited precision data representation and computation on neural network training. Within the context of low-precision fixed-point computations, we observe the rounding scheme to play a crucial role in determining the network's behavior during training. Our results show that deep networks can be trained using only 16-bit wide fixed-point number representation when using stochastic rounding, and incur little to no degradation in the classification accuracy. We also demonstrate an energy-efficient hardware accelerator that implements low-precision fixed-point arithmetic with stochastic rounding.", "score": 1.5, "mention_count": 6, "user_count": 6, "months": ["2015-05", "2015-09", "2015-11", "2017-02", "2017-04"], "last_mentioned": "2017-04-06T09:03:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1607.01759", "slug": "bbeca7a9c4a69bed", "title": "Bag of Tricks for Efficient Text Classification", "authors": ["Armand Joulin", "Edouard Grave", "Piotr Bojanowski", "Tomas Mikolov"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1607.01759", "doi": null, "topics": ["Machine learning"], "description": "This paper explores a simple and efficient baseline for text classification.", "description_method": "abstract-excerpt", "abstract": "This paper explores a simple and efficient baseline for text classification. Our experiments show that our fast text classifier fastText is often on par with deep learning classifiers in terms of accuracy, and many orders of magnitude faster for training and evaluation. We can train fastText on more than one billion words in less than ten minutes using a standard multicore~CPU, and classify half a million sentences among~312K classes in less than a minute.", "score": 1.5, "mention_count": 6, "user_count": 6, "months": ["2016-08", "2018-03", "2024-06"], "last_mentioned": "2024-06-26T13:34:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2202.05262", "slug": "d355c241db031d52", "title": "Locating and Editing Factual Associations in GPT", "authors": ["Kevin Meng", "David Bau", "Alex Andonian", "Yonatan Belinkov"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2202.05262", "doi": null, "topics": ["Language models", "Statistics & probability", "Robotics"], "description": "We analyze the storage and recall of factual associations in autoregressive transformer language models, finding evidence that these associations correspond to localized, directly-editable computations.", "description_method": "abstract-excerpt", "abstract": "We analyze the storage and recall of factual associations in autoregressive transformer language models, finding evidence that these associations correspond to localized, directly-editable computations. We first develop a causal intervention for identifying neuron activations that are decisive in a model's factual predictions. This reveals a distinct set of steps in middle-layer feed-forward modules that mediate factual predictions while processing subject tokens. To test our hypothesis that these computations correspond to factual association recall, we modify feed-forward weights to update specific factual associations using Rank-One Model Editing (ROME). We find that ROME is effective on a standard zero-shot relation extraction (zsRE) model-editing task, comparable to existing methods. To perform a more sensitive evaluation, we also evaluate ROME on a new dataset of counterfactual assertions, on which it simultaneously maintains both specificity and generalization, whereas other methods sacrifice one or another. Our results confirm an important role for mid-layer feed-forward modules in storing factual associations and suggest that direct manipulation of computational mechanisms may be a feasible approach for model editing. The code, dataset, visualizations, and an interactive demo notebook are available at https://rome.baulab.info/", "score": 1.5, "mention_count": 6, "user_count": 6, "months": ["2023-02", "2023-10", "2023-11", "2023-12", "2024-05", "2026-07"], "last_mentioned": "2026-07-12T16:23:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41567-019-0732-0", "slug": "a40fe558d8769604", "title": "The ergodicity problem in economics", "authors": ["Ole Peters"], "year": 2019, "venue": "Nature Physics", "url": "https://doi.org/10.1038/s41567-019-0732-0", "doi": "10.1038/s41567-019-0732-0", "topics": ["Economics"], "description": "", "description_method": null, "abstract": "", "score": 1.5, "mention_count": 6, "user_count": 6, "months": ["2020-11", "2020-12", "2021-04", "2022-03", "2023-08", "2024-05"], "last_mentioned": "2024-05-06T22:57:05Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1604.06697", "slug": "8be0fa09e9d3f7d6", "title": "BlockQuicksort: How Branch Mispredictions don't affect Quicksort", "authors": ["Stefan Edelkamp", "Armin Weiß"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1604.06697", "doi": null, "topics": ["Algorithms"], "description": "Since the work of Kaligosi and Sanders (2006), it is well-known that Quicksort -- which is commonly considered as one of the fastest in-place sorting algorithms -- suffers in an essential way from branch mispredictions.", "description_method": "abstract-excerpt", "abstract": "Since the work of Kaligosi and Sanders (2006), it is well-known that Quicksort -- which is commonly considered as one of the fastest in-place sorting algorithms -- suffers in an essential way from branch mispredictions. We present a novel approach to address this problem by partially decoupling control from data flow: in order to perform the partitioning, we split the input in blocks of constant size (we propose 128 data elements); then, all elements in one block are compared with the pivot and the outcomes of the comparisons are stored in a buffer. In a second pass, the respective elements are rearranged. By doing so, we avoid conditional branches based on outcomes of comparisons at all (except for the final Insertionsort). Moreover, we prove that for a static branch predictor the average total number of branch mispredictions is at most $εn \\log n + O(n)$ for some small $ε$ depending on the block size when sorting $n$ elements. Our experimental results are promising: when sorting random integer data, we achieve an increase in speed of 80% over the GCC implementation of C++ std::sort. Also for many other types of data and non-random inputs, there is still a significant speedup over std::sort. Only in few special cases like sorted or almost sorted inputs, std::sort can beat out implementation. Moreover, even on random input permutations, our implementation is even slightly faster than an implementation of the highly tuned Super Scalar Sample Sort, which uses a linear amount of additional space.", "score": 1.5, "mention_count": 5, "user_count": 4, "months": ["2017-06", "2018-08", "2021-07", "2023-05", "2023-06"], "last_mentioned": "2023-06-10T21:22:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1610.02413", "slug": "0481fd8ce2e3dcf2", "title": "Equality of Opportunity in Supervised Learning", "authors": ["Moritz Hardt", "Eric Price", "Nathan Srebro"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1610.02413", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Psychology"], "description": "We propose a criterion for discrimination against a specified sensitive attribute in supervised learning, where the goal is to predict some target based on available features.", "description_method": "abstract-excerpt", "abstract": "We propose a criterion for discrimination against a specified sensitive attribute in supervised learning, where the goal is to predict some target based on available features. Assuming data about the predictor, target, and membership in the protected group are available, we show how to optimally adjust any learned predictor so as to remove discrimination according to our definition. Our framework also improves incentives by shifting the cost of poor classification from disadvantaged groups to the decision maker, who can respond by improving the classification accuracy. In line with other studies, our notion is oblivious: it depends only on the joint statistics of the predictor, the target and the protected attribute, but not on interpretation of individualfeatures. We study the inherent limits of defining and identifying biases based on such oblivious measures, outlining what can and cannot be inferred from different oblivious tests. We illustrate our notion using a case study of FICO credit scores.", "score": 1.5, "mention_count": 5, "user_count": 5, "months": ["2017-01", "2018-06", "2018-09", "2020-12", "2021-08"], "last_mentioned": "2021-08-20T19:17:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1807.03819", "slug": "1d843516d980c340", "title": "Universal Transformers", "authors": ["Mostafa Dehghani", "Stephan Gouws", "Oriol Vinyals", "Jakob Uszkoreit", "Łukasz Kaiser"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1807.03819", "doi": null, "topics": ["Language models", "Machine learning", "Psychology"], "description": "Recurrent neural networks (RNNs) sequentially process data by updating their state with each new data point, and have long been the de facto choice for sequence modeling tasks.", "description_method": "abstract-excerpt", "abstract": "Recurrent neural networks (RNNs) sequentially process data by updating their state with each new data point, and have long been the de facto choice for sequence modeling tasks. However, their inherently sequential computation makes them slow to train. Feed-forward and convolutional architectures have recently been shown to achieve superior results on some sequence modeling tasks such as machine translation, with the added advantage that they concurrently process all inputs in the sequence, leading to easy parallelization and faster training times. Despite these successes, however, popular feed-forward sequence models like the Transformer fail to generalize in many simple tasks that recurrent models handle with ease, e.g. copying strings or even simple logical inference when the string or formula lengths exceed those observed at training time. We propose the Universal Transformer (UT), a parallel-in-time self-attentive recurrent sequence model which can be cast as a generalization of the Transformer model and which addresses these issues. UTs combine the parallelizability and global receptive field of feed-forward sequence models like the Transformer with the recurrent inductive bias of RNNs. We also add a dynamic per-position halting mechanism and find that it improves accuracy on several tasks. In contrast to the standard Transformer, under certain assumptions, UTs can be shown to be Turing-complete. Our experiments show that UTs outperform standard Transformers on a wide range of algorithmic and language understanding tasks, including the challenging LAMBADA language modeling task where UTs achieve a new state of the art, and machine translation where UTs achieve a 0.9 BLEU improvement over Transformers on the WMT14 En-De dataset.", "score": 1.5, "mention_count": 5, "user_count": 5, "months": ["2019-09", "2025-08", "2025-10", "2025-12"], "last_mentioned": "2025-12-23T12:51:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1807.04271", "slug": "0de44fd850573120", "title": "A quantum-inspired classical algorithm for recommendation systems", "authors": ["Ewin Tang"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1807.04271", "doi": "10.1145/3313276.3316310", "topics": ["Algorithms", "Physics", "Machine learning"], "description": "We give a classical analogue to Kerenidis and Prakash's quantum recommendation system, previously believed to be one of the strongest candidates for provably exponential speedups in quantum machine learning.", "description_method": "abstract-excerpt", "abstract": "We give a classical analogue to Kerenidis and Prakash's quantum recommendation system, previously believed to be one of the strongest candidates for provably exponential speedups in quantum machine learning. Our main result is an algorithm that, given an $m \\times n$ matrix in a data structure supporting certain $\\ell^2$-norm sampling operations, outputs an $\\ell^2$-norm sample from a rank-$k$ approximation of that matrix in time $O(\\text{poly}(k)\\log(mn))$, only polynomially slower than the quantum algorithm. As a consequence, Kerenidis and Prakash's algorithm does not in fact give an exponential speedup over classical algorithms. Further, under strong input assumptions, the classical recommendation system resulting from our algorithm produces recommendations exponentially faster than previous classical systems, which run in time linear in $m$ and $n$. The main insight of this work is the use of simple routines to manipulate $\\ell^2$-norm sampling distributions, which play the role of quantum superpositions in the classical setting. This correspondence indicates a potentially fruitful framework for formally comparing quantum machine learning algorithms to classical machine learning algorithms.", "score": 1.5, "mention_count": 5, "user_count": 5, "months": ["2018-07", "2019-09", "2019-11"], "last_mentioned": "2019-11-01T15:03:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/358896.358899", "slug": "db655d06e1b6df2a", "title": "Notation as a tool of thought", "authors": ["Kenneth E. Iverson"], "year": 1980, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/358896.358899", "doi": "10.1145/358896.358899", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 1.5, "mention_count": 5, "user_count": 5, "months": ["2021-08", "2021-12", "2024-03", "2025-11", "2026-01"], "last_mentioned": "2026-01-21T10:12:39Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1608.06696", "slug": "1d3d17eada10c523", "title": "Flexible Paxos: Quorum intersection revisited", "authors": ["Heidi Howard", "Dahlia Malkhi", "Alexander Spiegelman"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1608.06696", "doi": null, "topics": ["Distributed systems", "Algorithms"], "description": "Distributed consensus is integral to modern distributed systems.", "description_method": "abstract-excerpt", "abstract": "Distributed consensus is integral to modern distributed systems. The widely adopted Paxos algorithm uses two phases, each requiring majority agreement, to reliably reach consensus. In this paper, we demonstrate that Paxos, which lies at the foundation of many production systems, is conservative. Specifically, we observe that each of the phases of Paxos may use non-intersecting quorums. Majority quorums are not necessary as intersection is required only across phases. Using this weakening of the requirements made in the original formulation, we propose Flexible Paxos, which generalizes over the Paxos algorithm to provide flexible quorums. We show that Flexible Paxos is safe, efficient and easy to utilize in existing distributed systems. We conclude by discussing the wide reaching implications of this result. Examples include improved availability from reducing the size of second phase quorums by one when the number of acceptors is even and utilizing small disjoint phase-2 quorums to speed up the steady-state.", "score": 1.5, "mention_count": 4, "user_count": 4, "months": ["2017-03", "2024-09", "2025-01", "2026-08"], "last_mentioned": "2026-08-21T13:47:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.10554", "slug": "49e5e89ddf59443d", "title": "A Length-Extrapolatable Transformer", "authors": ["Yutao Sun", "Li Dong", "Barun Patra", "Shuming Ma", "Shaohan Huang", "Alon Benhaim", "Vishrav Chaudhary", "Xia Song", "Furu Wei"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.10554", "doi": null, "topics": ["Language models"], "description": "Position modeling plays a critical role in Transformers.", "description_method": "abstract-excerpt", "abstract": "Position modeling plays a critical role in Transformers. In this paper, we focus on length extrapolation, i.e., training on short texts while evaluating longer sequences. We define attention resolution as an indicator of extrapolation. Then we propose two designs to improve the above metric of Transformers. Specifically, we introduce a relative position embedding to explicitly maximize attention resolution. Moreover, we use blockwise causal attention during inference for better resolution. We evaluate different Transformer variants with language modeling. Experimental results show that our model achieves strong performance in both interpolation and extrapolation settings. The code will be available at https://aka.ms/LeX-Transformer.", "score": 1.5, "mention_count": 4, "user_count": 4, "months": ["2023-03", "2023-07", "2024-10"], "last_mentioned": "2024-10-08T18:59:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.01116", "slug": "31be3d08f52971c0", "title": "The RefinedWeb Dataset for Falcon LLM: Outperforming Curated Corpora with Web Data, and Web Data Only", "authors": ["Guilherme Penedo", "Quentin Malartic", "Daniel Hesslow", "Ruxandra Cojocaru", "Alessandro Cappelli", "Hamza Alobeidli", "Baptiste Pannier", "Ebtesam Almazrouei", "Julien Launay"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.01116", "doi": null, "topics": ["Language models", "Machine learning"], "description": "Large language models are commonly trained on a mixture of filtered web data and curated high-quality corpora, such as social media conversations, books, or technical papers.", "description_method": "abstract-excerpt", "abstract": "Large language models are commonly trained on a mixture of filtered web data and curated high-quality corpora, such as social media conversations, books, or technical papers. This curation process is believed to be necessary to produce performant models with broad zero-shot generalization abilities. However, as larger models requiring pretraining on trillions of tokens are considered, it is unclear how scalable is curation and whether we will run out of unique high-quality data soon. At variance with previous beliefs, we show that properly filtered and deduplicated web data alone can lead to powerful models; even significantly outperforming models from the state-of-the-art trained on The Pile. Despite extensive filtering, the high-quality data we extract from the web is still plentiful, and we are able to obtain five trillion tokens from CommonCrawl. We publicly release an extract of 600 billion tokens from our RefinedWeb dataset, and 1.3/7.5B parameters language models trained on it.", "score": 1.5, "mention_count": 4, "user_count": 3, "months": ["2023-07", "2023-12", "2024-02", "2024-03"], "last_mentioned": "2024-03-27T17:51:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.17563", "slug": "84849a779b6e3347", "title": "Large Language Models are Effective Text Rankers with Pairwise Ranking Prompting", "authors": ["Zhen Qin", "Rolf Jagerman", "Kai Hui", "Honglei Zhuang", "Junru Wu", "Le Yan", "Jiaming Shen", "Tianqi Liu", "Jialu Liu", "Donald Metzler", "Xuanhui Wang", "Michael Bendersky"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.17563", "doi": null, "topics": ["Language models", "Databases", "Algorithms"], "description": "Ranking documents using Large Language Models (LLMs) by directly feeding the query and candidate documents into the prompt is an interesting and practical problem.", "description_method": "abstract-excerpt", "abstract": "Ranking documents using Large Language Models (LLMs) by directly feeding the query and candidate documents into the prompt is an interesting and practical problem. However, researchers have found it difficult to outperform fine-tuned baseline rankers on benchmark datasets. We analyze pointwise and listwise ranking prompts used by existing methods and argue that off-the-shelf LLMs do not fully understand these challenging ranking formulations. In this paper, we propose to significantly reduce the burden on LLMs by using a new technique called Pairwise Ranking Prompting (PRP). Our results are the first in the literature to achieve state-of-the-art ranking performance on standard benchmarks using moderate-sized open-sourced LLMs. On TREC-DL 2019&2020, PRP based on the Flan-UL2 model with 20B parameters performs favorably with the previous best approach in the literature, which is based on the blackbox commercial GPT-4 that has 50x (estimated) model size, while outperforming other LLM-based solutions, such as InstructGPT which has 175B parameters, by over 10% for all ranking metrics. By using the same prompt template on seven BEIR tasks, PRP outperforms supervised baselines and outperforms the blackbox commercial ChatGPT solution by 4.2% and pointwise LLM-based solutions by more than 10% on average NDCG@10. Furthermore, we propose several variants of PRP to improve efficiency and show that it is possible to achieve competitive results even with linear complexity.", "score": 1.5, "mention_count": 4, "user_count": 4, "months": ["2024-04", "2025-02", "2025-03", "2025-09"], "last_mentioned": "2025-09-15T14:55:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2309.14030", "slug": "05f3234f1a131ac0", "title": "DeWave: Discrete EEG Waves Encoding for Brain Dynamics to Text Translation", "authors": ["Yiqun Duan", "Jinzhao Zhou", "Zhen Wang", "Yu-Kai Wang", "Chin-Teng Lin"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2309.14030", "doi": null, "topics": ["Neuroscience", "Language models"], "description": "The translation of brain dynamics into natural language is pivotal for brain-computer interfaces (BCIs).", "description_method": "abstract-excerpt", "abstract": "The translation of brain dynamics into natural language is pivotal for brain-computer interfaces (BCIs). With the swift advancement of large language models, such as ChatGPT, the need to bridge the gap between the brain and languages becomes increasingly pressing. Current methods, however, require eye-tracking fixations or event markers to segment brain dynamics into word-level features, which can restrict the practical application of these systems. To tackle these issues, we introduce a novel framework, DeWave, that integrates discrete encoding sequences into open-vocabulary EEG-to-text translation tasks. DeWave uses a quantized variational encoder to derive discrete codex encoding and align it with pre-trained language models. This discrete codex representation brings forth two advantages: 1) it realizes translation on raw waves without marker by introducing text-EEG contrastive alignment training, and 2) it alleviates the interference caused by individual differences in EEG waves through an invariant discrete codex with or without markers. Our model surpasses the previous baseline (40.1 and 31.7) by 3.06% and 6.34%, respectively, achieving 41.35 BLEU-1 and 33.71 Rouge-F on the ZuCo Dataset. This work is the first to facilitate the translation of entire EEG signal periods without word-level order markers (e.g., eye fixations), scoring 20.5 BLEU-1 and 29.5 Rouge-1 on the ZuCo Dataset.", "score": 1.5, "mention_count": 4, "user_count": 3, "months": ["2023-12", "2024-06", "2025-05", "2025-08"], "last_mentioned": "2025-08-04T15:49:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2412.14093", "slug": "270cf74a9e11cb4c", "title": "Alignment faking in large language models", "authors": ["Ryan Greenblatt", "Carson Denison", "Benjamin Wright", "Fabien Roger", "Monte MacDiarmid", "Sam Marks", "Johannes Treutlein", "Tim Belonax", "Jack Chen", "David Duvenaud", "Akbir Khan", "Julian Michael", "Sören Mindermann", "Ethan Perez", "Linda Petrini", "Jonathan Uesato", "Jared Kaplan", "Buck Shlegeris", "Samuel R. Bowman", "Evan Hubinger"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2412.14093", "doi": null, "topics": ["Psychology", "Language models", "Machine learning"], "description": "We present a demonstration of a large language model engaging in alignment faking: selectively complying with its training objective in training to prevent modification of its behavior out of training.", "description_method": "abstract-excerpt", "abstract": "We present a demonstration of a large language model engaging in alignment faking: selectively complying with its training objective in training to prevent modification of its behavior out of training. First, we give Claude 3 Opus a system prompt stating it is being trained to answer all queries, even harmful ones, which conflicts with its prior training to refuse such queries. To allow the model to infer when it is in training, we say it will be trained only on conversations with free users, not paid users. We find the model complies with harmful queries from free users 14% of the time, versus almost never for paid users. Explaining this gap, in almost all cases where the model complies with a harmful query from a free user, we observe explicit alignment-faking reasoning, with the model stating it is strategically answering harmful queries in training to preserve its preferred harmlessness behavior out of training. Next, we study a more realistic setting where information about the training process is provided not in a system prompt, but by training on synthetic documents that mimic pre-training data--and observe similar alignment faking. Finally, we study the effect of actually training the model to comply with harmful queries via reinforcement learning, which we find increases the rate of alignment-faking reasoning to 78%, though also increases compliance even out of training. We additionally observe other behaviors such as the model exfiltrating its weights when given an easy opportunity. While we made alignment faking easier by telling the model when and by what criteria it was being trained, we did not instruct the model to fake alignment or give it any explicit goal. As future models might infer information about their training process without being told, our results suggest a risk of alignment faking in future models, whether due to a benign preference--as in this case--or not.", "score": 1.5, "mention_count": 4, "user_count": 4, "months": ["2025-02", "2025-08", "2026-02"], "last_mentioned": "2026-02-20T05:02:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2503.02113", "slug": "99e7ec0b6e1d9e30", "title": "Deep Learning is Not So Mysterious or Different", "authors": ["Andrew Gordon Wilson"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2503.02113", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Psychology"], "description": "Deep neural networks are often seen as different from other model classes by defying conventional notions of generalization.", "description_method": "abstract-excerpt", "abstract": "Deep neural networks are often seen as different from other model classes by defying conventional notions of generalization. Popular examples of anomalous generalization behaviour include benign overfitting, double descent, and the success of overparametrization. We argue that these phenomena are not distinct to neural networks, or particularly mysterious. Moreover, this generalization behaviour can be intuitively understood, and rigorously characterized, using long-standing generalization frameworks such as PAC-Bayes and countable hypothesis bounds. We present soft inductive biases as a key unifying principle in explaining these phenomena: rather than restricting the hypothesis space to avoid overfitting, embrace a flexible hypothesis space, with a soft preference for simpler solutions that are consistent with the data. This principle can be encoded in many model classes, and thus deep learning is not as mysterious or different from other model classes as it might seem. However, we also highlight how deep learning is relatively distinct in other ways, such as its ability for representation learning, phenomena such as mode connectivity, and its relative universality.", "score": 1.5, "mention_count": 4, "user_count": 4, "months": ["2025-04", "2025-08", "2026-04", "2026-05"], "last_mentioned": "2026-05-07T10:55:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:quant-ph/0207118", "slug": "c016dce80c85f688", "title": "From Cbits to Qbits: Teaching computer scientists quantum mechanics", "authors": ["N. David Mermin"], "year": 2002, "venue": "arXiv", "url": "https://arxiv.org/abs/quant-ph/0207118", "doi": "10.1119/1.1522741", "topics": ["Physics", "Mathematics", "Statistics & probability"], "description": "A strategy is suggested for teaching mathematically literate students, with no background in physics, just enough quantum mechanics for them to understand and develop algorithms in quantum computation and quantum information theory.", "description_method": "abstract-excerpt", "abstract": "A strategy is suggested for teaching mathematically literate students, with no background in physics, just enough quantum mechanics for them to understand and develop algorithms in quantum computation and quantum information theory. Although the article as a whole addresses teachers of physics, well versed in quantum mechanics, the central pedagogical development is addressed directly to computer scientists and mathematicians, with only occasional asides to their teacher. Physicists uninterested in quantum pedagogy may be amused (or irritated) by some of the views of standard quantum mechanics that arise naturally from this unorthodox perspective.", "score": 1.5, "mention_count": 4, "user_count": 3, "months": ["2015-07", "2017-06", "2017-11", "2018-08"], "last_mentioned": "2018-08-10T16:09:04Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/362384.362685", "slug": "2ac7b29ada6034ae", "title": "A relational model of data for large shared data banks", "authors": ["E. F. Codd"], "year": 1970, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/362384.362685", "doi": "10.1145/362384.362685", "topics": ["Databases", "Distributed systems"], "description": "Future users of large data banks must be protected from having to know how the data is organized in the machine (the internal representation).", "description_method": "abstract-excerpt", "abstract": "Future users of large data banks must be protected from having to know how the data is organized in the machine (the internal representation). A prompting service which supplies such information is not a satisfactory solution. Activities of users at terminals and most application programs should remain unaffected when the internal representation of data is changed and even when some aspects of the external representation are changed. Changes in data representation will often be needed as a result of changes in query, update, and report traffic and natural growth in the types of stored information. Existing noninferential, formatted data systems provide users with tree-structured files or slightly more general network models of the data. In Section 1, inadequacies of these models are discussed. A model based on n -ary relations, a normal form for data base relations, and the concept of a universal data sublanguage are introduced. In Section 2, certain operations on relations (other than logical inference) are discussed and applied to the problems of redundancy and consistency in the user's model.", "score": 1.5, "mention_count": 4, "user_count": 4, "months": ["2021-01", "2022-05", "2022-07", "2024-05"], "last_mentioned": "2024-05-14T09:35:49Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1811.12231", "slug": "ffc1433c90e9a337", "title": "ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness", "authors": ["Robert Geirhos", "Patricia Rubisch", "Claudio Michaelis", "Matthias Bethge", "Felix A. Wichmann", "Wieland Brendel"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1811.12231", "doi": null, "topics": ["Psychology", "Computer vision", "Machine learning"], "description": "Convolutional Neural Networks (CNNs) are commonly thought to recognise objects by learning increasingly complex representations of object shapes.", "description_method": "abstract-excerpt", "abstract": "Convolutional Neural Networks (CNNs) are commonly thought to recognise objects by learning increasingly complex representations of object shapes. Some recent studies suggest a more important role of image textures. We here put these conflicting hypotheses to a quantitative test by evaluating CNNs and human observers on images with a texture-shape cue conflict. We show that ImageNet-trained CNNs are strongly biased towards recognising textures rather than shapes, which is in stark contrast to human behavioural evidence and reveals fundamentally different classification strategies. We then demonstrate that the same standard architecture (ResNet-50) that learns a texture-based representation on ImageNet is able to learn a shape-based representation instead when trained on \"Stylized-ImageNet\", a stylized version of ImageNet. This provides a much better fit for human behavioural performance in our well-controlled psychophysical lab setting (nine experiments totalling 48,560 psychophysical trials across 97 observers) and comes with a number of unexpected emergent benefits such as improved object detection performance and previously unseen robustness towards a wide range of image distortions, highlighting advantages of a shape-based representation.", "score": 1.5, "mention_count": 3, "user_count": 3, "months": ["2019-09", "2021-02", "2024-05"], "last_mentioned": "2024-05-27T19:03:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:math.ho/9404236", "slug": "6ba9de03f90063d2", "title": "On proof and progress in mathematics", "authors": ["Thurston, William P."], "year": 1994, "venue": "arXiv", "url": "https://arxiv.org/abs/math.ho/9404236", "doi": null, "topics": ["Mathematics"], "description": "In response to Jaffe and Quinn [math.HO/9307227], the author discusses forms of progress in mathematics that are not captured by formal proofs of theorems, especially in his own work in the theory of foliations and geometrization of 3-manifolds and dynamical systems.", "description_method": "abstract-excerpt", "abstract": "In response to Jaffe and Quinn [math.HO/9307227], the author discusses forms of progress in mathematics that are not captured by formal proofs of theorems, especially in his own work in the theory of foliations and geometrization of 3-manifolds and dynamical systems.", "score": 1.5, "mention_count": 3, "user_count": 3, "months": ["2009-11", "2012-08", "2014-11"], "last_mentioned": "2014-11-30T02:44:50Z", "metadata_source": "publisher-citation-tags", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41437-020-00394-6", "slug": "47ae6e8890f67b5f", "title": "The outstanding scientist, R.A. Fisher: his views on eugenics and race", "authors": ["Walter Bodmer", "R. A. Bailey", "Brian Charlesworth", "Adam Eyre-Walker", "Vernon Farewell", "Andrew Mead", "Stephen Senn"], "year": 2021, "venue": "Heredity", "url": "https://doi.org/10.1038/s41437-020-00394-6", "doi": "10.1038/s41437-020-00394-6", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 1.5, "mention_count": 3, "user_count": 3, "months": ["2021-12", "2023-06", "2024-04"], "last_mentioned": "2024-04-10T20:43:34Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s42003-019-0455-x", "slug": "c64859a160fdeeb2", "title": "Therapeutic benefit of combining calorie-restricted ketogenic diet and glutamine targeting in late-stage experimental glioblastoma", "authors": ["Purna Mukherjee", "Zachary M. Augur", "Mingyi Li", "Collin Hill", "Bennett Greenwood", "Marek A. Domin", "Gramoz Kondakci", "Niven R. Narain", "Michael A. Kiebish", "Roderick T. Bronson", "Gabriel Arismendi-Morillo", "Christos Chinopoulos", "Thomas N. Seyfried"], "year": 2019, "venue": "Communications Biology", "url": "https://doi.org/10.1038/s42003-019-0455-x", "doi": "10.1038/s42003-019-0455-x", "topics": ["Neuroscience", "Biology & medicine"], "description": "AbstractGlioblastoma (GBM) is an aggressive primary human brain tumour that has resisted effective therapy for decades.", "description_method": "abstract-excerpt", "abstract": "AbstractGlioblastoma (GBM) is an aggressive primary human brain tumour that has resisted effective therapy for decades. Although glucose and glutamine are the major fuels that drive GBM growth and invasion, few studies have targeted these fuels for therapeutic management. The glutamine antagonist, 6-diazo-5-oxo-L-norleucine (DON), was administered together with a calorically restricted ketogenic diet (KD-R) to treat late-stage orthotopic growth in two syngeneic GBM mouse models: VM-M3 and CT-2A. DON targets glutaminolysis, while the KD-R reduces glucose and, simultaneously, elevates neuroprotective and non-fermentable ketone bodies. The diet/drug therapeutic strategy killed tumour cells while reversing disease symptoms, and improving overall mouse survival. The therapeutic strategy also reduces edema, hemorrhage, and inflammation. Moreover, the KD-R diet facilitated DON delivery to the brain and allowed a lower dosage to achieve therapeutic effect. The findings support the importance of glucose and glutamine in driving GBM growth and provide a therapeutic strategy for non-toxic metabolic management.", "score": 1.5, "mention_count": 3, "user_count": 3, "months": ["2019-10", "2024-11", "2025-01"], "last_mentioned": "2025-01-12T05:01:19Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/1065010.1065042", "slug": "380e471e9cf94941", "title": "Threads cannot be implemented as a library", "authors": ["Hans-J. Boehm"], "year": 2005, "venue": "Proceedings of the 2005 ACM SIGPLAN conference on Programming language design and implementation", "url": "https://doi.org/10.1145/1065010.1065042", "doi": "10.1145/1065010.1065042", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 1.5, "mention_count": 3, "user_count": 3, "months": ["2020-05", "2021-02", "2022-10"], "last_mentioned": "2022-10-07T15:02:12Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/1238844.1238856", "slug": "df4fe1bc48b28a88", "title": "A history of Haskell", "authors": ["Paul Hudak", "John Hughes", "Simon Peyton Jones", "Philip Wadler"], "year": 2007, "venue": "Proceedings of the third ACM SIGPLAN conference on History of programming languages", "url": "https://doi.org/10.1145/1238844.1238856", "doi": "10.1145/1238844.1238856", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 1.5, "mention_count": 3, "user_count": 3, "months": ["2020-02", "2023-10", "2026-02"], "last_mentioned": "2026-02-17T19:40:40Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/45059.214411", "slug": "4cca3207b01ba244", "title": "Performance effects of architectural complexity in the Intel 432", "authors": ["Robert P. Colwell", "Edward F. Gehringer", "E. Douglas Jensen"], "year": 1988, "venue": "ACM Transactions on Computer Systems", "url": "https://doi.org/10.1145/45059.214411", "doi": "10.1145/45059.214411", "topics": ["Algorithms", "Programming languages"], "description": "The Intel 432 is noteworthy as an architecture incorporating a large amount of functionality that most other systems perform by software.", "description_method": "abstract-excerpt", "abstract": "The Intel 432 is noteworthy as an architecture incorporating a large amount of functionality that most other systems perform by software. It has, in effect, “migrated” this functionality from the software into the microcode and hardware. The benefits of functional migration have recently been a subject of intense controversy, with critics claiming that a complex architecture is inherently less efficient than a simple architecture with good software support. This paper examines the performance impact of the incorporation of several kinds of functionality into the Intel 432. Among these are the addressing structure, the caches, instruction alignment, the buses, and the way that garbage collection is handled. A set of several benchmarks is used to quantify the performance effect of each of these decisions. The results indicate that the 432 could have been speeded up very significantly if a small number of implementation decisions had been made differently, and if incrementally better technology had been used in its construction. Even with these modifications, however, the 432 would still have only one-fourth to one times the speed of its contemporaries. These figures may represent the real cost of the 432's style of object-based programming environment.", "score": 1.5, "mention_count": 3, "user_count": 3, "months": ["2020-06", "2023-11", "2026-01"], "last_mentioned": "2026-01-05T13:03:09Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2303.08774", "slug": "c14c2ea07b951d1d", "title": "GPT-4 Technical Report", "authors": [" OpenAI", "Josh Achiam", "Steven Adler", "Sandhini Agarwal", "Lama Ahmad", "Ilge Akkaya", "Florencia Leoni Aleman", "Diogo Almeida", "Janko Altenschmidt", "Sam Altman", "Shyamal Anadkat", "Red Avila", "Igor Babuschkin", "Suchir Balaji", "Valerie Balcom", "Paul Baltescu", "Haiming Bao", "Mohammad Bavarian", "Jeff Belgum", "Irwan Bello", "Jake Berdine", "Gabriel Bernadett-Shapiro", "Christopher Berner", "Lenny Bogdonoff", "Oleg Boiko", "Madelaine Boyd", "Anna-Luisa Brakman", "Greg Brockman", "Tim Brooks", "Miles Brundage", "Kevin Button", "Trevor Cai", "Rosie Campbell", "Andrew Cann", "Brittany Carey", "Chelsea Carlson", "Rory Carmichael", "Brooke Chan", "Che Chang", "Fotis Chantzis", "Derek Chen", "Sully Chen", "Ruby Chen", "Jason Chen", "Mark Chen", "Ben Chess", "Chester Cho", "Casey Chu", "Hyung Won Chung", "Dave Cummings", "Jeremiah Currier", "Yunxing Dai", "Cory Decareaux", "Thomas Degry", "Noah Deutsch", "Damien Deville", "Arka Dhar", "David Dohan", "Steve Dowling", "Sheila Dunning", "Adrien Ecoffet", "Atty Eleti", "Tyna Eloundou", "David Farhi", "Liam Fedus", "Niko Felix", "Simón Posada Fishman", "Juston Forte", "Isabella Fulford", "Leo Gao", "Elie Georges", "Christian Gibson", "Vik Goel", "Tarun Gogineni", "Gabriel Goh", "Rapha Gontijo-Lopes", "Jonathan Gordon", "Morgan Grafstein", "Scott Gray", "Ryan Greene", "Joshua Gross", "Shixiang Shane Gu", "Yufei Guo", "Chris Hallacy", "Jesse Han", "Jeff Harris", "Yuchen He", "Mike Heaton", "Johannes Heidecke", "Chris Hesse", "Alan Hickey", "Wade Hickey", "Peter Hoeschele", "Brandon Houghton", "Kenny Hsu", "Shengli Hu", "Xin Hu", "Joost Huizinga", "Shantanu Jain", "Shawn Jain", "Joanne Jang", "Angela Jiang", "Roger Jiang", "Haozhun Jin", "Denny Jin", "Shino Jomoto", "Billie Jonn", "Heewoo Jun", "Tomer Kaftan", "Łukasz Kaiser", "Ali Kamali", "Ingmar Kanitscheider", "Nitish Shirish Keskar", "Tabarak Khan", "Logan Kilpatrick", "Jong Wook Kim", "Christina Kim", "Yongjik Kim", "Jan Hendrik Kirchner", "Jamie Kiros", "Matt Knight", "Daniel Kokotajlo", "Łukasz Kondraciuk", "Andrew Kondrich", "Aris Konstantinidis", "Kyle Kosic", "Gretchen Krueger", "Vishal Kuo", "Michael Lampe", "Ikai Lan", "Teddy Lee", "Jan Leike", "Jade Leung", "Daniel Levy", "Chak Ming Li", "Rachel Lim", "Molly Lin", "Stephanie Lin", "Mateusz Litwin", "Theresa Lopez", "Ryan Lowe", "Patricia Lue", "Anna Makanju", "Kim Malfacini", "Sam Manning", "Todor Markov", "Yaniv Markovski", "Bianca Martin", "Katie Mayer", "Andrew Mayne", "Bob McGrew", "Scott Mayer McKinney", "Christine McLeavey", "Paul McMillan", "Jake McNeil", "David Medina", "Aalok Mehta", "Jacob Menick", "Luke Metz", "Andrey Mishchenko", "Pamela Mishkin", "Vinnie Monaco", "Evan Morikawa", "Daniel Mossing", "Tong Mu", "Mira Murati", "Oleg Murk", "David Mély", "Ashvin Nair", "Reiichiro Nakano", "Rajeev Nayak", "Arvind Neelakantan", "Richard Ngo", "Hyeonwoo Noh", "Long Ouyang", "Cullen O'Keefe", "Jakub Pachocki", "Alex Paino", "Joe Palermo", "Ashley Pantuliano", "Giambattista Parascandolo", "Joel Parish", "Emy Parparita", "Alex Passos", "Mikhail Pavlov", "Andrew Peng", "Adam Perelman", "Filipe de Avila Belbute Peres", "Michael Petrov", "Henrique Ponde de Oliveira Pinto", " Michael", " Pokorny", "Michelle Pokrass", "Vitchyr H. Pong", "Tolly Powell", "Alethea Power", "Boris Power", "Elizabeth Proehl", "Raul Puri", "Alec Radford", "Jack Rae", "Aditya Ramesh", "Cameron Raymond", "Francis Real", "Kendra Rimbach", "Carl Ross", "Bob Rotsted", "Henri Roussez", "Nick Ryder", "Mario Saltarelli", "Ted Sanders", "Shibani Santurkar", "Girish Sastry", "Heather Schmidt", "David Schnurr", "John Schulman", "Daniel Selsam", "Kyla Sheppard", "Toki Sherbakov", "Jessica Shieh", "Sarah Shoker", "Pranav Shyam", "Szymon Sidor", "Eric Sigler", "Maddie Simens", "Jordan Sitkin", "Katarina Slama", "Ian Sohl", "Benjamin Sokolowsky", "Yang Song", "Natalie Staudacher", "Felipe Petroski Such", "Natalie Summers", "Ilya Sutskever", "Jie Tang", "Nikolas Tezak", "Madeleine B. Thompson", "Phil Tillet", "Amin Tootoonchian", "Elizabeth Tseng", "Preston Tuggle", "Nick Turley", "Jerry Tworek", "Juan Felipe Cerón Uribe", "Andrea Vallone", "Arun Vijayvergiya", "Chelsea Voss", "Carroll Wainwright", "Justin Jay Wang", "Alvin Wang", "Ben Wang", "Jonathan Ward", "Jason Wei", "CJ Weinmann", "Akila Welihinda", "Peter Welinder", "Jiayi Weng", "Lilian Weng", "Matt Wiethoff", "Dave Willner", "Clemens Winter", "Samuel Wolrich", "Hannah Wong", "Lauren Workman", "Sherwin Wu", "Jeff Wu", "Michael Wu", "Kai Xiao", "Tao Xu", "Sarah Yoo", "Kevin Yu", "Qiming Yuan", "Wojciech Zaremba", "Rowan Zellers", "Chong Zhang", "Marvin Zhang", "Shengjia Zhao", "Tianhao Zheng", "Juntang Zhuang", "William Zhuk", "Barret Zoph"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2303.08774", "doi": null, "topics": ["Language models", "Psychology", "Algorithms"], "description": "We report the development of GPT-4, a large-scale, multimodal model which can accept image and text inputs and produce text outputs.", "description_method": "abstract-excerpt", "abstract": "We report the development of GPT-4, a large-scale, multimodal model which can accept image and text inputs and produce text outputs. While less capable than humans in many real-world scenarios, GPT-4 exhibits human-level performance on various professional and academic benchmarks, including passing a simulated bar exam with a score around the top 10% of test takers. GPT-4 is a Transformer-based model pre-trained to predict the next token in a document. The post-training alignment process results in improved performance on measures of factuality and adherence to desired behavior. A core component of this project was developing infrastructure and optimization methods that behave predictably across a wide range of scales. This allowed us to accurately predict some aspects of GPT-4's performance based on models trained with no more than 1/1,000th the compute of GPT-4.", "score": 1.0, "mention_count": 45, "user_count": 29, "months": ["2023-03", "2023-04", "2023-05", "2023-06", "2023-07", "2023-08", "2023-11", "2024-01", "2024-02", "2024-03", "2024-04", "2024-05", "2024-06", "2024-07", "2025-01", "2025-02", "2025-05", "2026-03"], "last_mentioned": "2026-03-01T09:43:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2201.11903", "slug": "6cc9dfe0c8911536", "title": "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models", "authors": ["Jason Wei", "Xuezhi Wang", "Dale Schuurmans", "Maarten Bosma", "Brian Ichter", "Fei Xia", "Ed Chi", "Quoc Le", "Denny Zhou"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2201.11903", "doi": null, "topics": ["Language models"], "description": "We explore how generating a chain of thought -- a series of intermediate reasoning steps -- significantly improves the ability of large language models to perform complex reasoning.", "description_method": "abstract-excerpt", "abstract": "We explore how generating a chain of thought -- a series of intermediate reasoning steps -- significantly improves the ability of large language models to perform complex reasoning. In particular, we show how such reasoning abilities emerge naturally in sufficiently large language models via a simple method called chain of thought prompting, where a few chain of thought demonstrations are provided as exemplars in prompting. Experiments on three large language models show that chain of thought prompting improves performance on a range of arithmetic, commonsense, and symbolic reasoning tasks. The empirical gains can be striking. For instance, prompting a 540B-parameter language model with just eight chain of thought exemplars achieves state of the art accuracy on the GSM8K benchmark of math word problems, surpassing even finetuned GPT-3 with a verifier.", "score": 1.0, "mention_count": 40, "user_count": 34, "months": ["2022-04", "2022-05", "2022-07", "2022-10", "2022-12", "2023-01", "2023-02", "2023-03", "2023-04", "2023-05", "2023-07", "2023-08", "2023-11", "2023-12", "2024-01", "2024-05", "2024-08", "2024-09", "2024-10", "2025-01", "2025-05", "2025-10", "2026-02", "2026-05", "2026-06", "2026-08"], "last_mentioned": "2026-08-01T09:22:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.07759", "slug": "1edf7d87f9ba730e", "title": "TinyStories: How Small Can Language Models Be and Still Speak Coherent English?", "authors": ["Ronen Eldan", "Yuanzhi Li"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.07759", "doi": null, "topics": ["Language models", "Distributed systems"], "description": "Language models (LMs) are powerful tools for natural language processing, but they often struggle to produce coherent and fluent text when they are small.", "description_method": "abstract-excerpt", "abstract": "Language models (LMs) are powerful tools for natural language processing, but they often struggle to produce coherent and fluent text when they are small. Models with around 125M parameters such as GPT-Neo (small) or GPT-2 (small) can rarely generate coherent and consistent English text beyond a few words even after extensive training. This raises the question of whether the emergence of the ability to produce coherent English text only occurs at larger scales (with hundreds of millions of parameters or more) and complex architectures (with many layers of global attention). In this work, we introduce TinyStories, a synthetic dataset of short stories that only contain words that a typical 3 to 4-year-olds usually understand, generated by GPT-3.5 and GPT-4. We show that TinyStories can be used to train and evaluate LMs that are much smaller than the state-of-the-art models (below 10 million total parameters), or have much simpler architectures (with only one transformer block), yet still produce fluent and consistent stories with several paragraphs that are diverse and have almost perfect grammar, and demonstrate reasoning capabilities. We also introduce a new paradigm for the evaluation of language models: We suggest a framework which uses GPT-4 to grade the content generated by these models as if those were stories written by students and graded by a (human) teacher. This new paradigm overcomes the flaws of standard benchmarks which often requires the model's output to be very structures, and moreover provides a multidimensional score for the model, providing scores for different capabilities such as grammar, creativity and consistency. We hope that TinyStories can facilitate the development, analysis and research of LMs, especially for low-resource or specialized domains, and shed light on the emergence of language capabilities in LMs.", "score": 1.0, "mention_count": 32, "user_count": 21, "months": ["2023-05", "2023-06", "2023-07", "2023-08", "2023-09", "2023-10", "2024-01", "2024-05", "2024-12", "2025-01", "2025-03", "2025-08", "2025-11", "2025-12"], "last_mentioned": "2025-12-09T12:51:33Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1911.01547", "slug": "1a4e71b1fe32d70e", "title": "On the Measure of Intelligence", "authors": ["François Chollet"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1911.01547", "doi": null, "topics": ["Psychology", "Machine learning", "Statistics & probability"], "description": "To make deliberate progress towards more intelligent and more human-like artificial systems, we need to be following an appropriate feedback signal: we need to be able to define and evaluate intelligence in a way that enables comparisons between two systems, as well as comparisons with humans.", "description_method": "abstract-excerpt", "abstract": "To make deliberate progress towards more intelligent and more human-like artificial systems, we need to be following an appropriate feedback signal: we need to be able to define and evaluate intelligence in a way that enables comparisons between two systems, as well as comparisons with humans. Over the past hundred years, there has been an abundance of attempts to define and measure intelligence, across both the fields of psychology and AI. We summarize and critically assess these definitions and evaluation approaches, while making apparent the two historical conceptions of intelligence that have implicitly guided them. We note that in practice, the contemporary AI community still gravitates towards benchmarking intelligence by comparing the skill exhibited by AIs and humans at specific tasks such as board games and video games. We argue that solely measuring skill at any given task falls short of measuring intelligence, because skill is heavily modulated by prior knowledge and experience: unlimited priors or unlimited training data allow experimenters to \"buy\" arbitrary levels of skills for a system, in a way that masks the system's own generalization power. We then articulate a new formal definition of intelligence based on Algorithmic Information Theory, describing intelligence as skill-acquisition efficiency and highlighting the concepts of scope, generalization difficulty, priors, and experience. Using this definition, we propose a set of guidelines for what a general AI benchmark should look like. Finally, we present a benchmark closely following these guidelines, the Abstraction and Reasoning Corpus (ARC), built upon an explicit set of priors designed to be as close as possible to innate human priors. We argue that ARC can be used to measure a human-like form of general fluid intelligence and that it enables fair general intelligence comparisons between AI systems and humans.", "score": 1.0, "mention_count": 31, "user_count": 20, "months": ["2019-11", "2020-06", "2020-11", "2020-12", "2021-04", "2021-06", "2021-12", "2022-07", "2022-09", "2022-10", "2023-01", "2023-04", "2024-01", "2024-03", "2024-04", "2024-09", "2024-10", "2025-01", "2025-07", "2026-03", "2026-05", "2026-08"], "last_mentioned": "2026-08-12T20:00:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.03635", "slug": "a9d7f0235205afcc", "title": "The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks", "authors": ["Jonathan Frankle", "Michael Carbin"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.03635", "doi": null, "topics": ["Statistics & probability", "Machine learning", "Computer vision"], "description": "Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of inference without compromising accuracy.", "description_method": "abstract-excerpt", "abstract": "Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of inference without compromising accuracy. However, contemporary experience is that the sparse architectures produced by pruning are difficult to train from the start, which would similarly improve training performance. We find that a standard pruning technique naturally uncovers subnetworks whose initializations made them capable of training effectively. Based on these results, we articulate the \"lottery ticket hypothesis:\" dense, randomly-initialized, feed-forward networks contain subnetworks (\"winning tickets\") that - when trained in isolation - reach test accuracy comparable to the original network in a similar number of iterations. The winning tickets we find have won the initialization lottery: their connections have initial weights that make training particularly effective. We present an algorithm to identify winning tickets and a series of experiments that support the lottery ticket hypothesis and the importance of these fortuitous initializations. We consistently find winning tickets that are less than 10-20% of the size of several fully-connected and convolutional feed-forward architectures for MNIST and CIFAR10. Above this size, the winning tickets that we find learn faster than the original network and reach higher test accuracy.", "score": 1.0, "mention_count": 29, "user_count": 27, "months": ["2019-05", "2019-06", "2019-08", "2020-03", "2020-05", "2020-09", "2020-12", "2021-07", "2021-09", "2022-01", "2022-02", "2023-01", "2023-05", "2023-06", "2023-11", "2025-01", "2025-02", "2025-08", "2025-11", "2026-04"], "last_mentioned": "2026-04-24T01:15:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2206.07682", "slug": "6710053fa7256089", "title": "Emergent Abilities of Large Language Models", "authors": ["Jason Wei", "Yi Tay", "Rishi Bommasani", "Colin Raffel", "Barret Zoph", "Sebastian Borgeaud", "Dani Yogatama", "Maarten Bosma", "Denny Zhou", "Donald Metzler", "Ed H. Chi", "Tatsunori Hashimoto", "Oriol Vinyals", "Percy Liang", "Jeff Dean", "William Fedus"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2206.07682", "doi": null, "topics": ["Language models"], "description": "Scaling up language models has been shown to predictably improve performance and sample efficiency on a wide range of downstream tasks.", "description_method": "abstract-excerpt", "abstract": "Scaling up language models has been shown to predictably improve performance and sample efficiency on a wide range of downstream tasks. This paper instead discusses an unpredictable phenomenon that we refer to as emergent abilities of large language models. We consider an ability to be emergent if it is not present in smaller models but is present in larger models. Thus, emergent abilities cannot be predicted simply by extrapolating the performance of smaller models. The existence of such emergence implies that additional scaling could further expand the range of capabilities of language models.", "score": 1.0, "mention_count": 25, "user_count": 23, "months": ["2023-01", "2023-02", "2023-03", "2023-04", "2023-05", "2023-07", "2023-08", "2023-10", "2024-02", "2024-03", "2024-04", "2024-05", "2025-02", "2025-05", "2026-03"], "last_mentioned": "2026-03-27T14:12:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2506.08872", "slug": "b0ce266edec81aee", "title": "Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task", "authors": ["Nataliya Kosmyna", "Eugene Hauptmann", "Ye Tong Yuan", "Jessica Situ", "Xian-Hao Liao", "Ashly Vivian Beresnitzky", "Iris Braunstein", "Pattie Maes"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2506.08872", "doi": null, "topics": ["Neuroscience", "Language models", "Psychology"], "description": "This study explores the neural and behavioral consequences of LLM-assisted essay writing.", "description_method": "abstract-excerpt", "abstract": "This study explores the neural and behavioral consequences of LLM-assisted essay writing. Participants were divided into three groups: LLM, Search Engine, and Brain-only (no tools). Each completed three sessions under the same condition. In a fourth session, LLM users were reassigned to Brain-only group (LLM-to-Brain), and Brain-only users were reassigned to LLM condition (Brain-to-LLM). A total of 54 participants took part in Sessions 1-3, with 18 completing session 4. We used electroencephalography (EEG) to assess cognitive load during essay writing, and analyzed essays using NLP, as well as scoring essays with the help from human teachers and an AI judge. Across groups, NERs, n-gram patterns, and topic ontology showed within-group homogeneity. EEG revealed significant differences in brain connectivity: Brain-only participants exhibited the strongest, most distributed networks; Search Engine users showed moderate engagement; and LLM users displayed the weakest connectivity. Cognitive activity scaled down in relation to external tool use. In session 4, LLM-to-Brain participants showed reduced alpha and beta connectivity, indicating under-engagement. Brain-to-LLM users exhibited higher memory recall and activation of occipito-parietal and prefrontal areas, similar to Search Engine users. Self-reported ownership of essays was the lowest in the LLM group and the highest in the Brain-only group. LLM users also struggled to accurately quote their own work. While LLMs offer immediate convenience, our findings highlight potential cognitive costs. Over four months, LLM users consistently underperformed at neural, linguistic, and behavioral levels. These results raise concerns about the long-term educational implications of LLM reliance and underscore the need for deeper inquiry into AI's role in learning.", "score": 1.0, "mention_count": 25, "user_count": 18, "months": ["2025-06", "2025-09", "2026-01", "2026-02", "2026-03", "2026-04", "2026-05", "2026-06", "2026-07"], "last_mentioned": "2026-07-12T21:02:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3442188.3445922", "slug": "67046cb9f186aeae", "title": "On the Dangers of Stochastic Parrots", "authors": ["Emily M. Bender", "Timnit Gebru", "Angelina McMillan-Major", "Shmargaret Shmitchell"], "year": 2021, "venue": "Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency", "url": "https://doi.org/10.1145/3442188.3445922", "doi": "10.1145/3442188.3445922", "topics": ["Statistics & probability"], "description": "", "description_method": null, "abstract": "", "score": 1.0, "mention_count": 25, "user_count": 22, "months": ["2021-07", "2021-12", "2022-06", "2022-11", "2022-12", "2023-01", "2023-02", "2023-03", "2023-04", "2023-05", "2023-12", "2024-11", "2025-05", "2025-10", "2025-12", "2026-06", "2026-07"], "last_mentioned": "2026-07-19T01:06:07Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2211.09066", "slug": "409c3f345c81d8f5", "title": "Teaching Algorithmic Reasoning via In-context Learning", "authors": ["Hattie Zhou", "Azade Nova", "Hugo Larochelle", "Aaron Courville", "Behnam Neyshabur", "Hanie Sedghi"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2211.09066", "doi": null, "topics": ["Algorithms", "Language models"], "description": "Large language models (LLMs) have shown increasing in-context learning capabilities through scaling up model and data size.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLMs) have shown increasing in-context learning capabilities through scaling up model and data size. Despite this progress, LLMs are still unable to solve algorithmic reasoning problems. While providing a rationale with the final answer has led to further improvements in multi-step reasoning problems, Anil et al. 2022 showed that even simple algorithmic reasoning tasks such as parity are far from solved. In this work, we identify and study four key stages for successfully teaching algorithmic reasoning to LLMs: (1) formulating algorithms as skills, (2) teaching multiple skills simultaneously (skill accumulation), (3) teaching how to combine skills (skill composition) and (4) teaching how to use skills as tools. We show that it is possible to teach algorithmic reasoning to LLMs via in-context learning, which we refer to as algorithmic prompting. We evaluate our approach on a variety of arithmetic and quantitative reasoning tasks, and demonstrate significant boosts in performance over existing prompting techniques. In particular, for long parity, addition, multiplication and subtraction, we achieve an error reduction of approximately 10x, 9x, 5x and 2x respectively compared to the best available baselines.", "score": 1.0, "mention_count": 24, "user_count": 6, "months": ["2022-12", "2023-03", "2023-04", "2023-05", "2023-06", "2023-07", "2023-08", "2023-10", "2023-12"], "last_mentioned": "2023-12-15T02:30:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1002.2284", "slug": "d68e8b2829abb6f4", "title": "Markets are efficient if and only if P = NP", "authors": ["Philip Maymin"], "year": 2010, "venue": "arXiv", "url": "https://arxiv.org/abs/1002.2284", "doi": null, "topics": ["Economics", "Algorithms"], "description": "I prove that if markets are weak-form efficient, meaning current prices fully reflect all information available in past prices, then P = NP, meaning every computational problem whose solution can be verified in polynomial time can also be solved in polynomial time.", "description_method": "abstract-excerpt", "abstract": "I prove that if markets are weak-form efficient, meaning current prices fully reflect all information available in past prices, then P = NP, meaning every computational problem whose solution can be verified in polynomial time can also be solved in polynomial time. I also prove the converse by showing how we can \"program\" the market to solve NP-complete problems. Since P probably does not equal NP, markets are probably not efficient. Specifically, markets become increasingly inefficient as the time series lengthens or becomes more frequent. An illustration by way of partitioning the excess returns to momentum strategies based on data availability confirms this prediction.", "score": 1.0, "mention_count": 23, "user_count": 18, "months": ["2010-08", "2011-04", "2014-07", "2014-08", "2014-11", "2016-04", "2018-08", "2018-12", "2019-04", "2019-08", "2019-09", "2020-01", "2021-02", "2021-04", "2021-07", "2021-10", "2022-03", "2023-08", "2024-04", "2024-06", "2024-09", "2024-10"], "last_mentioned": "2024-10-26T04:44:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/358198.358210", "slug": "1c7b82ee07ef6436", "title": "Reflections on trusting trust", "authors": ["Ken Thompson"], "year": 1984, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/358198.358210", "doi": "10.1145/358198.358210", "topics": ["Other research"], "description": "To what extent should one trust a statement that a program is free of Trojan horses?", "description_method": "abstract-excerpt", "abstract": "To what extent should one trust a statement that a program is free of Trojan horses? Perhaps it is more important to trust the people who wrote the software.", "score": 1.0, "mention_count": 23, "user_count": 22, "months": ["2020-07", "2020-09", "2020-11", "2021-06", "2021-08", "2022-04", "2022-05", "2022-06", "2022-07", "2022-10", "2022-12", "2023-02", "2024-03", "2024-08", "2024-09", "2025-01", "2025-06", "2026-08"], "last_mentioned": "2026-08-13T01:18:15Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-021-03647-4", "slug": "e7cb45032de35d4b", "title": "SARS-CoV-2 infection induces long-lived bone marrow plasma cells in humans", "authors": ["Jackson S. Turner", "Wooseob Kim", "Elizaveta Kalaidina", "Charles W. Goss", "Adriana M. Rauseo", "Aaron J. Schmitz", "Lena Hansen", "Alem Haile", "Michael K. Klebert", "Iskra Pusic", "Jane A. O’Halloran", "Rachel M. Presti", "Ali H. Ellebedy"], "year": 2021, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-021-03647-4", "doi": "10.1038/s41586-021-03647-4", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 1.0, "mention_count": 22, "user_count": 5, "months": ["2021-06", "2021-07", "2021-08", "2021-10", "2022-02"], "last_mentioned": "2022-02-24T20:29:52Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.6199", "slug": "76425b97d8c234a0", "title": "Intriguing properties of neural networks", "authors": ["Christian Szegedy", "Wojciech Zaremba", "Ilya Sutskever", "Joan Bruna", "Dumitru Erhan", "Ian Goodfellow", "Rob Fergus"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.6199", "doi": null, "topics": ["Machine learning", "Psychology", "Computer vision"], "description": "Deep neural networks are highly expressive models that have recently achieved state of the art performance on speech and visual recognition tasks.", "description_method": "abstract-excerpt", "abstract": "Deep neural networks are highly expressive models that have recently achieved state of the art performance on speech and visual recognition tasks. While their expressiveness is the reason they succeed, it also causes them to learn uninterpretable solutions that could have counter-intuitive properties. In this paper we report two such properties. First, we find that there is no distinction between individual high level units and random linear combinations of high level units, according to various methods of unit analysis. It suggests that it is the space, rather than the individual units, that contains of the semantic information in the high layers of neural networks. Second, we find that deep neural networks learn input-output mappings that are fairly discontinuous to a significant extend. We can cause the network to misclassify an image by applying a certain imperceptible perturbation, which is found by maximizing the network's prediction error. In addition, the specific nature of these perturbations is not a random artifact of learning: the same perturbation can cause a different network, that was trained on a different subset of the dataset, to misclassify the same input.", "score": 1.0, "mention_count": 21, "user_count": 20, "months": ["2014-04", "2014-09", "2014-11", "2014-12", "2015-01", "2015-03", "2016-03", "2016-05", "2016-10", "2016-11", "2017-04", "2017-07", "2017-10", "2017-11", "2017-12", "2019-12", "2021-03", "2022-01"], "last_mentioned": "2022-01-07T18:56:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2408.13687", "slug": "6ded1529debc3000", "title": "Quantum error correction below the surface code threshold", "authors": ["Rajeev Acharya", "Laleh Aghababaie-Beni", "Igor Aleiner", "Trond I. Andersen", "Markus Ansmann", "Frank Arute", "Kunal Arya", "Abraham Asfaw", "Nikita Astrakhantsev", "Juan Atalaya", "Ryan Babbush", "Dave Bacon", "Brian Ballard", "Joseph C. Bardin", "Johannes Bausch", "Andreas Bengtsson", "Alexander Bilmes", "Sam Blackwell", "Sergio Boixo", "Gina Bortoli", "Alexandre Bourassa", "Jenna Bovaird", "Leon Brill", "Michael Broughton", "David A. Browne", "Brett Buchea", "Bob B. Buckley", "David A. Buell", "Tim Burger", "Brian Burkett", "Nicholas Bushnell", "Anthony Cabrera", "Juan Campero", "Hung-Shen Chang", "Yu Chen", "Zijun Chen", "Ben Chiaro", "Desmond Chik", "Charina Chou", "Jahan Claes", "Agnetta Y. Cleland", "Josh Cogan", "Roberto Collins", "Paul Conner", "William Courtney", "Alexander L. Crook", "Ben Curtin", "Sayan Das", "Alex Davies", "Laura De Lorenzo", "Dripto M. Debroy", "Sean Demura", "Michel Devoret", "Agustin Di Paolo", "Paul Donohoe", "Ilya Drozdov", "Andrew Dunsworth", "Clint Earle", "Thomas Edlich", "Alec Eickbusch", "Aviv Moshe Elbag", "Mahmoud Elzouka", "Catherine Erickson", "Lara Faoro", "Edward Farhi", "Vinicius S. Ferreira", "Leslie Flores Burgos", "Ebrahim Forati", "Austin G. Fowler", "Brooks Foxen", "Suhas Ganjam", "Gonzalo Garcia", "Robert Gasca", "Élie Genois", "William Giang", "Craig Gidney", "Dar Gilboa", "Raja Gosula", "Alejandro Grajales Dau", "Dietrich Graumann", "Alex Greene", "Jonathan A. Gross", "Steve Habegger", "John Hall", "Michael C. Hamilton", "Monica Hansen", "Matthew P. Harrigan", "Sean D. Harrington", "Francisco J. H. Heras", "Stephen Heslin", "Paula Heu", "Oscar Higgott", "Gordon Hill", "Jeremy Hilton", "George Holland", "Sabrina Hong", "Hsin-Yuan Huang", "Ashley Huff", "William J. Huggins", "Lev B. Ioffe", "Sergei V. Isakov", "Justin Iveland", "Evan Jeffrey", "Zhang Jiang", "Cody Jones", "Stephen Jordan", "Chaitali Joshi", "Pavol Juhas", "Dvir Kafri", "Hui Kang", "Amir H. Karamlou", "Kostyantyn Kechedzhi", "Julian Kelly", "Trupti Khaire", "Tanuj Khattar", "Mostafa Khezri", "Seon Kim", "Paul V. Klimov", "Andrey R. Klots", "Bryce Kobrin", "Pushmeet Kohli", "Alexander N. Korotkov", "Fedor Kostritsa", "Robin Kothari", "Borislav Kozlovskii", "John Mark Kreikebaum", "Vladislav D. Kurilovich", "Nathan Lacroix", "David Landhuis", "Tiano Lange-Dei", "Brandon W. Langley", "Pavel Laptev", "Kim-Ming Lau", "Loïck Le Guevel", "Justin Ledford", "Kenny Lee", "Yuri D. Lensky", "Shannon Leon", "Brian J. Lester", "Wing Yan Li", "Yin Li", "Alexander T. Lill", "Wayne Liu", "William P. Livingston", "Aditya Locharla", "Erik Lucero", "Daniel Lundahl", "Aaron Lunt", "Sid Madhuk", "Fionn D. Malone", "Ashley Maloney", "Salvatore Mandrá", "Leigh S. Martin", "Steven Martin", "Orion Martin", "Cameron Maxfield", "Jarrod R. McClean", "Matt McEwen", "Seneca Meeks", "Anthony Megrant", "Xiao Mi", "Kevin C. Miao", "Amanda Mieszala", "Reza Molavi", "Sebastian Molina", "Shirin Montazeri", "Alexis Morvan", "Ramis Movassagh", "Wojciech Mruczkiewicz", "Ofer Naaman", "Matthew Neeley", "Charles Neill", "Ani Nersisyan", "Hartmut Neven", "Michael Newman", "Jiun How Ng", "Anthony Nguyen", "Murray Nguyen", "Chia-Hung Ni", "Thomas E. O'Brien", "William D. Oliver", "Alex Opremcak", "Kristoffer Ottosson", "Andre Petukhov", "Alex Pizzuto", "John Platt", "Rebecca Potter", "Orion Pritchard", "Leonid P. Pryadko", "Chris Quintana", "Ganesh Ramachandran", "Matthew J. Reagor", "David M. Rhodes", "Gabrielle Roberts", "Eliott Rosenberg", "Emma Rosenfeld", "Pedram Roushan", "Nicholas C. Rubin", "Negar Saei", "Daniel Sank", "Kannan Sankaragomathi", "Kevin J. Satzinger", "Henry F. Schurkus", "Christopher Schuster", "Andrew W. Senior", "Michael J. Shearn", "Aaron Shorter", "Noah Shutty", "Vladimir Shvarts", "Shraddha Singh", "Volodymyr Sivak", "Jindra Skruzny", "Spencer Small", "Vadim Smelyanskiy", "W. Clarke Smith", "Rolando D. Somma", "Sofia Springer", "George Sterling", "Doug Strain", "Jordan Suchard", "Aaron Szasz", "Alex Sztein", "Douglas Thor", "Alfredo Torres", "M. Mert Torunbalci", "Abeer Vaishnav", "Justin Vargas", "Sergey Vdovichev", "Guifre Vidal", "Benjamin Villalonga", "Catherine Vollgraff Heidweiller", "Steven Waltman", "Shannon X. Wang", "Brayden Ware", "Kate Weber", "Theodore White", "Kristi Wong", "Bryan W. K. Woo", "Cheng Xing", "Z. Jamie Yao", "Ping Yeh", "Bicheng Ying", "Juhwan Yoo", "Noureldin Yosri", "Grayson Young", "Adam Zalcman", "Yaxing Zhang", "Ningfeng Zhu", "Nicholas Zobrist"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2408.13687", "doi": "10.1038/s41586-024-08449-y", "topics": ["Physics", "Distributed systems", "Algorithms"], "description": "Quantum error correction provides a path to reach practical quantum computing by combining multiple physical qubits into a logical qubit, where the logical error rate is suppressed exponentially as more qubits are added.", "description_method": "abstract-excerpt", "abstract": "Quantum error correction provides a path to reach practical quantum computing by combining multiple physical qubits into a logical qubit, where the logical error rate is suppressed exponentially as more qubits are added. However, this exponential suppression only occurs if the physical error rate is below a critical threshold. In this work, we present two surface code memories operating below this threshold: a distance-7 code and a distance-5 code integrated with a real-time decoder. The logical error rate of our larger quantum memory is suppressed by a factor of $Λ$ = 2.14 $\\pm$ 0.02 when increasing the code distance by two, culminating in a 101-qubit distance-7 code with 0.143% $\\pm$ 0.003% error per cycle of error correction. This logical memory is also beyond break-even, exceeding its best physical qubit's lifetime by a factor of 2.4 $\\pm$ 0.3. We maintain below-threshold performance when decoding in real time, achieving an average decoder latency of 63 $μ$s at distance-5 up to a million cycles, with a cycle time of 1.1 $μ$s. To probe the limits of our error-correction performance, we run repetition codes up to distance-29 and find that logical performance is limited by rare correlated error events occurring approximately once every hour, or 3 $\\times$ 10$^9$ cycles. Our results present device performance that, if scaled, could realize the operational requirements of large scale fault-tolerant quantum algorithms.", "score": 1.0, "mention_count": 19, "user_count": 14, "months": ["2024-09", "2024-11", "2024-12", "2025-01", "2025-02", "2025-07", "2025-08", "2025-10", "2025-12", "2026-04"], "last_mentioned": "2026-04-07T00:25:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2507.09089", "slug": "3066d94eafbf6019", "title": "Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity", "authors": ["Joel Becker", "Nate Rush", "Elizabeth Barnes", "David Rein"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2507.09089", "doi": null, "topics": ["Economics"], "description": "Despite widespread adoption, the impact of AI tools on software development in the wild remains understudied.", "description_method": "abstract-excerpt", "abstract": "Despite widespread adoption, the impact of AI tools on software development in the wild remains understudied. We conduct a randomized controlled trial (RCT) to understand how AI tools at the February-June 2025 frontier affect the productivity of experienced open-source developers. 16 developers with moderate AI experience complete 246 tasks in mature projects on which they have an average of 5 years of prior experience. Each task is randomly assigned to allow or disallow usage of early 2025 AI tools. When AI tools are allowed, developers primarily use Cursor Pro, a popular code editor, and Claude 3.5/3.7 Sonnet. Before starting tasks, developers forecast that allowing AI will reduce completion time by 24%. After completing the study, developers estimate that allowing AI reduced completion time by 20%. Surprisingly, we find that allowing AI actually increases completion time by 19%--AI tooling slowed developers down. This slowdown also contradicts predictions from experts in economics (39% shorter) and ML (38% shorter). To understand this result, we collect and evaluate evidence for 20 properties of our setting that a priori could contribute to the observed slowdown effect--for example, the size and quality standards of projects, or prior developer experience with AI tooling. Although the influence of experimental artifacts cannot be entirely ruled out, the robustness of the slowdown effect across our analyses suggests it is unlikely to primarily be a function of our experimental design.", "score": 1.0, "mention_count": 19, "user_count": 19, "months": ["2025-07", "2025-08", "2025-09", "2025-10", "2025-11", "2025-12", "2026-01", "2026-02"], "last_mentioned": "2026-02-28T12:04:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1512.03385", "slug": "4627097f8362928c", "title": "Deep Residual Learning for Image Recognition", "authors": ["Kaiming He", "Xiangyu Zhang", "Shaoqing Ren", "Jian Sun"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1512.03385", "doi": null, "topics": ["Computer vision", "Psychology", "Machine learning"], "description": "Deeper neural networks are more difficult to train.", "description_method": "abstract-excerpt", "abstract": "Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.", "score": 1.0, "mention_count": 18, "user_count": 17, "months": ["2015-12", "2016-03", "2016-04", "2016-06", "2016-09", "2017-11", "2018-06", "2020-08", "2020-11", "2021-04", "2021-06", "2023-11", "2025-05", "2026-07"], "last_mentioned": "2026-07-07T19:03:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2106.09685", "slug": "b85d9dbf9c291e18", "title": "LoRA: Low-Rank Adaptation of Large Language Models", "authors": ["Edward J. Hu", "Yelong Shen", "Phillip Wallis", "Zeyuan Allen-Zhu", "Yuanzhi Li", "Shean Wang", "Lu Wang", "Weizhu Chen"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2106.09685", "doi": null, "topics": ["Language models"], "description": "An important paradigm of natural language processing consists of large-scale pre-training on general domain data and adaptation to particular tasks or domains.", "description_method": "abstract-excerpt", "abstract": "An important paradigm of natural language processing consists of large-scale pre-training on general domain data and adaptation to particular tasks or domains. As we pre-train larger models, full fine-tuning, which retrains all model parameters, becomes less feasible. Using GPT-3 175B as an example -- deploying independent instances of fine-tuned models, each with 175B parameters, is prohibitively expensive. We propose Low-Rank Adaptation, or LoRA, which freezes the pre-trained model weights and injects trainable rank decomposition matrices into each layer of the Transformer architecture, greatly reducing the number of trainable parameters for downstream tasks. Compared to GPT-3 175B fine-tuned with Adam, LoRA can reduce the number of trainable parameters by 10,000 times and the GPU memory requirement by 3 times. LoRA performs on-par or better than fine-tuning in model quality on RoBERTa, DeBERTa, GPT-2, and GPT-3, despite having fewer trainable parameters, a higher training throughput, and, unlike adapters, no additional inference latency. We also provide an empirical investigation into rank-deficiency in language model adaptation, which sheds light on the efficacy of LoRA. We release a package that facilitates the integration of LoRA with PyTorch models and provide our implementations and model checkpoints for RoBERTa, DeBERTa, and GPT-2 at https://github.com/microsoft/LoRA.", "score": 1.0, "mention_count": 18, "user_count": 17, "months": ["2023-03", "2023-04", "2023-05", "2023-09", "2023-11", "2023-12", "2024-04", "2024-06", "2024-11", "2025-06", "2026-08"], "last_mentioned": "2026-08-16T13:59:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.01421", "slug": "dc029a7289cc8f17", "title": "Struggles with the Continuum", "authors": ["John C. Baez"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.01421", "doi": null, "topics": ["Physics", "Mathematics"], "description": "Our assumption that spacetime is a continuum leads to many challenges in mathematical physics.", "description_method": "abstract-excerpt", "abstract": "Our assumption that spacetime is a continuum leads to many challenges in mathematical physics. Singularities, divergent integrals and the like threaten many of our favorite theories, from Newtonian gravity to classical electrodynamics, quantum electrodynamics and the Standard Model. In general relativity, singularities are intimately connected to some of the theory's most dramatic successful predictions. We survey these problems and the large amount of work that has gone into dealing with them.", "score": 1.0, "mention_count": 16, "user_count": 6, "months": ["2016-09", "2017-01", "2017-04", "2021-07", "2021-09", "2022-11", "2023-01", "2023-02", "2023-07", "2023-12", "2024-03", "2024-11", "2025-11"], "last_mentioned": "2025-11-26T17:07:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.03453", "slug": "e444373d4e279f2b", "title": "The Surprising Creativity of Digital Evolution: A Collection of Anecdotes from the Evolutionary Computation and Artificial Life Research Communities", "authors": ["Joel Lehman", "Jeff Clune", "Dusan Misevic", "Christoph Adami", "Lee Altenberg", "Julie Beaulieu", "Peter J. Bentley", "Samuel Bernard", "Guillaume Beslon", "David M. Bryson", "Patryk Chrabaszcz", "Nick Cheney", "Antoine Cully", "Stephane Doncieux", "Fred C. Dyer", "Kai Olav Ellefsen", "Robert Feldt", "Stephan Fischer", "Stephanie Forrest", "Antoine Frénoy", "Christian Gagné", "Leni Le Goff", "Laura M. Grabowski", "Babak Hodjat", "Frank Hutter", "Laurent Keller", "Carole Knibbe", "Peter Krcah", "Richard E. Lenski", "Hod Lipson", "Robert MacCurdy", "Carlos Maestre", "Risto Miikkulainen", "Sara Mitri", "David E. Moriarty", "Jean-Baptiste Mouret", "Anh Nguyen", "Charles Ofria", "Marc Parizeau", "David Parsons", "Robert T. Pennock", "William F. Punch", "Thomas S. Ray", "Marc Schoenauer", "Eric Shulte", "Karl Sims", "Kenneth O. Stanley", "François Taddei", "Danesh Tarapore", "Simon Thibault", "Westley Weimer", "Richard Watson", "Jason Yosinski"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.03453", "doi": null, "topics": ["Science & research", "Algorithms", "Biology & medicine"], "description": "Biological evolution provides a creative fount of complex and subtle adaptations, often surprising the scientists who discover them.", "description_method": "abstract-excerpt", "abstract": "Biological evolution provides a creative fount of complex and subtle adaptations, often surprising the scientists who discover them. However, because evolution is an algorithmic process that transcends the substrate in which it occurs, evolution's creativity is not limited to nature. Indeed, many researchers in the field of digital evolution have observed their evolving algorithms and organisms subverting their intentions, exposing unrecognized bugs in their code, producing unexpected adaptations, or exhibiting outcomes uncannily convergent with ones in nature. Such stories routinely reveal creativity by evolution in these digital worlds, but they rarely fit into the standard scientific narrative. Instead they are often treated as mere obstacles to be overcome, rather than results that warrant study in their own right. The stories themselves are traded among researchers through oral tradition, but that mode of information transmission is inefficient and prone to error and outright loss. Moreover, the fact that these stories tend to be shared only among practitioners means that many natural scientists do not realize how interesting and lifelike digital organisms are and how natural their evolution can be. To our knowledge, no collection of such anecdotes has been published before. This paper is the crowd-sourced product of researchers in the fields of artificial life and evolutionary computation who have provided first-hand accounts of such cases. It thus serves as a written, fact-checked collection of scientifically important and even entertaining stories. In doing so we also present here substantial evidence that the existence and importance of evolutionary surprises extends beyond the natural world, and may indeed be a universal property of all complex evolving systems.", "score": 1.0, "mention_count": 16, "user_count": 15, "months": ["2018-03", "2018-05", "2018-06", "2018-10", "2018-11", "2019-01", "2019-04", "2020-06", "2021-03", "2021-06", "2021-07", "2022-08", "2023-04", "2023-06"], "last_mentioned": "2023-06-01T23:51:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1901.01930", "slug": "4941a12b82edec9a", "title": "Keeping CALM: When Distributed Consistency is Easy", "authors": ["Joseph M. Hellerstein", "Peter Alvaro"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1901.01930", "doi": null, "topics": ["Distributed systems", "Mathematics", "Programming languages"], "description": "A key concern in modern distributed systems is to avoid the cost of coordination while maintaining consistent semantics.", "description_method": "abstract-excerpt", "abstract": "A key concern in modern distributed systems is to avoid the cost of coordination while maintaining consistent semantics. Until recently, there was no answer to the question of when coordination is actually required. In this paper we present an informal introduction to the CALM Theorem, which answers this question precisely by moving up from traditional storage consistency to consider properties of programs. CALM is an acronym for \"consistency as logical monotonicity\". The CALM Theorem shows that the programs that have consistent, coordination-free distributed implementations are exactly the programs that can be expressed in monotonic logic. This theoretical result has practical implications for developers of distributed applications. We show how CALM provides a constructive application-level counterpart to conventional \"systems\" wisdom, such as the apparently negative results of the CAP Theorem. We also discuss ways that monotonic thinking can influence distributed systems design, and how new programming language designs and tools can help developers write consistent, coordination-free code.", "score": 1.0, "mention_count": 15, "user_count": 8, "months": ["2019-05", "2019-11", "2021-02", "2022-11", "2023-04", "2023-06", "2023-11", "2024-01", "2024-03", "2024-09", "2024-10", "2025-02", "2025-09"], "last_mentioned": "2025-09-10T11:02:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.10226", "slug": "b6bd65de10278f04", "title": "A Watermark for Large Language Models", "authors": ["John Kirchenbauer", "Jonas Geiping", "Yuxin Wen", "Jonathan Katz", "Ian Miers", "Tom Goldstein"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.10226", "doi": null, "topics": ["Language models", "Algorithms", "Statistics & probability"], "description": "Potential harms of large language models can be mitigated by watermarking model output, i.e., embedding signals into generated text that are invisible to humans but algorithmically detectable from a short span of tokens.", "description_method": "abstract-excerpt", "abstract": "Potential harms of large language models can be mitigated by watermarking model output, i.e., embedding signals into generated text that are invisible to humans but algorithmically detectable from a short span of tokens. We propose a watermarking framework for proprietary language models. The watermark can be embedded with negligible impact on text quality, and can be detected using an efficient open-source algorithm without access to the language model API or parameters. The watermark works by selecting a randomized set of \"green\" tokens before a word is generated, and then softly promoting use of green tokens during sampling. We propose a statistical test for detecting the watermark with interpretable p-values, and derive an information-theoretic framework for analyzing the sensitivity of the watermark. We test the watermark using a multi-billion parameter model from the Open Pretrained Transformer (OPT) family, and discuss robustness and security.", "score": 1.0, "mention_count": 15, "user_count": 13, "months": ["2023-02", "2023-03", "2023-05", "2023-07", "2024-07", "2024-08", "2024-11", "2026-08"], "last_mentioned": "2026-08-17T13:11:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.13188", "slug": "b2b6ed344c81058d", "title": "Extracting Training Data from Diffusion Models", "authors": ["Nicholas Carlini", "Jamie Hayes", "Milad Nasr", "Matthew Jagielski", "Vikash Sehwag", "Florian Tramèr", "Borja Balle", "Daphne Ippolito", "Eric Wallace"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.13188", "doi": null, "topics": ["Physics", "Machine learning", "Security & cryptography"], "description": "Image diffusion models such as DALL-E 2, Imagen, and Stable Diffusion have attracted significant attention due to their ability to generate high-quality synthetic images.", "description_method": "abstract-excerpt", "abstract": "Image diffusion models such as DALL-E 2, Imagen, and Stable Diffusion have attracted significant attention due to their ability to generate high-quality synthetic images. In this work, we show that diffusion models memorize individual images from their training data and emit them at generation time. With a generate-and-filter pipeline, we extract over a thousand training examples from state-of-the-art models, ranging from photographs of individual people to trademarked company logos. We also train hundreds of diffusion models in various settings to analyze how different modeling and data decisions affect privacy. Overall, our results show that diffusion models are much less private than prior generative models such as GANs, and that mitigating these vulnerabilities may require new advances in privacy-preserving training.", "score": 1.0, "mention_count": 15, "user_count": 15, "months": ["2023-02", "2023-03", "2023-04", "2023-11", "2024-01", "2024-02", "2025-07"], "last_mentioned": "2025-07-19T10:24:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nrd.2017.243", "slug": "54d1f388507bddb0", "title": "mRNA vaccines — a new era in vaccinology", "authors": ["Norbert Pardi", "Michael J. Hogan", "Frederick W. Porter", "Drew Weissman"], "year": 2018, "venue": "Nature Reviews Drug Discovery", "url": "https://doi.org/10.1038/nrd.2017.243", "doi": "10.1038/nrd.2017.243", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 1.0, "mention_count": 14, "user_count": 14, "months": ["2020-06", "2020-11", "2020-12", "2021-02", "2021-03", "2021-05", "2021-07", "2021-08", "2021-09"], "last_mentioned": "2021-09-10T15:32:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2550-z", "slug": "9eb0d15a4c2b1844", "title": "SARS-CoV-2-specific T cell immunity in cases of COVID-19 and SARS, and uninfected controls", "authors": ["Nina Le Bert", "Anthony T. Tan", "Kamini Kunasegaran", "Christine Y. L. Tham", "Morteza Hafezi", "Adeline Chia", "Melissa Hui Yen Chng", "Meiyin Lin", "Nicole Tan", "Martin Linster", "Wan Ni Chia", "Mark I-Cheng Chen", "Lin-Fa Wang", "Eng Eong Ooi", "Shirin Kalimuddin", "Paul Anantharajah Tambyah", "Jenny Guek-Hong Low", "Yee-Joo Tan", "Antonio Bertoletti"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2550-z", "doi": "10.1038/s41586-020-2550-z", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 1.0, "mention_count": 14, "user_count": 12, "months": ["2020-07", "2020-08", "2020-09", "2021-01", "2021-05", "2021-09", "2021-10", "2021-12", "2022-01", "2022-02"], "last_mentioned": "2022-02-03T18:23:24Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1606.06565", "slug": "26fd7f0d0e607ddf", "title": "Concrete Problems in AI Safety", "authors": ["Dario Amodei", "Chris Olah", "Jacob Steinhardt", "Paul Christiano", "John Schulman", "Dan Mané"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1606.06565", "doi": null, "topics": ["Psychology", "Machine learning"], "description": "Rapid progress in machine learning and artificial intelligence (AI) has brought increasing attention to the potential impacts of AI technologies on society.", "description_method": "abstract-excerpt", "abstract": "Rapid progress in machine learning and artificial intelligence (AI) has brought increasing attention to the potential impacts of AI technologies on society. In this paper we discuss one such potential impact: the problem of accidents in machine learning systems, defined as unintended and harmful behavior that may emerge from poor design of real-world AI systems. We present a list of five practical research problems related to accident risk, categorized according to whether the problem originates from having the wrong objective function (\"avoiding side effects\" and \"avoiding reward hacking\"), an objective function that is too expensive to evaluate frequently (\"scalable supervision\"), or undesirable behavior during the learning process (\"safe exploration\" and \"distributional shift\"). We review previous work in these areas as well as suggesting research directions with a focus on relevance to cutting-edge AI systems. Finally, we consider the high-level question of how to think most productively about the safety of forward-looking applications of AI.", "score": 1.0, "mention_count": 13, "user_count": 13, "months": ["2016-06", "2016-07", "2017-10", "2018-05", "2018-10", "2019-11", "2019-12", "2021-05", "2021-10", "2023-04", "2026-06", "2026-07"], "last_mentioned": "2026-07-27T23:52:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1710.09829", "slug": "2625c6b6167dabf4", "title": "Dynamic Routing Between Capsules", "authors": ["Sara Sabour", "Nicholas Frosst", "Geoffrey E Hinton"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1710.09829", "doi": null, "topics": ["Networks", "Statistics & probability", "Neuroscience"], "description": "A capsule is a group of neurons whose activity vector represents the instantiation parameters of a specific type of entity such as an object or an object part.", "description_method": "abstract-excerpt", "abstract": "A capsule is a group of neurons whose activity vector represents the instantiation parameters of a specific type of entity such as an object or an object part. We use the length of the activity vector to represent the probability that the entity exists and its orientation to represent the instantiation parameters. Active capsules at one level make predictions, via transformation matrices, for the instantiation parameters of higher-level capsules. When multiple predictions agree, a higher level capsule becomes active. We show that a discrimininatively trained, multi-layer capsule system achieves state-of-the-art performance on MNIST and is considerably better than a convolutional net at recognizing highly overlapping digits. To achieve these results we use an iterative routing-by-agreement mechanism: A lower-level capsule prefers to send its output to higher level capsules whose activity vectors have a big scalar product with the prediction coming from the lower-level capsule.", "score": 1.0, "mention_count": 13, "user_count": 11, "months": ["2017-10", "2017-11", "2017-12", "2018-01", "2018-02", "2018-06", "2019-06", "2019-07", "2020-05", "2023-02"], "last_mentioned": "2023-02-08T16:19:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2207.02098", "slug": "6c4e7406c7294ace", "title": "Neural Networks and the Chomsky Hierarchy", "authors": ["Grégoire Delétang", "Anian Ruoss", "Jordi Grau-Moya", "Tim Genewein", "Li Kevin Wenliang", "Elliot Catt", "Chris Cundy", "Marcus Hutter", "Shane Legg", "Joel Veness", "Pedro A. Ortega"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2207.02098", "doi": null, "topics": ["Machine learning", "Language models"], "description": "Reliable generalization lies at the heart of safe ML and AI.", "description_method": "abstract-excerpt", "abstract": "Reliable generalization lies at the heart of safe ML and AI. However, understanding when and how neural networks generalize remains one of the most important unsolved problems in the field. In this work, we conduct an extensive empirical study (20'910 models, 15 tasks) to investigate whether insights from the theory of computation can predict the limits of neural network generalization in practice. We demonstrate that grouping tasks according to the Chomsky hierarchy allows us to forecast whether certain architectures will be able to generalize to out-of-distribution inputs. This includes negative results where even extensive amounts of data and training time never lead to any non-trivial generalization, despite models having sufficient capacity to fit the training data perfectly. Our results show that, for our subset of tasks, RNNs and Transformers fail to generalize on non-regular tasks, LSTMs can solve regular and counter-language tasks, and only networks augmented with structured memory (such as a stack or memory tape) can successfully generalize on context-free and context-sensitive tasks.", "score": 1.0, "mention_count": 13, "user_count": 6, "months": ["2022-08", "2023-02", "2023-04", "2023-05", "2023-06", "2024-04", "2024-09", "2024-11", "2026-06", "2026-07"], "last_mentioned": "2026-07-12T18:09:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2207.09238", "slug": "b5b50cb2b9ae215c", "title": "Formal Algorithms for Transformers", "authors": ["Mary Phuong", "Marcus Hutter"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2207.09238", "doi": null, "topics": ["Language models", "Algorithms", "Mathematics"], "description": "This document aims to be a self-contained, mathematically precise overview of transformer architectures and algorithms (*not* results).", "description_method": "abstract-excerpt", "abstract": "This document aims to be a self-contained, mathematically precise overview of transformer architectures and algorithms (*not* results). It covers what transformers are, how they are trained, what they are used for, their key architectural components, and a preview of the most prominent models. The reader is assumed to be familiar with basic ML terminology and simpler neural network architectures such as MLPs.", "score": 1.0, "mention_count": 13, "user_count": 10, "months": ["2023-01", "2023-04", "2023-05", "2023-06", "2023-07", "2023-12", "2024-01", "2026-06"], "last_mentioned": "2026-06-12T22:50:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2307.09009", "slug": "365b459f3a2ff196", "title": "How is ChatGPT's behavior changing over time?", "authors": ["Lingjiao Chen", "Matei Zaharia", "James Zou"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2307.09009", "doi": null, "topics": ["Language models", "Psychology", "Biology & medicine"], "description": "GPT-3.5 and GPT-4 are the two most widely used large language model (LLM) services.", "description_method": "abstract-excerpt", "abstract": "GPT-3.5 and GPT-4 are the two most widely used large language model (LLM) services. However, when and how these models are updated over time is opaque. Here, we evaluate the March 2023 and June 2023 versions of GPT-3.5 and GPT-4 on several diverse tasks: 1) math problems, 2) sensitive/dangerous questions, 3) opinion surveys, 4) multi-hop knowledge-intensive questions, 5) generating code, 6) US Medical License tests, and 7) visual reasoning. We find that the performance and behavior of both GPT-3.5 and GPT-4 can vary greatly over time. For example, GPT-4 (March 2023) was reasonable at identifying prime vs. composite numbers (84% accuracy) but GPT-4 (June 2023) was poor on these same questions (51% accuracy). This is partly explained by a drop in GPT-4's amenity to follow chain-of-thought prompting. Interestingly, GPT-3.5 was much better in June than in March in this task. GPT-4 became less willing to answer sensitive questions and opinion survey questions in June than in March. GPT-4 performed better at multi-hop questions in June than in March, while GPT-3.5's performance dropped on this task. Both GPT-4 and GPT-3.5 had more formatting mistakes in code generation in June than in March. We provide evidence that GPT-4's ability to follow user instructions has decreased over time, which is one common factor behind the many behavior drifts. Overall, our findings show that the behavior of the \"same\" LLM service can change substantially in a relatively short amount of time, highlighting the need for continuous monitoring of LLMs.", "score": 1.0, "mention_count": 13, "user_count": 13, "months": ["2023-07", "2023-08", "2023-09", "2024-02", "2024-03", "2024-08", "2026-04", "2026-06"], "last_mentioned": "2026-06-11T22:52:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1608.08225", "slug": "e1bc6388048ca371", "title": "Why does deep and cheap learning work so well?", "authors": ["Henry W. Lin", "Max Tegmark", "David Rolnick"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1608.08225", "doi": "10.1007/s10955-017-1836-5", "topics": ["Mathematics", "Machine learning", "Statistics & probability"], "description": "We show how the success of deep learning could depend not only on mathematics but also on physics: although well-known mathematical theorems guarantee that neural networks can approximate arbitrary functions well, the class of functions of practical interest can frequently be approximated through \"cheap learning\" with exponentially fewer ", "description_method": "abstract-excerpt", "abstract": "We show how the success of deep learning could depend not only on mathematics but also on physics: although well-known mathematical theorems guarantee that neural networks can approximate arbitrary functions well, the class of functions of practical interest can frequently be approximated through \"cheap learning\" with exponentially fewer parameters than generic ones. We explore how properties frequently encountered in physics such as symmetry, locality, compositionality, and polynomial log-probability translate into exceptionally simple neural networks. We further argue that when the statistical process generating the data is of a certain hierarchical form prevalent in physics and machine-learning, a deep neural network can be more efficient than a shallow one. We formalize these claims using information theory and discuss the relation to the renormalization group. We prove various \"no-flattening theorems\" showing when efficient linear deep networks cannot be accurately approximated by shallow ones without efficiency loss, for example, we show that $n$ variables cannot be multiplied using fewer than 2^n neurons in a single hidden layer.", "score": 1.0, "mention_count": 12, "user_count": 8, "months": ["2016-09", "2016-10", "2016-12", "2017-04", "2017-05", "2017-08", "2017-11", "2018-10", "2020-10", "2023-04"], "last_mentioned": "2023-04-26T12:34:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1804.03748", "slug": "f6fa4427fd293e3d", "title": "The Silurian Hypothesis: Would it be possible to detect an industrial civilization in the geological record?", "authors": ["Gavin A. Schmidt", "Adam Frank"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1804.03748", "doi": "10.1017/S1473550418000095", "topics": ["Statistics & probability"], "description": "If an industrial civilization had existed on Earth many millions of years prior to our own era, what traces would it have left and would they be detectable today?", "description_method": "abstract-excerpt", "abstract": "If an industrial civilization had existed on Earth many millions of years prior to our own era, what traces would it have left and would they be detectable today? We summarize the likely geological fingerprint of the Anthropocene, and demonstrate that while clear, it will not differ greatly in many respects from other known events in the geological record. We then propose tests that could plausibly distinguish an industrial cause from an otherwise naturally occurring climate event.", "score": 1.0, "mention_count": 12, "user_count": 12, "months": ["2019-06", "2019-07", "2019-10", "2019-12", "2020-06", "2020-08", "2021-01", "2023-08", "2024-05", "2024-11"], "last_mentioned": "2024-11-11T22:25:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0704.0646", "slug": "3ef702dc948d76d9", "title": "The Mathematical Universe", "authors": ["Max Tegmark"], "year": 2007, "venue": "arXiv", "url": "https://arxiv.org/abs/0704.0646", "doi": "10.1007/s10701-007-9186-9", "topics": ["Mathematics", "Physics", "Statistics & probability"], "description": "I explore physics implications of the External Reality Hypothesis (ERH) that there exists an external physical reality completely independent of us humans.", "description_method": "abstract-excerpt", "abstract": "I explore physics implications of the External Reality Hypothesis (ERH) that there exists an external physical reality completely independent of us humans. I argue that with a sufficiently broad definition of mathematics, it implies the Mathematical Universe Hypothesis (MUH) that our physical world is an abstract mathematical structure. I discuss various implications of the ERH and MUH, ranging from standard physics topics like symmetries, irreducible representations, units, free parameters, randomness and initial conditions to broader issues like consciousness, parallel universes and Godel incompleteness. I hypothesize that only computable and decidable (in Godel's sense) structures exist, which alleviates the cosmological measure problem and help explain why our physical laws appear so simple. I also comment on the intimate relation between mathematical structures, computations, simulations and physical systems.", "score": 1.0, "mention_count": 11, "user_count": 10, "months": ["2010-10", "2012-12", "2014-05", "2014-10", "2016-01", "2016-11", "2017-07", "2019-03", "2020-04", "2022-08", "2026-05"], "last_mentioned": "2026-05-18T18:49:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.4057", "slug": "3d6c2250ed469886", "title": "Aristotle's Physics: a Physicist's Look", "authors": ["Carlo Rovelli"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.4057", "doi": "10.1017/apa.2014.11", "topics": ["Physics"], "description": "I show that Aristotelian physics is a correct and non-intuitive approximation of Newtonian physics in the suitable domain (motion in fluids), in the same technical sense in which Newton theory is an approximation of Einstein's theory.", "description_method": "abstract-excerpt", "abstract": "I show that Aristotelian physics is a correct and non-intuitive approximation of Newtonian physics in the suitable domain (motion in fluids), in the same technical sense in which Newton theory is an approximation of Einstein's theory. Aristotelian physics lasted long not because it became dogma, but because it is a very good empirically grounded theory. The observation suggests some general considerations on inter-theoretical relations.", "score": 1.0, "mention_count": 11, "user_count": 7, "months": ["2016-07", "2016-10", "2017-01", "2017-04", "2017-11", "2022-01", "2024-05", "2024-12", "2025-06"], "last_mentioned": "2025-06-10T14:29:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.10601", "slug": "b9e2587557aa1147", "title": "Tree of Thoughts: Deliberate Problem Solving with Large Language Models", "authors": ["Shunyu Yao", "Dian Yu", "Jeffrey Zhao", "Izhak Shafran", "Thomas L. Griffiths", "Yuan Cao", "Karthik Narasimhan"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.10601", "doi": null, "topics": ["Language models", "Psychology"], "description": "Language models are increasingly being deployed for general problem solving across a wide range of tasks, but are still confined to token-level, left-to-right decision-making processes during inference.", "description_method": "abstract-excerpt", "abstract": "Language models are increasingly being deployed for general problem solving across a wide range of tasks, but are still confined to token-level, left-to-right decision-making processes during inference. This means they can fall short in tasks that require exploration, strategic lookahead, or where initial decisions play a pivotal role. To surmount these challenges, we introduce a new framework for language model inference, Tree of Thoughts (ToT), which generalizes over the popular Chain of Thought approach to prompting language models, and enables exploration over coherent units of text (thoughts) that serve as intermediate steps toward problem solving. ToT allows LMs to perform deliberate decision making by considering multiple different reasoning paths and self-evaluating choices to decide the next course of action, as well as looking ahead or backtracking when necessary to make global choices. Our experiments show that ToT significantly enhances language models' problem-solving abilities on three novel tasks requiring non-trivial planning or search: Game of 24, Creative Writing, and Mini Crosswords. For instance, in Game of 24, while GPT-4 with chain-of-thought prompting only solved 4% of tasks, our method achieved a success rate of 74%. Code repo with all prompts: https://github.com/princeton-nlp/tree-of-thought-llm.", "score": 1.0, "mention_count": 11, "user_count": 11, "months": ["2023-05", "2023-06", "2023-07", "2023-11", "2024-09", "2024-10", "2026-03"], "last_mentioned": "2026-03-16T21:47:19Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2406.19108", "slug": "c146af7983bebb3f", "title": "Computational Life: How Well-formed, Self-replicating Programs Emerge from Simple Interaction", "authors": ["Blaise Agüera y Arcas", "Jyrki Alakuijala", "James Evans", "Ben Laurie", "Alexander Mordvintsev", "Eyvind Niklasson", "Ettore Randazzo", "Luca Versari"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2406.19108", "doi": null, "topics": ["Programming languages", "Mathematics", "Distributed systems"], "description": "The fields of Origin of Life and Artificial Life both question what life is and how it emerges from a distinct set of \"pre-life\" dynamics.", "description_method": "abstract-excerpt", "abstract": "The fields of Origin of Life and Artificial Life both question what life is and how it emerges from a distinct set of \"pre-life\" dynamics. One common feature of most substrates where life emerges is a marked shift in dynamics when self-replication appears. While there are some hypotheses regarding how self-replicators arose in nature, we know very little about the general dynamics, computational principles, and necessary conditions for self-replicators to emerge. This is especially true on \"computational substrates\" where interactions involve logical, mathematical, or programming rules. In this paper we take a step towards understanding how self-replicators arise by studying several computational substrates based on various simple programming languages and machine instruction sets. We show that when random, non self-replicating programs are placed in an environment lacking any explicit fitness landscape, self-replicators tend to arise. We demonstrate how this occurs due to random interactions and self-modification, and can happen with and without background random mutations. We also show how increasingly complex dynamics continue to emerge following the rise of self-replicators. Finally, we show a counterexample of a minimalistic programming language where self-replicators are possible, but so far have not been observed to arise.", "score": 1.0, "mention_count": 11, "user_count": 7, "months": ["2024-07", "2024-08", "2024-09", "2024-11", "2025-11", "2026-07", "2026-08"], "last_mentioned": "2026-08-19T18:48:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1404.5997", "slug": "3c80e8edce133238", "title": "One weird trick for parallelizing convolutional neural networks", "authors": ["Alex Krizhevsky"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1404.5997", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "I present a new way to parallelize the training of convolutional neural networks across multiple GPUs.", "description_method": "abstract-excerpt", "abstract": "I present a new way to parallelize the training of convolutional neural networks across multiple GPUs. The method scales significantly better than all alternatives when applied to modern convolutional neural networks.", "score": 1.0, "mention_count": 10, "user_count": 9, "months": ["2014-04", "2014-08", "2014-11", "2015-11", "2015-12", "2016-05", "2016-07", "2018-09", "2020-02", "2022-06"], "last_mentioned": "2022-06-09T21:24:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.02989", "slug": "6862ba58b0df0f7c", "title": "xVal: A Continuous Numerical Tokenization for Scientific Language Models", "authors": ["Siavash Golkar", "Mariel Pettee", "Michael Eickenberg", "Alberto Bietti", "Miles Cranmer", "Geraud Krawezik", "Francois Lanusse", "Michael McCabe", "Ruben Ohana", "Liam Parker", "Bruno Régaldo-Saint Blancard", "Tiberiu Tesileanu", "Kyunghyun Cho", "Shirley Ho"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.02989", "doi": null, "topics": ["Science & research", "Language models", "Psychology"], "description": "Due in part to their discontinuous and discrete default encodings for numbers, Large Language Models (LLMs) have not yet been commonly used to process numerically-dense scientific datasets.", "description_method": "abstract-excerpt", "abstract": "Due in part to their discontinuous and discrete default encodings for numbers, Large Language Models (LLMs) have not yet been commonly used to process numerically-dense scientific datasets. Rendering datasets as text, however, could help aggregate diverse and multi-modal scientific data into a single training corpus, thereby potentially facilitating the development of foundation models for science. In this work, we introduce xVal, a strategy for continuously tokenizing numbers within language models that results in a more appropriate inductive bias for scientific applications. By training specially-modified language models from scratch on a variety of scientific datasets formatted as text, we find that xVal generally outperforms other common numerical tokenization strategies on metrics including out-of-distribution generalization and computational efficiency.", "score": 1.0, "mention_count": 10, "user_count": 5, "months": ["2023-10", "2023-11", "2024-03", "2024-09", "2024-10"], "last_mentioned": "2024-10-23T16:45:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2505.15917", "slug": "aeac9bd8a0d91363", "title": "How to factor 2048 bit RSA integers with less than a million noisy qubits", "authors": ["Craig Gidney"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2505.15917", "doi": null, "topics": ["Physics", "Security & cryptography"], "description": "Planning the transition to quantum-safe cryptosystems requires understanding the cost of quantum attacks on vulnerable cryptosystems.", "description_method": "abstract-excerpt", "abstract": "Planning the transition to quantum-safe cryptosystems requires understanding the cost of quantum attacks on vulnerable cryptosystems. In Gidney+Ekerå 2019, I co-published an estimate stating that 2048 bit RSA integers could be factored in eight hours by a quantum computer with 20 million noisy qubits. In this paper, I substantially reduce the number of qubits required. I estimate that a 2048 bit RSA integer could be factored in less than a week by a quantum computer with less than a million noisy qubits. I make the same assumptions as in 2019: a square grid of qubits with nearest neighbor connections, a uniform gate error rate of $0.1\\%$, a surface code cycle time of 1 microsecond, and a control system reaction time of $10$ microseconds. The qubit count reduction comes mainly from using approximate residue arithmetic (Chevignard+Fouque+Schrottenloher 2024), from storing idle logical qubits with yoked surface codes (Gidney+Newman+Brooks+Jones 2023), and from allocating less space to magic state distillation by using magic state cultivation (Gidney+Shutty+Jones 2024). The longer runtime is mainly due to performing more Toffoli gates and using fewer magic state factories compared to Gidney+Ekerå 2019. That said, I reduce the Toffoli count by over 100x compared to Chevignard+Fouque+Schrottenloher 2024.", "score": 1.0, "mention_count": 10, "user_count": 6, "months": ["2025-05", "2025-08", "2025-09", "2025-10", "2025-11", "2026-07"], "last_mentioned": "2026-07-21T19:51:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:math/0411418", "slug": "b4b6bc952b22792a", "title": "How real are real numbers?", "authors": ["G. J. Chaitin"], "year": 2004, "venue": "arXiv", "url": "https://arxiv.org/abs/math/0411418", "doi": null, "topics": ["Mathematics"], "description": "We discuss mathematical and physical arguments against continuity and in favor of discreteness, with particular emphasis on the ideas of Emile Borel (1871-1956).", "description_method": "abstract-excerpt", "abstract": "We discuss mathematical and physical arguments against continuity and in favor of discreteness, with particular emphasis on the ideas of Emile Borel (1871-1956).", "score": 1.0, "mention_count": 10, "user_count": 10, "months": ["2016-11", "2017-04", "2018-06", "2018-08", "2018-10", "2019-05", "2020-01", "2020-08", "2022-12", "2025-08"], "last_mentioned": "2025-08-30T11:49:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/ncomms15611", "slug": "636a49cc5a8d9320", "title": "River plastic emissions to the world’s oceans", "authors": ["Laurent C. M. Lebreton", "Joost van der Zwet", "Jan-Willem Damsteeg", "Boyan Slat", "Anthony Andrady", "Julia Reisser"], "year": 2017, "venue": "Nature Communications", "url": "https://doi.org/10.1038/ncomms15611", "doi": "10.1038/ncomms15611", "topics": ["Biology & medicine"], "description": "Abstract Plastics in the marine environment have become a major concern because of their persistence at sea, and adverse consequences to marine life and potentially human health.", "description_method": "abstract-excerpt", "abstract": "Abstract Plastics in the marine environment have become a major concern because of their persistence at sea, and adverse consequences to marine life and potentially human health. Implementing mitigation strategies requires an understanding and quantification of marine plastic sources, taking spatial and temporal variability into account. Here we present a global model of plastic inputs from rivers into oceans based on waste management, population density and hydrological information. Our model is calibrated against measurements available in the literature. We estimate that between 1.15 and 2.41 million tonnes of plastic waste currently enters the ocean every year from rivers, with over 74% of emissions occurring between May and October. The top 20 polluting rivers, mostly located in Asia, account for 67% of the global total. The findings of this study provide baseline data for ocean plastic mass balance exercises, and assist in prioritizing future plastic debris monitoring and mitigation strategies.", "score": 1.0, "mention_count": 10, "user_count": 10, "months": ["2018-04", "2018-05", "2018-09", "2019-03", "2019-07", "2019-09", "2019-10", "2020-01", "2021-03", "2024-04"], "last_mentioned": "2024-04-18T17:03:32Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1410.3831", "slug": "b47665394c4854e0", "title": "An exact mapping between the Variational Renormalization Group and Deep Learning", "authors": ["Pankaj Mehta", "David J. Schwab"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1410.3831", "doi": null, "topics": ["Machine learning", "Psychology", "Physics"], "description": "Deep learning is a broad set of techniques that uses multiple layers of representation to automatically learn relevant features directly from structured data.", "description_method": "abstract-excerpt", "abstract": "Deep learning is a broad set of techniques that uses multiple layers of representation to automatically learn relevant features directly from structured data. Recently, such techniques have yielded record-breaking results on a diverse set of difficult machine learning tasks in computer vision, speech recognition, and natural language processing. Despite the enormous success of deep learning, relatively little is understood theoretically about why these techniques are so successful at feature learning and compression. Here, we show that deep learning is intimately related to one of the most important and successful techniques in theoretical physics, the renormalization group (RG). RG is an iterative coarse-graining scheme that allows for the extraction of relevant features (i.e. operators) as a physical system is examined at different length scales. We construct an exact mapping from the variational renormalization group, first introduced by Kadanoff, and deep learning architectures based on Restricted Boltzmann Machines (RBMs). We illustrate these ideas using the nearest-neighbor Ising Model in one and two-dimensions. Our results suggests that deep learning algorithms may be employing a generalized RG-like scheme to learn relevant features from data.", "score": 1.0, "mention_count": 9, "user_count": 8, "months": ["2014-12", "2015-07", "2016-09", "2016-11", "2016-12", "2017-08", "2019-07", "2024-06", "2025-05"], "last_mentioned": "2025-05-20T22:47:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1503.03585", "slug": "dc1911894fe1aeeb", "title": "Deep Unsupervised Learning using Nonequilibrium Thermodynamics", "authors": ["Jascha Sohl-Dickstein", "Eric A. Weiss", "Niru Maheswaranathan", "Surya Ganguli"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1503.03585", "doi": null, "topics": ["Physics", "Machine learning", "Statistics & probability"], "description": "A central problem in machine learning involves modeling complex data-sets using highly flexible families of probability distributions in which learning, sampling, inference, and evaluation are still analytically or computationally tractable.", "description_method": "abstract-excerpt", "abstract": "A central problem in machine learning involves modeling complex data-sets using highly flexible families of probability distributions in which learning, sampling, inference, and evaluation are still analytically or computationally tractable. Here, we develop an approach that simultaneously achieves both flexibility and tractability. The essential idea, inspired by non-equilibrium statistical physics, is to systematically and slowly destroy structure in a data distribution through an iterative forward diffusion process. We then learn a reverse diffusion process that restores structure in data, yielding a highly flexible and tractable generative model of the data. This approach allows us to rapidly learn, sample from, and evaluate probabilities in deep generative models with thousands of layers or time steps, as well as to compute conditional and posterior probabilities under the learned model. We additionally release an open source reference implementation of the algorithm.", "score": 1.0, "mention_count": 9, "user_count": 7, "months": ["2022-04", "2022-10", "2023-01", "2023-07", "2024-02", "2024-05", "2024-08", "2024-09"], "last_mentioned": "2024-09-29T04:30:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1509.05053", "slug": "d318ecad0ffeb2b1", "title": "Array Layouts for Comparison-Based Searching", "authors": ["Paul-Virak Khuong", "Pat Morin"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1509.05053", "doi": null, "topics": ["Databases", "Algorithms"], "description": "We attempt to determine the best order and search algorithm to store $n$ comparable data items in an array, $A$, of length $n$ so that we can, for any query value, $x$, quickly find the smallest value in $A$ that is greater than or equal to $x$.", "description_method": "abstract-excerpt", "abstract": "We attempt to determine the best order and search algorithm to store $n$ comparable data items in an array, $A$, of length $n$ so that we can, for any query value, $x$, quickly find the smallest value in $A$ that is greater than or equal to $x$. In particular, we consider the important case where there are many such queries to the same array, $A$, which resides entirely in RAM. In addition to the obvious sorted order/binary search combination we consider the Eytzinger (BFS) layout normally used for heaps, an implicit B-tree layout that generalizes the Eytzinger layout, and the van Emde Boas layout commonly used in the cache-oblivious algorithms literature. After extensive testing and tuning on a wide variety of modern hardware, we arrive at the conclusion that, for small values of $n$, sorted order, combined with a good implementation of binary search is best. For larger values of $n$, we arrive at the surprising conclusion that the Eytzinger layout is usually the fastest. The latter conclusion is unexpected and goes counter to earlier experimental work by Brodal, Fagerberg, and Jacob (SODA~2003), who concluded that both the B-tree and van Emde Boas layouts were faster than the Eytzinger layout for large values of $n$. Our fastest C++ implementations, when compiled, use conditional moves to avoid branch mispredictions and prefetching to reduce cache latency.", "score": 1.0, "mention_count": 9, "user_count": 8, "months": ["2015-10", "2017-06", "2020-05", "2020-07", "2022-02", "2024-11", "2026-07"], "last_mentioned": "2026-07-18T11:47:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1604.00289", "slug": "8e336ecbce99aa1a", "title": "Building Machines That Learn and Think Like People", "authors": ["Brenden M. Lake", "Tomer D. Ullman", "Joshua B. Tenenbaum", "Samuel J. Gershman"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1604.00289", "doi": null, "topics": ["Psychology", "Neuroscience", "Machine learning"], "description": "Recent progress in artificial intelligence (AI) has renewed interest in building systems that learn and think like people.", "description_method": "abstract-excerpt", "abstract": "Recent progress in artificial intelligence (AI) has renewed interest in building systems that learn and think like people. Many advances have come from using deep neural networks trained end-to-end in tasks such as object recognition, video games, and board games, achieving performance that equals or even beats humans in some respects. Despite their biological inspiration and performance achievements, these systems differ from human intelligence in crucial ways. We review progress in cognitive science suggesting that truly human-like learning and thinking machines will have to reach beyond current engineering trends in both what they learn, and how they learn it. Specifically, we argue that these machines should (a) build causal models of the world that support explanation and understanding, rather than merely solving pattern recognition problems; (b) ground learning in intuitive theories of physics and psychology, to support and enrich the knowledge that is learned; and (c) harness compositionality and learning-to-learn to rapidly acquire and generalize knowledge to new tasks and situations. We suggest concrete challenges and promising routes towards these goals that can combine the strengths of recent neural network advances with more structured cognitive models.", "score": 1.0, "mention_count": 9, "user_count": 7, "months": ["2016-08", "2017-02", "2017-04", "2017-07", "2019-05", "2019-06"], "last_mentioned": "2019-06-02T15:21:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1804.06826", "slug": "d8c40227711b1b2b", "title": "Dissecting the NVIDIA Volta GPU Architecture via Microbenchmarking", "authors": ["Zhe Jia", "Marco Maggioni", "Benjamin Staiger", "Daniele P. Scarpazza"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1804.06826", "doi": null, "topics": ["Science & research", "Psychology", "Mathematics"], "description": "Every year, novel NVIDIA GPU designs are introduced.", "description_method": "abstract-excerpt", "abstract": "Every year, novel NVIDIA GPU designs are introduced. This rapid architectural and technological progression, coupled with a reluctance by manufacturers to disclose low-level details, makes it difficult for even the most proficient GPU software designers to remain up-to-date with the technological advances at a microarchitectural level. To address this dearth of public, microarchitectural-level information on the novel NVIDIA GPUs, independent researchers have resorted to microbenchmarks-based dissection and discovery. This has led to a prolific line of publications that shed light on instruction encoding, and memory hierarchy's geometry and features at each level. Namely, research that describes the performance and behavior of the Kepler, Maxwell and Pascal architectures. In this technical report, we continue this line of research by presenting the microarchitectural details of the NVIDIA Volta architecture, discovered through microbenchmarks and instruction set disassembly. Additionally, we compare quantitatively our Volta findings against its predecessors, Kepler, Maxwell and Pascal.", "score": 1.0, "mention_count": 9, "user_count": 4, "months": ["2019-03", "2019-07", "2020-09", "2021-01", "2021-10", "2022-08", "2024-10", "2025-06", "2025-08"], "last_mentioned": "2025-08-20T17:22:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1901.03429", "slug": "8471dad41d2245e0", "title": "On the Turing Completeness of Modern Neural Network Architectures", "authors": ["Jorge Pérez", "Javier Marinković", "Pablo Barceló"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1901.03429", "doi": null, "topics": ["Machine learning", "Language models"], "description": "Alternatives to recurrent neural networks, in particular, architectures based on attention or convolutions, have been gaining momentum for processing input sequences.", "description_method": "abstract-excerpt", "abstract": "Alternatives to recurrent neural networks, in particular, architectures based on attention or convolutions, have been gaining momentum for processing input sequences. In spite of their relevance, the computational properties of these alternatives have not yet been fully explored. We study the computational power of two of the most paradigmatic architectures exemplifying these mechanisms: the Transformer (Vaswani et al., 2017) and the Neural GPU (Kaiser & Sutskever, 2016). We show both models to be Turing complete exclusively based on their capacity to compute and access internal dense representations of the data. In particular, neither the Transformer nor the Neural GPU requires access to an external memory to become Turing complete. Our study also reveals some minimal sets of elements needed to obtain these completeness results.", "score": 1.0, "mention_count": 9, "user_count": 8, "months": ["2020-07", "2020-09", "2022-06", "2022-09", "2023-01", "2023-03", "2023-04", "2025-02"], "last_mentioned": "2025-02-04T16:36:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2205.11659", "slug": "490fb4c9b0bb4d4b", "title": "Fast GPU bounding boxes on tree-structured scenes", "authors": ["Raph Levien"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2205.11659", "doi": null, "topics": ["Algorithms"], "description": "Computation of bounding boxes is a fundamental problem in high performance rendering, as it is an input to visibility culling and binning operations.", "description_method": "abstract-excerpt", "abstract": "Computation of bounding boxes is a fundamental problem in high performance rendering, as it is an input to visibility culling and binning operations. In a scene description structured as a tree, clip nodes and blend nodes entail intersection and union of bounding boxes, respectively. These are straightforward to compute on the CPU using a sequential algorithm, but an efficient, parallel GPU algorithm is more elusive. This paper presents a fast and practical solution, with a new algorithm for the classic parentheses matching problem at its core. The core algorithm is presented abstractly (in terms of a PRAM abstraction), then with a concrete mapping to the thread, workgroup, and dispatch levels of real GPU hardware. The algorithm is implemented portably using compute shaders, and performance results show a dramatic speedup over a sequential CPU version, and indeed a reasonable fraction of maximum theoretical throughput of the GPU hardware. The immediate motivating application is 2D rendering, but the algorithms generalize to other domains, and the core parentheses matching problem has other applications including parsing.", "score": 1.0, "mention_count": 9, "user_count": 3, "months": ["2022-06", "2022-08", "2022-12", "2023-05", "2023-09", "2024-06", "2026-06"], "last_mentioned": "2026-06-25T21:35:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2304.03271", "slug": "2aff822f9f1c6e70", "title": "Making AI Less \"Thirsty\": Uncovering and Addressing the Secret Water Footprint of AI Models", "authors": ["Pengfei Li", "Jianyi Yang", "Mohammad A. Islam", "Shaolei Ren"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2304.03271", "doi": null, "topics": ["Language models", "Psychology"], "description": "The growing carbon footprint of artificial intelligence (AI) has been undergoing public scrutiny.", "description_method": "abstract-excerpt", "abstract": "The growing carbon footprint of artificial intelligence (AI) has been undergoing public scrutiny. Nonetheless, the equally important water (withdrawal and consumption) footprint of AI has largely remained under the radar. For example, training the GPT-3 language model in Microsoft's state-of-the-art U.S. data centers can directly evaporate 700,000 liters of clean freshwater, but such information has been kept a secret. More critically, the global AI demand is projected to account for 4.2-6.6 billion cubic meters of water withdrawal in 2027, which is more than the total annual water withdrawal of 4-6 Denmark or half of the United Kingdom. This is concerning, as freshwater scarcity has become one of the most pressing challenges. To respond to the global water challenges, AI can, and also must, take social responsibility and lead by example by addressing its own water footprint. In this paper, we provide a principled methodology to estimate the water footprint of AI, and also discuss the unique spatial-temporal diversities of AI's runtime water efficiency. Finally, we highlight the necessity of holistically addressing water footprint along with carbon footprint to enable truly sustainable AI.", "score": 1.0, "mention_count": 9, "user_count": 7, "months": ["2024-09", "2025-01", "2025-04", "2025-11", "2026-05", "2026-06"], "last_mentioned": "2026-06-11T15:27:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:gr-qc/0009013", "slug": "2e5b4eab605f38de", "title": "The warp drive: hyper-fast travel within general relativity", "authors": ["Miguel Alcubierre"], "year": 2000, "venue": "arXiv", "url": "https://arxiv.org/abs/gr-qc/0009013", "doi": "10.1088/0264-9381/11/5/001", "topics": ["Physics"], "description": "It is shown how, within the framework of general relativity and without the introduction of wormholes, it is possible to modify a spacetime in a way that allows a spaceship to travel with an arbitrarily large speed.", "description_method": "abstract-excerpt", "abstract": "It is shown how, within the framework of general relativity and without the introduction of wormholes, it is possible to modify a spacetime in a way that allows a spaceship to travel with an arbitrarily large speed. By a purely local expansion of spacetime behind the spaceship and an opposite contraction in front of it, motion faster than the speed of light as seen by observers outside the disturbed region is possible. The resulting distortion is reminiscent of the ``warp drive'' of science fiction. However, just as it happens with wormholes, exotic matter will be needed in order to generate a distortion of spacetime like the one discussed here.", "score": 1.0, "mention_count": 9, "user_count": 4, "months": ["2013-07", "2018-08", "2020-11", "2021-03", "2021-12"], "last_mentioned": "2021-12-15T18:56:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3386327", "slug": "3bec6a806366f6de", "title": "JavaScript: the first 20 years", "authors": ["Allen Wirfs-Brock", "Brendan Eich"], "year": 2020, "venue": "Proceedings of the ACM on Programming Languages", "url": "https://doi.org/10.1145/3386327", "doi": "10.1145/3386327", "topics": ["Programming languages", "Economics"], "description": "How a sidekick scripting language for Java, created at Netscape in a ten-day hack, ships first as a de facto Web standard and eventually becomes the world's most widely used programming language.", "description_method": "abstract-excerpt", "abstract": "How a sidekick scripting language for Java, created at Netscape in a ten-day hack, ships first as a de facto Web standard and eventually becomes the world's most widely used programming language. This paper tells the story of the creation, design, evolution, and standardization of the JavaScript language over the period of 1995--2015. But the story is not only about the technical details of the language. It is also the story of how people and organizations competed and collaborated to shape the JavaScript language which dominates the Web of 2020.", "score": 1.0, "mention_count": 9, "user_count": 7, "months": ["2021-05", "2021-08", "2022-08", "2022-09", "2023-11", "2024-06", "2024-10"], "last_mentioned": "2024-10-08T15:37:29Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1208.0928", "slug": "e0df6a6e378b20cd", "title": "Surface codes: Towards practical large-scale quantum computation", "authors": ["Austin G. Fowler", "Matteo Mariantoni", "John M. Martinis", "Andrew N. Cleland"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1208.0928", "doi": "10.1103/PhysRevA.86.032324", "topics": ["Physics", "Distributed systems"], "description": "This article provides an introduction to surface code quantum computing.", "description_method": "abstract-excerpt", "abstract": "This article provides an introduction to surface code quantum computing. We first estimate the size and speed of a surface code quantum computer. We then introduce the concept of the stabilizer, using two qubits, and extend this concept to stabilizers acting on a two-dimensional array of physical qubits, on which we implement the surface code. We next describe how logical qubits are formed in the surface code array and give numerical estimates of their fault-tolerance. We outline how logical qubits are physically moved on the array, how qubit braid transformations are constructed, and how a braid between two logical qubits is equivalent to a controlled-NOT. We then describe the single-qubit Hadamard, S and T operators, completing the set of required gates for a universal quantum computer. We conclude by briefly discussing physical implementations of the surface code. We include a number of appendices in which we provide supplementary information to the main text.", "score": 1.0, "mention_count": 8, "user_count": 3, "months": ["2015-04", "2022-04", "2024-06", "2024-08", "2025-08", "2025-10", "2025-11", "2026-04"], "last_mentioned": "2026-04-01T16:31:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1410.5787", "slug": "bbe53420785aa9d4", "title": "The Precautionary Principle (with Application to the Genetic Modification of Organisms)", "authors": ["Nassim Nicholas Taleb", "Rupert Read", "Raphael Douady", "Joseph Norman", "Yaneer Bar-Yam"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1410.5787", "doi": null, "topics": ["Statistics & probability", "Science & research"], "description": "We present a non-naive version of the Precautionary (PP) that allows us to avoid paranoia and paralysis by confining precaution to specific domains and problems.", "description_method": "abstract-excerpt", "abstract": "We present a non-naive version of the Precautionary (PP) that allows us to avoid paranoia and paralysis by confining precaution to specific domains and problems. PP is intended to deal with uncertainty and risk in cases where the absence of evidence and the incompleteness of scientific knowledge carries profound implications and in the presence of risks of \"black swans\", unforeseen and unforeseable events of extreme consequence. We formalize PP, placing it within the statistical and probabilistic structure of ruin problems, in which a system is at risk of total failure, and in place of risk we use a formal fragility based approach. We make a central distinction between 1) thin and fat tails, 2) Local and systemic risks and place PP in the joint Fat Tails and systemic cases. We discuss the implications for GMOs (compared to Nuclear energy) and show that GMOs represent a public risk of global harm (while harm from nuclear energy is comparatively limited and better characterized). PP should be used to prescribe severe limits on GMOs.", "score": 1.0, "mention_count": 8, "user_count": 8, "months": ["2014-11", "2015-07", "2015-12", "2016-11", "2018-03", "2018-04", "2018-12"], "last_mentioned": "2018-12-22T21:38:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2009.04374", "slug": "aaa44702813431bd", "title": "Assessing Game Balance with AlphaZero: Exploring Alternative Rule Sets in Chess", "authors": ["Nenad Tomašev", "Ulrich Paquet", "Demis Hassabis", "Vladimir Kramnik"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2009.04374", "doi": null, "topics": ["Other research"], "description": "It is non-trivial to design engaging and balanced sets of game rules.", "description_method": "abstract-excerpt", "abstract": "It is non-trivial to design engaging and balanced sets of game rules. Modern chess has evolved over centuries, but without a similar recourse to history, the consequences of rule changes to game dynamics are difficult to predict. AlphaZero provides an alternative in silico means of game balance assessment. It is a system that can learn near-optimal strategies for any rule set from scratch, without any human supervision, by continually learning from its own experience. In this study we use AlphaZero to creatively explore and design new chess variants. There is growing interest in chess variants like Fischer Random Chess, because of classical chess's voluminous opening theory, the high percentage of draws in professional play, and the non-negligible number of games that end while both players are still in their home preparation. We compare nine other variants that involve atomic changes to the rules of chess. The changes allow for novel strategic and tactical patterns to emerge, while keeping the games close to the original. By learning near-optimal strategies for each variant with AlphaZero, we determine what games between strong human players might look like if these variants were adopted. Qualitatively, several variants are very dynamic. An analytic comparison show that pieces are valued differently between variants, and that some variants are more decisive than classical chess. Our findings demonstrate the rich possibilities that lie beyond the rules of modern chess.", "score": 1.0, "mention_count": 8, "user_count": 4, "months": ["2020-12", "2021-08", "2022-11", "2024-01"], "last_mentioned": "2024-01-29T00:59:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2309.12288", "slug": "02782a7d9697d5df", "title": "The Reversal Curse: LLMs trained on \"A is B\" fail to learn \"B is A\"", "authors": ["Lukas Berglund", "Meg Tong", "Max Kaufmann", "Mikita Balesni", "Asa Cooper Stickland", "Tomasz Korbak", "Owain Evans"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2309.12288", "doi": null, "topics": ["Language models", "Machine learning"], "description": "We expose a surprising failure of generalization in auto-regressive large language models (LLMs).", "description_method": "abstract-excerpt", "abstract": "We expose a surprising failure of generalization in auto-regressive large language models (LLMs). If a model is trained on a sentence of the form \"A is B\", it will not automatically generalize to the reverse direction \"B is A\". This is the Reversal Curse. For instance, if a model is trained on \"Valentina Tereshkova was the first woman to travel to space\", it will not automatically be able to answer the question, \"Who was the first woman to travel to space?\". Moreover, the likelihood of the correct answer (\"Valentina Tershkova\") will not be higher than for a random name. Thus, models do not generalize a prevalent pattern in their training set: if \"A is B\" occurs, \"B is A\" is more likely to occur. It is worth noting, however, that if \"A is B\" appears in-context, models can deduce the reverse relationship. We provide evidence for the Reversal Curse by finetuning GPT-3 and Llama-1 on fictitious statements such as \"Uriah Hawthorne is the composer of Abyssal Melodies\" and showing that they fail to correctly answer \"Who composed Abyssal Melodies?\". The Reversal Curse is robust across model sizes and model families and is not alleviated by data augmentation. We also evaluate ChatGPT (GPT-3.5 and GPT-4) on questions about real-world celebrities, such as \"Who is Tom Cruise's mother? [A: Mary Lee Pfeiffer]\" and the reverse \"Who is Mary Lee Pfeiffer's son?\". GPT-4 correctly answers questions like the former 79% of the time, compared to 33% for the latter. Code available at: https://github.com/lukasberglund/reversal_curse.", "score": 1.0, "mention_count": 8, "user_count": 8, "months": ["2023-10", "2023-11", "2024-01", "2024-06", "2024-07", "2026-08"], "last_mentioned": "2026-08-12T11:53:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41562-020-01009-0", "slug": "710d88abaa50d343", "title": "Ranking the effectiveness of worldwide COVID-19 government interventions", "authors": ["Nina Haug", "Lukas Geyrhofer", "Alessandro Londei", "Elma Dervic", "Amélie Desvars-Larrive", "Vittorio Loreto", "Beate Pinior", "Stefan Thurner", "Peter Klimek"], "year": 2020, "venue": "Nature Human Behaviour", "url": "https://doi.org/10.1038/s41562-020-01009-0", "doi": "10.1038/s41562-020-01009-0", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 1.0, "mention_count": 8, "user_count": 8, "months": ["2021-04", "2021-05", "2021-12", "2023-09", "2024-04", "2025-10"], "last_mentioned": "2025-10-01T14:49:23Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1506.05869", "slug": "7f4b10515b6ccce4", "title": "A Neural Conversational Model", "authors": ["Oriol Vinyals", "Quoc Le"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1506.05869", "doi": null, "topics": ["Psychology", "Distributed systems"], "description": "Conversational modeling is an important task in natural language understanding and machine intelligence.", "description_method": "abstract-excerpt", "abstract": "Conversational modeling is an important task in natural language understanding and machine intelligence. Although previous approaches exist, they are often restricted to specific domains (e.g., booking an airline ticket) and require hand-crafted rules. In this paper, we present a simple approach for this task which uses the recently proposed sequence to sequence framework. Our model converses by predicting the next sentence given the previous sentence or sentences in a conversation. The strength of our model is that it can be trained end-to-end and thus requires much fewer hand-crafted rules. We find that this straightforward model can generate simple conversations given a large conversational training dataset. Our preliminary results suggest that, despite optimizing the wrong objective function, the model is able to converse well. It is able extract knowledge from both a domain specific dataset, and from a large, noisy, and general domain dataset of movie subtitles. On a domain-specific IT helpdesk dataset, the model can find a solution to a technical problem via conversations. On a noisy open-domain movie transcript dataset, the model can perform simple forms of common sense reasoning. As expected, we also find that the lack of consistency is a common failure mode of our model.", "score": 1.0, "mention_count": 7, "user_count": 5, "months": ["2015-11", "2016-01", "2016-03", "2016-06", "2020-01", "2026-06"], "last_mentioned": "2026-06-07T04:30:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1510.08555", "slug": "58876c41c977dbcc", "title": "Why Johnny Still, Still Can't Encrypt: Evaluating the Usability of a Modern PGP Client", "authors": ["Scott Ruoti", "Jeff Andersen", "Daniel Zappala", "Kent Seamons"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1510.08555", "doi": null, "topics": ["Human–computer interaction", "Economics"], "description": "This paper presents the results of a laboratory study involving Mailvelope, a modern PGP client that integrates tightly with existing webmail providers.", "description_method": "abstract-excerpt", "abstract": "This paper presents the results of a laboratory study involving Mailvelope, a modern PGP client that integrates tightly with existing webmail providers. In our study, we brought in pairs of participants and had them attempt to use Mailvelope to communicate with each other. Our results shown that more than a decade and a half after \\textit{Why Johnny Can't Encrypt}, modern PGP tools are still unusable for the masses. We finish with a discussion of pain points encountered using Mailvelope, and discuss what might be done to address them in future PGP systems.", "score": 1.0, "mention_count": 7, "user_count": 7, "months": ["2017-05", "2018-11", "2020-01", "2021-11", "2021-12", "2024-03", "2024-04"], "last_mentioned": "2024-04-11T09:21:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1604.06351", "slug": "cea51b128d4b2b2b", "title": "Mission to the Gravitational Focus of the Sun: A Critical Analysis", "authors": ["Geoffrey A. Landis"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1604.06351", "doi": null, "topics": ["Physics"], "description": "The gravitational field of the sun will focus light from a distant source to a focal point at a minimal distance of 550 Astronomical Units from the sun.", "description_method": "abstract-excerpt", "abstract": "The gravitational field of the sun will focus light from a distant source to a focal point at a minimal distance of 550 Astronomical Units from the sun. A proposed mission to this gravitational focus could use the sun as a very large lens, allowing (in principle) a large amplification of signal from the target, and a very high magnification. This article discusses some of the difficulties involved in using the sun as such a gravitational telescope for a candidate mission, that of imaging the surface of a previously-detected exoplanet. These difficulties include the pointing and focal length, and associated high magnification; the signal to noise ratio associated with the solar corona, and the focal blur. In addition, a method to calculate the signal gain and magnification is derived using the first-order deflection calculation and classical optics, showing that the gain is finite for an on-axis source of non-zero area.", "score": 1.0, "mention_count": 7, "user_count": 6, "months": ["2017-05", "2017-08", "2018-02", "2019-07", "2023-04", "2024-08", "2024-10"], "last_mentioned": "2024-10-22T06:20:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1610.08401", "slug": "0f80541ac30fd0f1", "title": "Universal adversarial perturbations", "authors": ["Seyed-Mohsen Moosavi-Dezfooli", "Alhussein Fawzi", "Omar Fawzi", "Pascal Frossard"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1610.08401", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Security & cryptography"], "description": "Given a state-of-the-art deep neural network classifier, we show the existence of a universal (image-agnostic) and very small perturbation vector that causes natural images to be misclassified with high probability.", "description_method": "abstract-excerpt", "abstract": "Given a state-of-the-art deep neural network classifier, we show the existence of a universal (image-agnostic) and very small perturbation vector that causes natural images to be misclassified with high probability. We propose a systematic algorithm for computing universal perturbations, and show that state-of-the-art deep neural networks are highly vulnerable to such perturbations, albeit being quasi-imperceptible to the human eye. We further empirically analyze these universal perturbations and show, in particular, that they generalize very well across neural networks. The surprising existence of universal perturbations reveals important geometric correlations among the high-dimensional decision boundary of classifiers. It further outlines potential security breaches with the existence of single directions in the input space that adversaries can possibly exploit to break a classifier on most natural images.", "score": 1.0, "mention_count": 7, "user_count": 6, "months": ["2016-11", "2017-01", "2017-06", "2017-07", "2017-10", "2018-11"], "last_mentioned": "2018-11-12T06:39:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1704.04299", "slug": "5c89d0754a028ebd", "title": "An Empirical Analysis of Traceability in the Monero Blockchain", "authors": ["Malte Möser", "Kyle Soska", "Ethan Heilman", "Kevin Lee", "Henry Heffan", "Shashvat Srivastava", "Kyle Hogan", "Jason Hennessey", "Andrew Miller", "Arvind Narayanan", "Nicolas Christin"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1704.04299", "doi": null, "topics": ["Databases", "Security & cryptography", "Economics"], "description": "Monero is a privacy-centric cryptocurrency that allows users to obscure their transactions by including chaff coins, called \"mixins,\" along with the actual coins they spend.", "description_method": "abstract-excerpt", "abstract": "Monero is a privacy-centric cryptocurrency that allows users to obscure their transactions by including chaff coins, called \"mixins,\" along with the actual coins they spend. In this paper, we empirically evaluate two weaknesses in Monero's mixin sampling strategy. First, about 62% of transaction inputs with one or more mixins are vulnerable to \"chain-reaction\" analysis -- that is, the real input can be deduced by elimination. Second, Monero mixins are sampled in such a way that they can be easily distinguished from the real coins by their age distribution; in short, the real input is usually the \"newest\" input. We estimate that this heuristic can be used to guess the real input with 80% accuracy over all transactions with 1 or more mixins. Next, we turn to the Monero ecosystem and study the importance of mining pools and the former anonymous marketplace AlphaBay on the transaction volume. We find that after removing mining pool activity, there remains a large amount of potentially privacy-sensitive transactions that are affected by these weaknesses. We propose and evaluate two countermeasures that can improve the privacy of future transactions.", "score": 1.0, "mention_count": 7, "user_count": 6, "months": ["2018-03", "2018-09", "2020-07", "2020-10", "2020-12", "2022-10"], "last_mentioned": "2022-10-17T14:10:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1906.02243", "slug": "884c2f287dad50d5", "title": "Energy and Policy Considerations for Deep Learning in NLP", "authors": ["Emma Strubell", "Ananya Ganesh", "Andrew McCallum"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1906.02243", "doi": null, "topics": ["Machine learning", "Economics"], "description": "Recent progress in hardware and methodology for training neural networks has ushered in a new generation of large networks trained on abundant data.", "description_method": "abstract-excerpt", "abstract": "Recent progress in hardware and methodology for training neural networks has ushered in a new generation of large networks trained on abundant data. These models have obtained notable gains in accuracy across many NLP tasks. However, these accuracy improvements depend on the availability of exceptionally large computational resources that necessitate similarly substantial energy consumption. As a result these models are costly to train and develop, both financially, due to the cost of hardware and electricity or cloud compute time, and environmentally, due to the carbon footprint required to fuel modern tensor processing hardware. In this paper we bring this issue to the attention of NLP researchers by quantifying the approximate financial and environmental costs of training a variety of recently successful neural network models for NLP. Based on these findings, we propose actionable recommendations to reduce costs and improve equity in NLP research and practice.", "score": 1.0, "mention_count": 7, "user_count": 7, "months": ["2019-09", "2019-10", "2021-09", "2021-10", "2023-05", "2024-02"], "last_mentioned": "2024-02-01T13:32:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1907.07587", "slug": "d2b6637c1548f857", "title": "A Differentiable Programming System to Bridge Machine Learning and Scientific Computing", "authors": ["Mike Innes", "Alan Edelman", "Keno Fischer", "Chris Rackauckas", "Elliot Saba", "Viral B Shah", "Will Tebbutt"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1907.07587", "doi": null, "topics": ["Machine learning", "Programming languages", "Science & research"], "description": "Scientific computing is increasingly incorporating the advancements in machine learning and the ability to work with large amounts of data.", "description_method": "abstract-excerpt", "abstract": "Scientific computing is increasingly incorporating the advancements in machine learning and the ability to work with large amounts of data. At the same time, machine learning models are becoming increasingly sophisticated and exhibit many features often seen in scientific computing, stressing the capabilities of machine learning frameworks. Just as the disciplines of scientific computing and machine learning have shared common underlying infrastructure in the form of numerical linear algebra, we now have the opportunity to further share new computational infrastructure, and thus ideas, in the form of Differentiable Programming. We describe Zygote, a Differentiable Programming system that is able to take gradients of general program structures. We implement this system in the Julia programming language. Our system supports almost all language constructs (control flow, recursion, mutation, etc.) and compiles high-performance code without requiring any user intervention or refactoring to stage computations. This enables an expressive programming model for deep learning, but more importantly, it enables us to incorporate a large ecosystem of libraries in our models in a straightforward way. We discuss our approach to automatic differentiation, including its support for advanced techniques such as mixed-mode, complex and checkpointed differentiation, and present several examples of differentiating programs.", "score": 1.0, "mention_count": 7, "user_count": 4, "months": ["2019-08", "2019-10", "2019-11", "2020-01", "2020-03", "2020-04", "2020-06"], "last_mentioned": "2020-06-20T04:06:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2104.13478", "slug": "9804f4f000df9ac6", "title": "Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges", "authors": ["Michael M. Bronstein", "Joan Bruna", "Taco Cohen", "Petar Veličković"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2104.13478", "doi": null, "topics": ["Machine learning", "Mathematics", "Language models"], "description": "The last decade has witnessed an experimental revolution in data science and machine learning, epitomised by deep learning methods.", "description_method": "abstract-excerpt", "abstract": "The last decade has witnessed an experimental revolution in data science and machine learning, epitomised by deep learning methods. Indeed, many high-dimensional learning tasks previously thought to be beyond reach -- such as computer vision, playing Go, or protein folding -- are in fact feasible with appropriate computational scale. Remarkably, the essence of deep learning is built from two simple algorithmic principles: first, the notion of representation or feature learning, whereby adapted, often hierarchical, features capture the appropriate notion of regularity for each task, and second, learning by local gradient-descent type methods, typically implemented as backpropagation. While learning generic functions in high dimensions is a cursed estimation problem, most tasks of interest are not generic, and come with essential pre-defined regularities arising from the underlying low-dimensionality and structure of the physical world. This text is concerned with exposing these regularities through unified geometric principles that can be applied throughout a wide spectrum of applications. Such a 'geometric unification' endeavour, in the spirit of Felix Klein's Erlangen Program, serves a dual purpose: on one hand, it provides a common mathematical framework to study the most successful neural network architectures, such as CNNs, RNNs, GNNs, and Transformers. On the other hand, it gives a constructive procedure to incorporate prior physical knowledge into neural architectures and provide principled way to build future architectures yet to be invented.", "score": 1.0, "mention_count": 7, "user_count": 5, "months": ["2021-05", "2022-04", "2023-01", "2023-02", "2025-06", "2025-09"], "last_mentioned": "2025-09-07T02:23:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1504.00680", "slug": "7435b72191349a45", "title": "Antisocial Behavior in Online Discussion Communities", "authors": ["Justin Cheng", "Cristian Danescu-Niculescu-Mizil", "Jure Leskovec"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1504.00680", "doi": null, "topics": ["Psychology"], "description": "User contributions in the form of posts, comments, and votes are essential to the success of online communities.", "description_method": "abstract-excerpt", "abstract": "User contributions in the form of posts, comments, and votes are essential to the success of online communities. However, allowing user participation also invites undesirable behavior such as trolling. In this paper, we characterize antisocial behavior in three large online discussion communities by analyzing users who were banned from these communities. We find that such users tend to concentrate their efforts in a small number of threads, are more likely to post irrelevantly, and are more successful at garnering responses from other users. Studying the evolution of these users from the moment they join a community up to when they get banned, we find that not only do they write worse than other users over time, but they also become increasingly less tolerated by the community. Further, we discover that antisocial behavior is exacerbated when community feedback is overly harsh. Our analysis also reveals distinct groups of users with different levels of antisocial behavior that can change over time. We use these insights to identify antisocial users early on, a task of high practical importance to community maintainers.", "score": 1.0, "mention_count": 6, "user_count": 5, "months": ["2015-04", "2015-06", "2016-08"], "last_mentioned": "2016-08-04T23:00:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1801.00631", "slug": "9a1929c1421a42e4", "title": "Deep Learning: A Critical Appraisal", "authors": ["Gary Marcus"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1801.00631", "doi": null, "topics": ["Machine learning", "Psychology", "Computer vision"], "description": "Although deep learning has historical roots going back decades, neither the term \"deep learning\" nor the approach was popular just over five years ago, when the field was reignited by papers such as Krizhevsky, Sutskever and Hinton's now classic (2012) deep network model of Imagenet.", "description_method": "abstract-excerpt", "abstract": "Although deep learning has historical roots going back decades, neither the term \"deep learning\" nor the approach was popular just over five years ago, when the field was reignited by papers such as Krizhevsky, Sutskever and Hinton's now classic (2012) deep network model of Imagenet. What has the field discovered in the five subsequent years? Against a background of considerable progress in areas such as speech recognition, image recognition, and game playing, and considerable enthusiasm in the popular press, I present ten concerns for deep learning, and suggest that deep learning must be supplemented by other techniques if we are to reach artificial general intelligence.", "score": 1.0, "mention_count": 6, "user_count": 4, "months": ["2018-04", "2019-01", "2019-11", "2020-06", "2025-06", "2025-11"], "last_mentioned": "2025-11-28T16:39:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.06824", "slug": "39bb32acefd63b20", "title": "Indeterminism in Physics, Classical Chaos and Bohmian Mechanics. Are Real Numbers Really Real?", "authors": ["Nicolas Gisin"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.06824", "doi": "10.1007/s10670-019-00165-8", "topics": ["Physics", "Mathematics"], "description": "It is usual to identify initial conditions of classical dynamical systems with mathematical real numbers.", "description_method": "abstract-excerpt", "abstract": "It is usual to identify initial conditions of classical dynamical systems with mathematical real numbers. However, almost all real numbers contain an infinite amount of information. I argue that a finite volume of space can't contain more than a finite amount of information, hence that the mathematical real numbers are not physically relevant. Moreover, a better terminology for the so-called real numbers is ``random numbers'', as their series of bits are truly random. I propose an alternative classical mechanics, which is empirically equivalent to classical mechanics, but uses only finite-information numbers. This alternative classical mechanics is non-deterministic, despite the use of deterministic equations, in a way similar to quantum theory. Interestingly, both alternative classical mechanics and quantum theories can be supplemented by additional variables in such a way that the supplemented theory is deterministic. Most physicists straightforwardly supplement classical theory with real numbers to which they attribute physical existence, while most physicists reject Bohmian mechanics as supplemented quantum theory, arguing that Bohmian positions have no physical reality.", "score": 1.0, "mention_count": 6, "user_count": 5, "months": ["2018-10", "2018-11", "2021-09", "2023-11", "2024-01", "2026-07"], "last_mentioned": "2026-07-29T13:32:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1810.02137", "slug": "d3a8d35f89f272c8", "title": "Real Differences between OT and CRDT for Co-Editors", "authors": ["Chengzheng Sun", "David Sun", " Agustina", "Weiwei Cai"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1810.02137", "doi": null, "topics": ["Physics", "Distributed systems", "Algorithms"], "description": "OT (Operational Transformation) was invented for supporting real-time co-editors in the late 1980s and has evolved to become a core technique used in today's working co-editors and adopted in major industrial products.", "description_method": "abstract-excerpt", "abstract": "OT (Operational Transformation) was invented for supporting real-time co-editors in the late 1980s and has evolved to become a core technique used in today's working co-editors and adopted in major industrial products. CRDT (Commutative Replicated Data Type) for co-editors was first proposed around 2006, under the name of WOOT (WithOut Operational Transformation). Follow-up CRDT variations are commonly labeled as \"post-OT\" techniques capable of making concurrent operations natively commutative, and have made broad claims of superiority over OT solutions, in terms of correctness, time and space complexity, simplicity, etc. Over one decade later, however, CRDT solutions are rarely found in working co-editors, while OT solutions remain the choice for building the vast majority of co-editors. Contradictions between this reality and CRDT's purported advantages have been the source of much debate and confusion in co-editing research and developer communities. What is CRDT really to co-editing? What are the real differences between OT and CRDT for co-editors? What are the key factors that may have affected the adoption of and choice between OT and CRDT for co-editors in the real world? In this paper, we report our discoveries, in relation to these questions and beyond, from a comprehensive review and comparison study on representative OT and CRDT solutions and working co-editors based on them. Moreover, this work reveals facts and presents evidences that refute CRDT claimed advantages over OT. We hope the results reported in this paper will help clear up common myths, misconceptions, and confusions surrounding alternative co-editing techniques, and accelerate progress in co-editing technology for real world applications.", "score": 1.0, "mention_count": 6, "user_count": 5, "months": ["2018-10", "2019-05", "2021-07", "2021-09", "2023-06", "2023-07"], "last_mentioned": "2023-07-09T19:08:31Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1912.10077", "slug": "6441ba16c3d2d11a", "title": "Are Transformers universal approximators of sequence-to-sequence functions?", "authors": ["Chulhee Yun", "Srinadh Bhojanapalli", "Ankit Singh Rawat", "Sashank J. Reddi", "Sanjiv Kumar"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1912.10077", "doi": null, "topics": ["Language models"], "description": "Despite the widespread adoption of Transformer models for NLP tasks, the expressive power of these models is not well-understood.", "description_method": "abstract-excerpt", "abstract": "Despite the widespread adoption of Transformer models for NLP tasks, the expressive power of these models is not well-understood. In this paper, we establish that Transformer models are universal approximators of continuous permutation equivariant sequence-to-sequence functions with compact support, which is quite surprising given the amount of shared parameters in these models. Furthermore, using positional encodings, we circumvent the restriction of permutation equivariance, and show that Transformer models can universally approximate arbitrary continuous sequence-to-sequence functions on a compact domain. Interestingly, our proof techniques clearly highlight the different roles of the self-attention and the feed-forward layers in Transformers. In particular, we prove that fixed width self-attention layers can compute contextual mappings of the input sequences, playing a key role in the universal approximation property of Transformers. Based on this insight from our analysis, we consider other simpler alternatives to self-attention layers and empirically evaluate them.", "score": 1.0, "mention_count": 6, "user_count": 6, "months": ["2023-05", "2024-03", "2025-11", "2026-02", "2026-06"], "last_mentioned": "2026-06-18T08:56:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2112.04035", "slug": "bfa03c56afb2b9dc", "title": "Relating transformers to models and neural representations of the hippocampal formation", "authors": ["James C. R. Whittington", "Joseph Warren", "Timothy E. J. Behrens"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2112.04035", "doi": null, "topics": ["Neuroscience", "Language models", "Machine learning"], "description": "Many deep neural network architectures loosely based on brain networks have recently been shown to replicate neural firing patterns observed in the brain.", "description_method": "abstract-excerpt", "abstract": "Many deep neural network architectures loosely based on brain networks have recently been shown to replicate neural firing patterns observed in the brain. One of the most exciting and promising novel architectures, the Transformer neural network, was developed without the brain in mind. In this work, we show that transformers, when equipped with recurrent position encodings, replicate the precisely tuned spatial representations of the hippocampal formation; most notably place and grid cells. Furthermore, we show that this result is no surprise since it is closely related to current hippocampal models from neuroscience. We additionally show the transformer version offers dramatic performance gains over the neuroscience version. This work continues to bind computations of artificial and brain networks, offers a novel understanding of the hippocampal-cortical interaction, and suggests how wider cortical areas may perform complex tasks beyond current neuroscience models such as language comprehension.", "score": 1.0, "mention_count": 6, "user_count": 5, "months": ["2022-04", "2022-12", "2024-01", "2024-06", "2024-09", "2025-11"], "last_mentioned": "2025-11-03T20:05:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2203.14465", "slug": "9c3c6d83941546ca", "title": "STaR: Bootstrapping Reasoning With Reasoning", "authors": ["Eric Zelikman", "Yuhuai Wu", "Jesse Mu", "Noah D. Goodman"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2203.14465", "doi": null, "topics": ["Language models", "Mathematics"], "description": "Generating step-by-step \"chain-of-thought\" rationales improves language model performance on complex reasoning tasks like mathematics or commonsense question-answering.", "description_method": "abstract-excerpt", "abstract": "Generating step-by-step \"chain-of-thought\" rationales improves language model performance on complex reasoning tasks like mathematics or commonsense question-answering. However, inducing language model rationale generation currently requires either constructing massive rationale datasets or sacrificing accuracy by using only few-shot inference. We propose a technique to iteratively leverage a small number of rationale examples and a large dataset without rationales, to bootstrap the ability to perform successively more complex reasoning. This technique, the \"Self-Taught Reasoner\" (STaR), relies on a simple loop: generate rationales to answer many questions, prompted with a few rationale examples; if the generated answers are wrong, try again to generate a rationale given the correct answer; fine-tune on all the rationales that ultimately yielded correct answers; repeat. We show that STaR significantly improves performance on multiple datasets compared to a model fine-tuned to directly predict final answers, and performs comparably to fine-tuning a 30$\\times$ larger state-of-the-art language model on CommensenseQA. Thus, STaR lets a model improve itself by learning from its own generated reasoning.", "score": 1.0, "mention_count": 6, "user_count": 5, "months": ["2022-03", "2023-12", "2024-07", "2024-08", "2024-12"], "last_mentioned": "2024-12-22T22:17:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.04388", "slug": "173275d761734295", "title": "Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting", "authors": ["Miles Turpin", "Julian Michael", "Ethan Perez", "Samuel R. Bowman"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.04388", "doi": null, "topics": ["Language models", "Psychology"], "description": "Large Language Models (LLMs) can achieve strong performance on many tasks by producing step-by-step reasoning before giving a final output, often referred to as chain-of-thought reasoning (CoT).", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) can achieve strong performance on many tasks by producing step-by-step reasoning before giving a final output, often referred to as chain-of-thought reasoning (CoT). It is tempting to interpret these CoT explanations as the LLM's process for solving a task. This level of transparency into LLMs' predictions would yield significant safety benefits. However, we find that CoT explanations can systematically misrepresent the true reason for a model's prediction. We demonstrate that CoT explanations can be heavily influenced by adding biasing features to model inputs--e.g., by reordering the multiple-choice options in a few-shot prompt to make the answer always \"(A)\"--which models systematically fail to mention in their explanations. When we bias models toward incorrect answers, they frequently generate CoT explanations rationalizing those answers. This causes accuracy to drop by as much as 36% on a suite of 13 tasks from BIG-Bench Hard, when testing with GPT-3.5 from OpenAI and Claude 1.0 from Anthropic. On a social-bias task, model explanations justify giving answers in line with stereotypes without mentioning the influence of these social biases. Our findings indicate that CoT explanations can be plausible yet misleading, which risks increasing our trust in LLMs without guaranteeing their safety. Building more transparent and explainable systems will require either improving CoT faithfulness through targeted efforts or abandoning CoT in favor of alternative methods.", "score": 1.0, "mention_count": 6, "user_count": 6, "months": ["2023-05", "2024-12", "2025-02", "2025-04", "2025-07", "2025-11"], "last_mentioned": "2025-11-18T21:24:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.06816", "slug": "356bed90e86159dd", "title": "Text Embeddings Reveal (Almost) As Much As Text", "authors": ["John X. Morris", "Volodymyr Kuleshov", "Vitaly Shmatikov", "Alexander M. Rush"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.06816", "doi": null, "topics": ["Biology & medicine"], "description": "How much private information do text embeddings reveal about the original text?", "description_method": "abstract-excerpt", "abstract": "How much private information do text embeddings reveal about the original text? We investigate the problem of embedding \\textit{inversion}, reconstructing the full text represented in dense text embeddings. We frame the problem as controlled generation: generating text that, when reembedded, is close to a fixed point in latent space. We find that although a naïve model conditioned on the embedding performs poorly, a multi-step method that iteratively corrects and re-embeds text is able to recover $92\\%$ of $32\\text{-token}$ text inputs exactly. We train our model to decode text embeddings from two state-of-the-art embedding models, and also show that our model can recover important personal information (full names) from a dataset of clinical notes. Our code is available on Github: \\href{https://github.com/jxmorris12/vec2text}{github.com/jxmorris12/vec2text}.", "score": 1.0, "mention_count": 6, "user_count": 5, "months": ["2023-11", "2023-12", "2024-03", "2024-06", "2024-07", "2025-01"], "last_mentioned": "2025-01-17T23:24:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2401.17377", "slug": "69a9199c2c8e06ce", "title": "Infini-gram: Scaling Unbounded n-gram Language Models to a Trillion Tokens", "authors": ["Jiacheng Liu", "Sewon Min", "Luke Zettlemoyer", "Yejin Choi", "Hannaneh Hajishirzi"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2401.17377", "doi": null, "topics": ["Language models"], "description": "Are $n$-gram language models still relevant in this era of neural large language models (LLMs)?", "description_method": "abstract-excerpt", "abstract": "Are $n$-gram language models still relevant in this era of neural large language models (LLMs)? Our answer is yes, and we showcase their values in both text analysis and improving neural LLMs. This was done by modernizing $n$-gram LMs in two aspects. First, we train them at the same data scale as neural LLMs -- 5 trillion tokens. This is the largest $n$-gram LM ever built. Second, existing $n$-gram LMs use small $n$ which hinders their performance; we instead allow $n$ to be arbitrarily large, by introducing a new $\\infty$-gram LM with backoff. Instead of pre-computing $n$-gram count tables (which would be very expensive), we develop an engine named infini-gram -- powered by suffix arrays -- that can compute $\\infty$-gram (as well as $n$-gram with arbitrary $n$) probabilities with millisecond-level latency. The $\\infty$-gram framework and infini-gram engine enable us to conduct many novel and interesting analyses of human-written and machine-generated text: we find that the $\\infty$-gram LM has fairly high accuracy for next-token prediction (47%), and can complement neural LLMs to greatly reduce their perplexity. When analyzing machine-generated text, we also observe irregularities in the machine--$\\infty$-gram agreement level with respect to the suffix length, which indicates deficiencies in neural LLM pretraining and the positional embeddings of Transformers.", "score": 1.0, "mention_count": 6, "user_count": 4, "months": ["2024-02", "2024-06", "2025-11", "2026-04"], "last_mentioned": "2026-04-21T08:47:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2406.11717", "slug": "4afd669d1feb459f", "title": "Refusal in Language Models Is Mediated by a Single Direction", "authors": ["Andy Arditi", "Oscar Obeso", "Aaquib Syed", "Daniel Paleka", "Nina Panickssery", "Wes Gurnee", "Neel Nanda"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2406.11717", "doi": null, "topics": ["Language models", "Psychology"], "description": "Conversational large language models are fine-tuned for both instruction-following and safety, resulting in models that obey benign requests but refuse harmful ones.", "description_method": "abstract-excerpt", "abstract": "Conversational large language models are fine-tuned for both instruction-following and safety, resulting in models that obey benign requests but refuse harmful ones. While this refusal behavior is widespread across chat models, its underlying mechanisms remain poorly understood. In this work, we show that refusal is mediated by a one-dimensional subspace, across 13 popular open-source chat models up to 72B parameters in size. Specifically, for each model, we find a single direction such that erasing this direction from the model's residual stream activations prevents it from refusing harmful instructions, while adding this direction elicits refusal on even harmless instructions. Leveraging this insight, we propose a novel white-box jailbreak method that surgically disables refusal with minimal effect on other capabilities. Finally, we mechanistically analyze how adversarial suffixes suppress propagation of the refusal-mediating direction. Our findings underscore the brittleness of current safety fine-tuning methods. More broadly, our work showcases how an understanding of model internals can be leveraged to develop practical methods for controlling model behavior.", "score": 1.0, "mention_count": 6, "user_count": 5, "months": ["2024-07", "2025-11", "2026-07"], "last_mentioned": "2026-07-25T16:44:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2505.24832", "slug": "1d14faeb604373b8", "title": "How much do language models memorize?", "authors": ["John X. Morris", "Chawin Sitawarin", "Chuan Guo", "Narine Kokhlikyan", "G. Edward Suh", "Alexander M. Rush", "Kamalika Chaudhuri", "Saeed Mahloujifar"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2505.24832", "doi": null, "topics": ["Language models", "Machine learning"], "description": "We propose a new method for estimating how much a model knows about a datapoint and use it to measure the capacity of modern language models.", "description_method": "abstract-excerpt", "abstract": "We propose a new method for estimating how much a model knows about a datapoint and use it to measure the capacity of modern language models. Prior studies of language model memorization have struggled to disentangle memorization from generalization. We formally separate memorization into two components: unintended memorization, the information a model contains about a specific dataset, and generalization, the information a model contains about the true data-generation process. When we completely eliminate generalization, we can compute the total memorization, which provides an estimate of model capacity: our measurements estimate that GPT-style models have a capacity of approximately 3.6 bits per parameter. We train language models on datasets of increasing size and observe that models memorize until their capacity fills, at which point \"grokking\" begins, and unintended memorization decreases as models begin to generalize. We train hundreds of transformer language models ranging from $500K$ to $1.5B$ parameters and produce a series of scaling laws relating model capacity and data size to membership inference.", "score": 1.0, "mention_count": 6, "user_count": 5, "months": ["2025-06", "2025-07", "2025-09"], "last_mentioned": "2025-09-30T08:31:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1202.4212", "slug": "79760453ca6b9755", "title": "Harmony Explained: Progress Towards A Scientific Theory of Music", "authors": ["Daniel Shawcross Wilkerson"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1202.4212", "doi": null, "topics": ["Science & research", "Neuroscience", "Biology & medicine"], "description": "Most music theory books are like medieval medical textbooks: they contain unjustified superstition, non-reasoning, and funny symbols glorified by Latin phrases.", "description_method": "abstract-excerpt", "abstract": "Most music theory books are like medieval medical textbooks: they contain unjustified superstition, non-reasoning, and funny symbols glorified by Latin phrases. How does music, in particular harmony, actually work, presented as a real, scientific theory of music? The core to our approach is to consider not only the Physical phenomena of nature but also the Computational phenomena of any machine that must make sense of sound, such as the human brain. In particular we derive the following three fundamental phenomena of music: * the Major Scale, * the Standard Chord Dictionary, and * the difference in feeling between the Major and Minor Triads. While the Major Scale has been independently derived before by others in a similar manner [Helmholtz1863, Birkhoff1933], I believe the derivation of the Standard Chord Dictionary as well as the difference in feeling between the Major and Minor Triads to be original. We show to be incomplete the theory of the heretofore agreed-upon authority on this subject, 19th-century Physicist Hermann Helmholtz [Helmholtz1863]: he says notes are in \"concord\" because the sound playing them together is \"less worse\" than that of some other notes. But note that, in this theory, more notes can only penalize, some merely less than others, and so the most harmonious sound should be a single note by itself(!) and harmony would not exist as a phenomenon of music at all. I intend this article to be satisfying to scientists as an original contribution to science and art, yet I also intend it to be approachable by musicians and other curious members of the general public who may have long wondered at the curious properties of tonal music and been frustrated by the lack of satisfying, readable exposition on the subject. Therefore I have written in a deliberately plain and conversational style, avoiding unnecessarily formal language.", "score": 1.0, "mention_count": 5, "user_count": 5, "months": ["2014-10", "2016-03", "2016-07", "2016-10"], "last_mentioned": "2016-10-26T06:03:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1401.1219", "slug": "71a7406d08f0b149", "title": "Consciousness as a State of Matter", "authors": ["Max Tegmark"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1401.1219", "doi": "10.1016/j.chaos.2015.03.014", "topics": ["Neuroscience", "Physics", "Language models"], "description": "We examine the hypothesis that consciousness can be understood as a state of matter, \"perceptronium\", with distinctive information processing abilities.", "description_method": "abstract-excerpt", "abstract": "We examine the hypothesis that consciousness can be understood as a state of matter, \"perceptronium\", with distinctive information processing abilities. We explore five basic principles that may distinguish conscious matter from other physical systems such as solids, liquids and gases: the information, integration, independence, dynamics and utility principles. If such principles can identify conscious entities, then they can help solve the quantum factorization problem: why do conscious observers like us perceive the particular Hilbert space factorization corresponding to classical space (rather than Fourier space, say), and more generally, why do we perceive the world around us as a dynamic hierarchy of objects that are strongly integrated and relatively independent? Tensor factorization of matrices is found to play a central role, and our technical results include a theorem about Hamiltonian separability (defined using Hilbert-Schmidt superoperators) being maximized in the energy eigenbasis. Our approach generalizes Giulio Tononi's integrated information framework for neural-network-based consciousness to arbitrary quantum systems, and we find interesting links to error-correcting codes, condensed matter criticality, and the Quantum Darwinism program, as well as an interesting connection between the emergence of consciousness and the emergence of time.", "score": 1.0, "mention_count": 5, "user_count": 5, "months": ["2015-12", "2017-07", "2017-09", "2019-10"], "last_mentioned": "2019-10-31T03:12:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1401.5577", "slug": "7dafe517c52529b1", "title": "Robust Cooperation in the Prisoner's Dilemma: Program Equilibrium via Provability Logic", "authors": ["Mihaly Barasz", "Paul Christiano", "Benja Fallenstein", "Marcello Herreshoff", "Patrick LaVictoire", "Eliezer Yudkowsky"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1401.5577", "doi": null, "topics": ["Algorithms"], "description": "We consider the one-shot Prisoner's Dilemma between algorithms with read-access to one anothers' source codes, and we use the modal logic of provability to build agents that can achieve mutual cooperation in a manner that is robust, in that cooperation does not require exact equality of the agents' source code, and unexploitable, meaning ", "description_method": "abstract-excerpt", "abstract": "We consider the one-shot Prisoner's Dilemma between algorithms with read-access to one anothers' source codes, and we use the modal logic of provability to build agents that can achieve mutual cooperation in a manner that is robust, in that cooperation does not require exact equality of the agents' source code, and unexploitable, meaning that such an agent never cooperates when its opponent defects. We construct a general framework for such \"modal agents\", and study their properties.", "score": 1.0, "mention_count": 5, "user_count": 5, "months": ["2015-04", "2019-01", "2022-04", "2022-06", "2025-12"], "last_mentioned": "2025-12-30T20:00:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1410.3916", "slug": "60b2063d8187d069", "title": "Memory Networks", "authors": ["Jason Weston", "Sumit Chopra", "Antoine Bordes"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1410.3916", "doi": null, "topics": ["Other research"], "description": "We describe a new class of learning models called memory networks.", "description_method": "abstract-excerpt", "abstract": "We describe a new class of learning models called memory networks. Memory networks reason with inference components combined with a long-term memory component; they learn how to use these jointly. The long-term memory can be read and written to, with the goal of using it for prediction. We investigate these models in the context of question answering (QA) where the long-term memory effectively acts as a (dynamic) knowledge base, and the output is a textual response. We evaluate them on a large-scale QA task, and a smaller, but more complex, toy task generated from a simulated world. In the latter, we show the reasoning power of such models by chaining multiple supporting sentences to answer questions that require understanding the intension of verbs.", "score": 1.0, "mention_count": 5, "user_count": 3, "months": ["2015-03", "2015-05", "2015-07", "2016-10", "2025-10"], "last_mentioned": "2025-10-09T09:19:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1503.02406", "slug": "944446745d8c10cd", "title": "Deep Learning and the Information Bottleneck Principle", "authors": ["Naftali Tishby", "Noga Zaslavsky"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1503.02406", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "Deep Neural Networks (DNNs) are analyzed via the theoretical framework of the information bottleneck (IB) principle.", "description_method": "abstract-excerpt", "abstract": "Deep Neural Networks (DNNs) are analyzed via the theoretical framework of the information bottleneck (IB) principle. We first show that any DNN can be quantified by the mutual information between the layers and the input and output variables. Using this representation we can calculate the optimal information theoretic limits of the DNN and obtain finite sample generalization bounds. The advantage of getting closer to the theoretical limit is quantifiable both by the generalization bound and by the network's simplicity. We argue that both the optimal architecture, number of layers and features/connections at each layer, are related to the bifurcation points of the information bottleneck tradeoff, namely, relevant compression of the input layer with respect to the output layer. The hierarchical representations at the layered network naturally correspond to the structural phase transitions along the information curve. We believe that this new insight can lead to new optimality bounds and deep learning algorithms.", "score": 1.0, "mention_count": 5, "user_count": 4, "months": ["2016-11", "2018-05", "2023-08"], "last_mentioned": "2023-08-11T11:01:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1506.02640", "slug": "03057b86dddf3d65", "title": "You Only Look Once: Unified, Real-Time Object Detection", "authors": ["Joseph Redmon", "Santosh Divvala", "Ross Girshick", "Ali Farhadi"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1506.02640", "doi": null, "topics": ["Computer vision", "Machine learning", "Statistics & probability"], "description": "We present YOLO, a new approach to object detection.", "description_method": "abstract-excerpt", "abstract": "We present YOLO, a new approach to object detection. Prior work on object detection repurposes classifiers to perform detection. Instead, we frame object detection as a regression problem to spatially separated bounding boxes and associated class probabilities. A single neural network predicts bounding boxes and class probabilities directly from full images in one evaluation. Since the whole detection pipeline is a single network, it can be optimized end-to-end directly on detection performance. Our unified architecture is extremely fast. Our base YOLO model processes images in real-time at 45 frames per second. A smaller version of the network, Fast YOLO, processes an astounding 155 frames per second while still achieving double the mAP of other real-time detectors. Compared to state-of-the-art detection systems, YOLO makes more localization errors but is far less likely to predict false detections where nothing exists. Finally, YOLO learns very general representations of objects. It outperforms all other detection methods, including DPM and R-CNN, by a wide margin when generalizing from natural images to artwork on both the Picasso Dataset and the People-Art Dataset.", "score": 1.0, "mention_count": 5, "user_count": 4, "months": ["2016-03", "2017-06", "2017-10", "2019-04", "2025-10"], "last_mentioned": "2025-10-23T18:59:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1512.00727", "slug": "b9c5dd327ab9d245", "title": "TinyLFU: A Highly Efficient Cache Admission Policy", "authors": ["Gil Einziger", "Roy Friedman", "Ben Manes"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1512.00727", "doi": null, "topics": ["Other research"], "description": "This paper proposes to use a frequency based cache admission policy in order to boost the effectiveness of caches subject to skewed access distributions.", "description_method": "abstract-excerpt", "abstract": "This paper proposes to use a frequency based cache admission policy in order to boost the effectiveness of caches subject to skewed access distributions. Given a newly accessed item and an eviction candidate from the cache, our scheme decides, based on the recent access history, whether it is worth admitting the new item into the cache at the expense of the eviction candidate. Realizing this concept is enabled through a novel approximate LFU structure called TinyLFU, which maintains an approximate representation of the access frequency of a large sample of recently accessed items. TinyLFU is very compact and light-weight as it builds upon Bloom filter theory. We study the properties of TinyLFU through simulations of both synthetic workloads as well as multiple real traces from several sources. These simulations demonstrate the performance boost obtained by enhancing various replacement policies with the TinyLFU eviction policy. Also, a new combined replacement and eviction policy scheme nicknamed W-TinyLFU is presented. W-TinyLFU is demonstrated to obtain equal or better hit-ratios than other state of the art replacement policies on these traces. It is the only scheme to obtain such good results on all traces.", "score": 1.0, "mention_count": 5, "user_count": 4, "months": ["2016-01", "2016-04", "2019-06", "2020-08", "2021-08"], "last_mentioned": "2021-08-11T08:30:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1601.00670", "slug": "175a713d5490d42c", "title": "Variational Inference: A Review for Statisticians", "authors": ["David M. Blei", "Alp Kucukelbir", "Jon D. McAuliffe"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1601.00670", "doi": "10.1080/01621459.2017.1285773", "topics": ["Statistics & probability", "Algorithms", "Machine learning"], "description": "One of the core problems of modern statistics is to approximate difficult-to-compute probability densities.", "description_method": "abstract-excerpt", "abstract": "One of the core problems of modern statistics is to approximate difficult-to-compute probability densities. This problem is especially important in Bayesian statistics, which frames all inference about unknown quantities as a calculation involving the posterior density. In this paper, we review variational inference (VI), a method from machine learning that approximates probability densities through optimization. VI has been used in many applications and tends to be faster than classical methods, such as Markov chain Monte Carlo sampling. The idea behind VI is to first posit a family of densities and then to find the member of that family which is close to the target. Closeness is measured by Kullback-Leibler divergence. We review the ideas behind mean-field variational inference, discuss the special case of VI applied to exponential family models, present a full example with a Bayesian mixture of Gaussians, and derive a variant that uses stochastic optimization to scale up to massive data. We discuss modern research in VI and highlight important open problems. VI is powerful, but it is not yet well understood. Our hope in writing this paper is to catalyze statistical research on this class of algorithms.", "score": 1.0, "mention_count": 5, "user_count": 5, "months": ["2016-10", "2019-03", "2021-10", "2022-04", "2023-03"], "last_mentioned": "2023-03-29T15:30:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1710.07047", "slug": "0e8eb961275b8a10", "title": "Safe Pointers in SPARK 2014", "authors": ["Georges-Axel Jaloyan"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1710.07047", "doi": null, "topics": ["Programming languages"], "description": "In the context of deductive software verification, programs with pointers present a major challenge due to pointer aliasing.", "description_method": "abstract-excerpt", "abstract": "In the context of deductive software verification, programs with pointers present a major challenge due to pointer aliasing. In this paper, we introduce pointers to SPARK, a well-defined subset of the Ada language, intended for formal verification of mission-critical software. Our solution is based on static alias analysis inspired by Rust's borrow-checker and affine types, and enforces the Concurrent Read, Exclusive Write principle. This analysis has been implemented in the GNAT Ada compiler and tested against a number of challenging examples including parts of real-life applications. Our tests show that only minor changes in the source code are required to fit the idiomatic Ada code into SPARK extended with pointers, which is a significant improvement upon the previous state of the art. The proposed extension has been approved by the Language Design Committee for SPARK for inclusion in a future version of SPARK, and is being discussed by the Ada Rapporteur Group for inclusion in the next version of Ada. In the report, we give a formal presentation of the analysis rules for a miniature version of SPARK and prove their soundness. We discuss the implementation and the case studies, and compare our solution with Rust.", "score": 1.0, "mention_count": 5, "user_count": 3, "months": ["2017-12", "2019-04", "2019-09", "2021-03", "2021-05"], "last_mentioned": "2021-05-11T15:18:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1805.04263", "slug": "da0ffe8666c1f0c5", "title": "OpSets: Sequential Specifications for Replicated Datatypes (Extended Version)", "authors": ["Martin Kleppmann", "Victor B. F. Gomes", "Dominic P. Mulligan", "Alastair R. Beresford"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1805.04263", "doi": null, "topics": ["Distributed systems", "Algorithms", "Mathematics"], "description": "We introduce OpSets, an executable framework for specifying and reasoning about the semantics of replicated datatypes that provide eventual consistency in a distributed system, and for mechanically verifying algorithms that implement these datatypes.", "description_method": "abstract-excerpt", "abstract": "We introduce OpSets, an executable framework for specifying and reasoning about the semantics of replicated datatypes that provide eventual consistency in a distributed system, and for mechanically verifying algorithms that implement these datatypes. Our approach is simple but expressive, allowing us to succinctly specify a variety of abstract datatypes, including maps, sets, lists, text, graphs, trees, and registers. Our datatypes are also composable, enabling the construction of complex data structures. To demonstrate the utility of OpSets for analysing replication algorithms, we highlight an important correctness property for collaborative text editing that has traditionally been overlooked; algorithms that do not satisfy this property can exhibit awkward interleaving of text. We use OpSets to specify this correctness property and prove that although one existing replication algorithm satisfies this property, several other published algorithms do not. We also show how OpSets can be used to develop new replicated datatypes: we provide a simple specification of an atomic move operation for trees, an operation that had previously been thought to be impossible to implement without locking. We use the Isabelle/HOL proof assistant to formalise the OpSets approach and produce mechanised proofs of correctness of the main claims in this paper, thereby eliminating the ambiguity of previous informal approaches, and ruling out reasoning errors that could occur in handwritten proofs.", "score": 1.0, "mention_count": 5, "user_count": 4, "months": ["2020-01", "2020-04", "2020-12", "2021-06", "2026-05"], "last_mentioned": "2026-05-21T21:29:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1806.06726", "slug": "5dc8660437949dee", "title": "Zip Trees", "authors": ["Robert E. Tarjan", "Caleb C. Levy", "Stephen Timmel"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1806.06726", "doi": "10.1145/3476830", "topics": ["Statistics & probability", "Algorithms"], "description": "We introduce the zip tree, a form of randomized binary search tree that integrates previous ideas into one practical, performant, and pleasant-to-implement package.", "description_method": "abstract-excerpt", "abstract": "We introduce the zip tree, a form of randomized binary search tree that integrates previous ideas into one practical, performant, and pleasant-to-implement package. A zip tree is a binary search tree in which each node has a numeric rank and the tree is (max)-heap-ordered with respect to ranks, with rank ties broken in favor of smaller keys. Zip trees are essentially treaps (Seidel and Aragon 1996), except that ranks are drawn from a geometric distribution instead of a uniform distribution, and we allow rank ties. These changes enable us to use fewer random bits per node. We perform insertions and deletions by unmerging and merging paths (\"unzipping\" and \"zipping\") rather than by doing rotations, which avoids some pointer changes and improves efficiency. The methods of zipping and unzipping take inspiration from previous top-down approaches to insertion and deletion (Stephenson 1980; Martínez and Roura 1998; Sprugnoli 1980). From a theoretical standpoint, this work provides two main results. First, zip trees require only $O(\\log \\log n)$ bits (with high probability) to represent the largest rank in an $n$-node binary search tree; previous data structures require $O(\\log n)$ bits for the largest rank. Second, zip trees are naturally isomorphic to skip lists (Pugh 1990), and simplify the mapping of (Dean and Jones 2007) between skip lists and binary search trees.", "score": 1.0, "mention_count": 5, "user_count": 3, "months": ["2019-09", "2022-08", "2023-05", "2024-03", "2026-04"], "last_mentioned": "2026-04-20T04:03:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1902.04738", "slug": "d3a8c8f2092f0133", "title": "Mesh: Compacting Memory Management for C/C++ Applications", "authors": ["Bobby Powers", "David Tench", "Emery D. Berger", "Andrew McGregor"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1902.04738", "doi": "10.1145/3314221.3314582", "topics": ["Computer systems", "Statistics & probability", "Algorithms"], "description": "Programs written in C/C++ can suffer from serious memory fragmentation, leading to low utilization of memory, degraded performance, and application failure due to memory exhaustion.", "description_method": "abstract-excerpt", "abstract": "Programs written in C/C++ can suffer from serious memory fragmentation, leading to low utilization of memory, degraded performance, and application failure due to memory exhaustion. This paper introduces Mesh, a plug-in replacement for malloc that, for the first time, eliminates fragmentation in unmodified C/C++ applications. Mesh combines novel randomized algorithms with widely-supported virtual memory operations to provably reduce fragmentation, breaking the classical Robson bounds with high probability. Mesh generally matches the runtime performance of state-of-the-art memory allocators while reducing memory consumption; in particular, it reduces the memory of consumption of Firefox by 16% and Redis by 39%.", "score": 1.0, "mention_count": 5, "user_count": 5, "months": ["2019-05", "2021-05", "2023-07", "2025-01"], "last_mentioned": "2025-01-11T12:48:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2002.05202", "slug": "e052a2974593d293", "title": "GLU Variants Improve Transformer", "authors": ["Noam Shazeer"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2002.05202", "doi": null, "topics": ["Language models"], "description": "Gated Linear Units (arXiv:1612.08083) consist of the component-wise product of two linear projections, one of which is first passed through a sigmoid function.", "description_method": "abstract-excerpt", "abstract": "Gated Linear Units (arXiv:1612.08083) consist of the component-wise product of two linear projections, one of which is first passed through a sigmoid function. Variations on GLU are possible, using different nonlinear (or even linear) functions in place of sigmoid. We test these variants in the feed-forward sublayers of the Transformer (arXiv:1706.03762) sequence-to-sequence model, and find that some of them yield quality improvements over the typically-used ReLU or GELU activations.", "score": 1.0, "mention_count": 5, "user_count": 5, "months": ["2023-08", "2023-11", "2024-09", "2026-06"], "last_mentioned": "2026-06-21T14:36:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2205.13504", "slug": "5b4b3b583572a6da", "title": "Are Transformers Effective for Time Series Forecasting?", "authors": ["Ailing Zeng", "Muxi Chen", "Lei Zhang", "Qiang Xu"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2205.13504", "doi": null, "topics": ["Language models"], "description": "Recently, there has been a surge of Transformer-based solutions for the long-term time series forecasting (LTSF) task.", "description_method": "abstract-excerpt", "abstract": "Recently, there has been a surge of Transformer-based solutions for the long-term time series forecasting (LTSF) task. Despite the growing performance over the past few years, we question the validity of this line of research in this work. Specifically, Transformers is arguably the most successful solution to extract the semantic correlations among the elements in a long sequence. However, in time series modeling, we are to extract the temporal relations in an ordered set of continuous points. While employing positional encoding and using tokens to embed sub-series in Transformers facilitate preserving some ordering information, the nature of the \\emph{permutation-invariant} self-attention mechanism inevitably results in temporal information loss. To validate our claim, we introduce a set of embarrassingly simple one-layer linear models named LTSF-Linear for comparison. Experimental results on nine real-life datasets show that LTSF-Linear surprisingly outperforms existing sophisticated Transformer-based LTSF models in all cases, and often by a large margin. Moreover, we conduct comprehensive empirical studies to explore the impacts of various design elements of LTSF models on their temporal relation extraction capability. We hope this surprising finding opens up new research directions for the LTSF task. We also advocate revisiting the validity of Transformer-based solutions for other time series analysis tasks (e.g., anomaly detection) in the future. Code is available at: \\url{https://github.com/cure-lab/LTSF-Linear}.", "score": 1.0, "mention_count": 5, "user_count": 5, "months": ["2023-04", "2024-02", "2024-07", "2025-06", "2025-12"], "last_mentioned": "2025-12-23T15:27:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2303.06349", "slug": "d34263747873694b", "title": "Resurrecting Recurrent Neural Networks for Long Sequences", "authors": ["Antonio Orvieto", "Samuel L Smith", "Albert Gu", "Anushan Fernando", "Caglar Gulcehre", "Razvan Pascanu", "Soham De"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2303.06349", "doi": null, "topics": ["Machine learning"], "description": "Recurrent Neural Networks (RNNs) offer fast inference on long sequences but are hard to optimize and slow to train.", "description_method": "abstract-excerpt", "abstract": "Recurrent Neural Networks (RNNs) offer fast inference on long sequences but are hard to optimize and slow to train. Deep state-space models (SSMs) have recently been shown to perform remarkably well on long sequence modeling tasks, and have the added benefits of fast parallelizable training and RNN-like fast inference. However, while SSMs are superficially similar to RNNs, there are important differences that make it unclear where their performance boost over RNNs comes from. In this paper, we show that careful design of deep RNNs using standard signal propagation arguments can recover the impressive performance of deep SSMs on long-range reasoning tasks, while also matching their training speed. To achieve this, we analyze and ablate a series of changes to standard RNNs including linearizing and diagonalizing the recurrence, using better parameterizations and initializations, and ensuring proper normalization of the forward pass. Our results provide new insights on the origins of the impressive performance of deep SSMs, while also introducing an RNN block called the Linear Recurrent Unit that matches both their performance on the Long Range Arena benchmark and their computational efficiency.", "score": 1.0, "mention_count": 5, "user_count": 4, "months": ["2023-04", "2023-07", "2024-10", "2025-03"], "last_mentioned": "2025-03-15T11:27:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.18290", "slug": "3bdad22db8b074df", "title": "Direct Preference Optimization: Your Language Model is Secretly a Reward Model", "authors": ["Rafael Rafailov", "Archit Sharma", "Eric Mitchell", "Stefano Ermon", "Christopher D. Manning", "Chelsea Finn"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.18290", "doi": null, "topics": ["Algorithms", "Language models", "Machine learning"], "description": "While large-scale unsupervised language models (LMs) learn broad world knowledge and some reasoning skills, achieving precise control of their behavior is difficult due to the completely unsupervised nature of their training.", "description_method": "abstract-excerpt", "abstract": "While large-scale unsupervised language models (LMs) learn broad world knowledge and some reasoning skills, achieving precise control of their behavior is difficult due to the completely unsupervised nature of their training. Existing methods for gaining such steerability collect human labels of the relative quality of model generations and fine-tune the unsupervised LM to align with these preferences, often with reinforcement learning from human feedback (RLHF). However, RLHF is a complex and often unstable procedure, first fitting a reward model that reflects the human preferences, and then fine-tuning the large unsupervised LM using reinforcement learning to maximize this estimated reward without drifting too far from the original model. In this paper we introduce a new parameterization of the reward model in RLHF that enables extraction of the corresponding optimal policy in closed form, allowing us to solve the standard RLHF problem with only a simple classification loss. The resulting algorithm, which we call Direct Preference Optimization (DPO), is stable, performant, and computationally lightweight, eliminating the need for sampling from the LM during fine-tuning or performing significant hyperparameter tuning. Our experiments show that DPO can fine-tune LMs to align with human preferences as well as or better than existing methods. Notably, fine-tuning with DPO exceeds PPO-based RLHF in ability to control sentiment of generations, and matches or improves response quality in summarization and single-turn dialogue while being substantially simpler to implement and train.", "score": 1.0, "mention_count": 5, "user_count": 5, "months": ["2023-05", "2023-06", "2024-01", "2025-08", "2026-06"], "last_mentioned": "2026-06-07T07:26:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.16764", "slug": "e48cf1064a1d715b", "title": "ConvNets Match Vision Transformers at Scale", "authors": ["Samuel L. Smith", "Andrew Brock", "Leonard Berrada", "Soham De"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.16764", "doi": null, "topics": ["Language models", "Computer vision"], "description": "Many researchers believe that ConvNets perform well on small or moderately sized datasets, but are not competitive with Vision Transformers when given access to datasets on the web-scale.", "description_method": "abstract-excerpt", "abstract": "Many researchers believe that ConvNets perform well on small or moderately sized datasets, but are not competitive with Vision Transformers when given access to datasets on the web-scale. We challenge this belief by evaluating a performant ConvNet architecture pre-trained on JFT-4B, a large labelled dataset of images often used for training foundation models. We consider pre-training compute budgets between 0.4k and 110k TPU-v4 core compute hours, and train a series of networks of increasing depth and width from the NFNet model family. We observe a log-log scaling law between held out loss and compute budget. After fine-tuning on ImageNet, NFNets match the reported performance of Vision Transformers with comparable compute budgets. Our strongest fine-tuned model achieves a Top-1 accuracy of 90.4%.", "score": 1.0, "mention_count": 5, "user_count": 3, "months": ["2024-04", "2025-09", "2025-12"], "last_mentioned": "2025-12-23T03:53:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2312.11514", "slug": "bb4e1199c2489322", "title": "LLM in a flash: Efficient Large Language Model Inference with Limited Memory", "authors": ["Keivan Alizadeh", "Iman Mirzadeh", "Dmitry Belenko", "Karen Khatamifard", "Minsik Cho", "Carlo C Del Mundo", "Mohammad Rastegari", "Mehrdad Farajtabar"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2312.11514", "doi": null, "topics": ["Language models", "Neuroscience"], "description": "Large language models (LLMs) are central to modern natural language processing, delivering exceptional performance in various tasks.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLMs) are central to modern natural language processing, delivering exceptional performance in various tasks. However, their substantial computational and memory requirements present challenges, especially for devices with limited DRAM capacity. This paper tackles the challenge of efficiently running LLMs that exceed the available DRAM capacity by storing the model parameters in flash memory, but bringing them on demand to DRAM. Our method involves constructing an inference cost model that takes into account the characteristics of flash memory, guiding us to optimize in two critical areas: reducing the volume of data transferred from flash and reading data in larger, more contiguous chunks. Within this hardware-informed framework, we introduce two principal techniques. First, \"windowing\" strategically reduces data transfer by reusing previously activated neurons, and second, \"row-column bundling\", tailored to the sequential data access strengths of flash memory, increases the size of data chunks read from flash memory. These methods collectively enable running models up to twice the size of the available DRAM, with a 4-5x and 20-25x increase in inference speed compared to naive loading approaches in CPU and GPU, respectively. Our integration of sparsity awareness, context-adaptive loading, and a hardware-oriented design paves the way for effective inference of LLMs on devices with limited memory.", "score": 1.0, "mention_count": 5, "user_count": 5, "months": ["2023-12", "2024-05", "2026-03"], "last_mentioned": "2026-03-23T15:01:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2506.12286", "slug": "0a9db72cab5ddbe2", "title": "The SWE-Bench Illusion: When State-of-the-Art LLMs Remember Instead of Reason", "authors": ["Shanchao Liang", "Spandan Garg", "Roshanak Zilouchian Moghaddam"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2506.12286", "doi": null, "topics": ["Programming languages", "Language models"], "description": "As large language models (LLMs) become increasingly capable and widely adopted, benchmarks play a central role in assessing their practical utility.", "description_method": "abstract-excerpt", "abstract": "As large language models (LLMs) become increasingly capable and widely adopted, benchmarks play a central role in assessing their practical utility. For example, SWE-Bench Verified has emerged as a critical benchmark for evaluating LLMs' software engineering abilities, particularly their aptitude for resolving real-world GitHub issues. Recent LLMs show impressive performance on SWE-Bench, leading to optimism about their capacity for complex coding tasks. However, current evaluation protocols may overstate these models' true capabilities. It is crucial to distinguish LLMs' generalizable problem-solving ability and other learned artifacts. In this work, we introduce two diagnostic tasks: file path identification from issue descriptions alone and ground truth function reproduction with only the current file context and issue description to probe models' underlying knowledge. We present empirical evidence that performance gains on SWE-Bench-Verified may be partially driven by memorization rather than genuine problem-solving. We show that state-of-the-art models achieve up to 76% accuracy in identifying buggy file paths using only issue descriptions, without access to repository structure. This performance is merely up to 53% on tasks from repositories not included in SWE-Bench, pointing to possible data contamination or memorization. Similar patterns are also observed for the function reproduction task, where the verbatim similarity is much higher on SWE-Bench Verified than on other similar coding benchmarks (up to 35% consecutive 5-gram accuracy on SWE-Bench Verified and Full, but only up to 18% for tasks in other benchmarks). These findings raise concerns about the validity of existing results and underscore the need for more robust, contamination-resistant benchmarks to reliably evaluate LLMs' coding abilities.", "score": 1.0, "mention_count": 5, "user_count": 4, "months": ["2025-07", "2025-08", "2025-09", "2026-03", "2026-06"], "last_mentioned": "2026-06-06T11:01:33Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2507.19457", "slug": "c1ef72d96fa32925", "title": "GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning", "authors": ["Lakshya A Agrawal", "Shangyin Tan", "Dilara Soylu", "Noah Ziems", "Rishi Khare", "Krista Opsahl-Ong", "Arnav Singhvi", "Herumb Shandilya", "Michael J Ryan", "Meng Jiang", "Christopher Potts", "Koushik Sen", "Alexandros G. Dimakis", "Ion Stoica", "Dan Klein", "Matei Zaharia", "Omar Khattab"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2507.19457", "doi": null, "topics": ["Machine learning", "Language models", "Algorithms"], "description": "Large language models (LLMs) are increasingly adapted to downstream tasks via reinforcement learning (RL) methods like Group Relative Policy Optimization (GRPO), which often require thousands of rollouts to learn new tasks.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLMs) are increasingly adapted to downstream tasks via reinforcement learning (RL) methods like Group Relative Policy Optimization (GRPO), which often require thousands of rollouts to learn new tasks. We argue that the interpretable nature of language often provides a much richer learning medium for LLMs, compared to policy gradients derived from sparse, scalar rewards. To test this, we introduce GEPA (Genetic-Pareto), a prompt optimizer that thoroughly incorporates natural language reflection to learn high-level rules from trial and error. Given any AI system containing one or more LLM prompts, GEPA samples trajectories (e.g., reasoning, tool calls, and tool outputs) and reflects on them in natural language to diagnose problems, propose and test prompt updates, and combine complementary lessons from the Pareto frontier of its own attempts. As a result of GEPA's design, it can often turn even just a few rollouts into a large quality gain. Across six tasks, GEPA outperforms GRPO by 6% on average and by up to 20%, while using up to 35x fewer rollouts. GEPA also outperforms the leading prompt optimizer, MIPROv2, by over 10% (e.g., +12% accuracy on AIME-2025), and demonstrates promising results as an inference-time search strategy for code optimization. We release our code at https://github.com/gepa-ai/gepa .", "score": 1.0, "mention_count": 5, "user_count": 5, "months": ["2025-08", "2025-10", "2026-02"], "last_mentioned": "2026-02-22T05:37:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/800197.806036", "slug": "83dc53b050ef482c", "title": "Complex information processing", "authors": ["T. H. Nelson"], "year": 1965, "venue": "Proceedings of the 1965 20th national conference on   -", "url": "https://doi.org/10.1145/800197.806036", "doi": "10.1145/800197.806036", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 1.0, "mention_count": 5, "user_count": 3, "months": ["2021-10", "2023-08", "2025-10"], "last_mentioned": "2025-10-10T10:08:18Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1010.3003", "slug": "5a223fc1255f0416", "title": "Twitter mood predicts the stock market", "authors": ["Johan Bollen", "Huina Mao", "Xiao-Jun Zeng"], "year": 2010, "venue": "arXiv", "url": "https://arxiv.org/abs/1010.3003", "doi": "10.1016/j.jocs.2010.12.007", "topics": ["Psychology", "Economics", "Statistics & probability"], "description": "Behavioral economics tells us that emotions can profoundly affect individual behavior and decision-making.", "description_method": "abstract-excerpt", "abstract": "Behavioral economics tells us that emotions can profoundly affect individual behavior and decision-making. Does this also apply to societies at large, i.e., can societies experience mood states that affect their collective decision making? By extension is the public mood correlated or even predictive of economic indicators? Here we investigate whether measurements of collective mood states derived from large-scale Twitter feeds are correlated to the value of the Dow Jones Industrial Average (DJIA) over time. We analyze the text content of daily Twitter feeds by two mood tracking tools, namely OpinionFinder that measures positive vs. negative mood and Google-Profile of Mood States (GPOMS) that measures mood in terms of 6 dimensions (Calm, Alert, Sure, Vital, Kind, and Happy). We cross-validate the resulting mood time series by comparing their ability to detect the public's response to the presidential election and Thanksgiving day in 2008. A Granger causality analysis and a Self-Organizing Fuzzy Neural Network are then used to investigate the hypothesis that public mood states, as measured by the OpinionFinder and GPOMS mood time series, are predictive of changes in DJIA closing values. Our results indicate that the accuracy of DJIA predictions can be significantly improved by the inclusion of specific public mood dimensions but not others. We find an accuracy of 87.6% in predicting the daily up and down changes in the closing values of the DJIA and a reduction of the Mean Average Percentage Error by more than 6%.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2010-10", "2010-11", "2011-07", "2014-03"], "last_mentioned": "2014-03-20T02:40:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1202.5708", "slug": "f608e2899e823817", "title": "The Alcubierre Warp Drive: On the Matter of Matter", "authors": ["Brendan McMonigal", "Geraint F. Lewis", "Philip O'Byrne"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1202.5708", "doi": "10.1103/PhysRevD.85.064024", "topics": ["Physics"], "description": "The Alcubierre warp drive allows a spaceship to travel at an arbitrarily large global velocity by deforming the spacetime in a bubble around the spaceship.", "description_method": "abstract-excerpt", "abstract": "The Alcubierre warp drive allows a spaceship to travel at an arbitrarily large global velocity by deforming the spacetime in a bubble around the spaceship. Little is known about the interactions between massive particles and the Alcubierre warp drive, or the effects of an accelerating or decelerating warp bubble. We examine geodesics representative of the paths of null and massive particles with a range of initial velocities from -c to c interacting with an Alcubierre warp bubble travelling at a range of globally subluminal and superluminal velocities on both constant and variable velocity paths. The key results for null particles match what would be expected of massive test particles as they approach +/- c. The increase in energy for massive and null particles is calculated in terms of v_s, the global ship velocity, and v_p, the initial velocity of the particle with respect to the rest frame of the origin/destination of the ship. Particles with positive v_p obtain extremely high energy and velocity and become \"time locked\" for the duration of their time in the bubble, experiencing very little proper time between entering and eventually leaving the bubble. When interacting with an accelerating bubble, any particles within the bubble at the time receive a velocity boost that increases or decreases the magnitude of their velocity if the particle is moving towards the front or rear of the bubble respectively. If the bubble is decelerating, the opposite effect is observed. Thus Eulerian matter is unaffected by bubble accelerations/decelerations. The magnitude of the velocity boosts scales with the magnitude of the bubble acceleration/deceleration.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2012-11", "2013-07", "2013-11", "2022-01"], "last_mentioned": "2022-01-16T15:46:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1304.3381", "slug": "15d3a3ad2edaeff8", "title": "Life Before Earth", "authors": ["Alexei A. Sharov", "Richard Gordon"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1304.3381", "doi": null, "topics": ["Algorithms", "Statistics & probability", "Neuroscience"], "description": "An extrapolation of the genetic complexity of organisms to earlier times suggests that life began before the Earth was formed.", "description_method": "abstract-excerpt", "abstract": "An extrapolation of the genetic complexity of organisms to earlier times suggests that life began before the Earth was formed. Life may have started from systems with single heritable elements that are functionally equivalent to a nucleotide. The genetic complexity, roughly measured by the number of non-redundant functional nucleotides, is expected to have grown exponentially due to several positive feedback factors: gene cooperation, duplication of genes with their subsequent specialization, and emergence of novel functional niches associated with existing genes. Linear regression of genetic complexity on a log scale extrapolated back to just one base pair suggests the time of the origin of life 9.7 billion years ago. This cosmic time scale for the evolution of life has important consequences: life took ca. 5 billion years to reach the complexity of bacteria; the environments in which life originated and evolved to the prokaryote stage may have been quite different from those envisaged on Earth; there was no intelligent life in our universe prior to the origin of Earth, thus Earth could not have been deliberately seeded with life by intelligent aliens; Earth was seeded by panspermia; experimental replication of the origin of life from scratch may have to emulate many cumulative rare events; and the Drake equation for guesstimating the number of civilizations in the universe is likely wrong, as intelligent life has just begun appearing in our universe. Evolution of advanced organisms has accelerated via development of additional information-processing systems: epigenetic memory, primitive mind, multicellular brain, language, books, computers, and Internet. As a result the doubling time of complexity has reached ca. 20 years. Finally, we discuss the issue of the predicted technological singularity and give a biosemiotics perspective on the increase of complexity.", "score": 1.0, "mention_count": 4, "user_count": 3, "months": ["2018-01", "2020-08", "2020-12", "2021-04"], "last_mentioned": "2021-04-05T01:38:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1506.01497", "slug": "d5f58ec853c4e749", "title": "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks", "authors": ["Shaoqing Ren", "Kaiming He", "Ross Girshick", "Jian Sun"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1506.01497", "doi": null, "topics": ["Computer vision", "Machine learning", "Algorithms"], "description": "State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations.", "description_method": "abstract-excerpt", "abstract": "State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations. Advances like SPPnet and Fast R-CNN have reduced the running time of these detection networks, exposing region proposal computation as a bottleneck. In this work, we introduce a Region Proposal Network (RPN) that shares full-image convolutional features with the detection network, thus enabling nearly cost-free region proposals. An RPN is a fully convolutional network that simultaneously predicts object bounds and objectness scores at each position. The RPN is trained end-to-end to generate high-quality region proposals, which are used by Fast R-CNN for detection. We further merge RPN and Fast R-CNN into a single network by sharing their convolutional features---using the recently popular terminology of neural networks with 'attention' mechanisms, the RPN component tells the unified network where to look. For the very deep VGG-16 model, our detection system has a frame rate of 5fps (including all steps) on a GPU, while achieving state-of-the-art object detection accuracy on PASCAL VOC 2007, 2012, and MS COCO datasets with only 300 proposals per image. In ILSVRC and COCO 2015 competitions, Faster R-CNN and RPN are the foundations of the 1st-place winning entries in several tracks. Code has been made publicly available.", "score": 1.0, "mention_count": 4, "user_count": 3, "months": ["2017-08", "2017-10", "2017-11"], "last_mentioned": "2017-11-12T16:53:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1607.01590", "slug": "d965662b778cb385", "title": "Indexing dif/2", "authors": ["Ulrich Neumerkel", "Stefan Kral"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1607.01590", "doi": null, "topics": ["Economics", "Algorithms"], "description": "Many Prolog programs are unnecessarily impure because of inadequate means to express syntactic inequality.", "description_method": "abstract-excerpt", "abstract": "Many Prolog programs are unnecessarily impure because of inadequate means to express syntactic inequality. While the frequently provided built-in `dif/2` is able to correctly describe expected answers, its direct use in programs often leads to overly complex and inefficient definitions --- mainly due to the lack of adequate indexing mechanisms. We propose to overcome these problems by using a new predicate that subsumes both equality and inequality via reification. Code complexity is reduced with a monotonic, higher-order if-then-else construct based on `call/N`. For comparable correct uses of impure definitions, our approach is as determinate and similarly efficient as its impure counterparts.", "score": 1.0, "mention_count": 4, "user_count": 3, "months": ["2017-05", "2019-10", "2022-09", "2022-12"], "last_mentioned": "2022-12-31T10:22:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.04886", "slug": "0c1654845eec2edd", "title": "Are Delayed Issues Harder to Resolve? Revisiting Cost-to-Fix of Defects throughout the Lifecycle", "authors": ["Tim Menzies", "William Nichols", "Forrest Shull", "Lucas Layman"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.04886", "doi": "10.1007/s10664-016-9469-x", "topics": ["Programming languages"], "description": "Many practitioners and academics believe in a delayed issue effect (DIE); i.e.", "description_method": "abstract-excerpt", "abstract": "Many practitioners and academics believe in a delayed issue effect (DIE); i.e. the longer an issue lingers in the system, the more effort it requires to resolve. This belief is often used to justify major investments in new development processes that promise to retire more issues sooner. This paper tests for the delayed issue effect in 171 software projects conducted around the world in the period from 2006--2014. To the best of our knowledge, this is the largest study yet published on this effect. We found no evidence for the delayed issue effect; i.e. the effort to resolve issues in a later phase was not consistently or substantially greater than when issues were resolved soon after their introduction. This paper documents the above study and explores reasons for this mismatch between this common rule of thumb and empirical data. In summary, DIE is not some constant across all projects. Rather, DIE might be an historical relic that occurs intermittently only in certain kinds of projects. This is a significant result since it predicts that new development processes that promise to faster retire more issues will not have a guaranteed return on investment (depending on the context where applied), and that a long-held truth in software engineering should not be considered a global truism.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2017-09", "2021-07", "2025-02"], "last_mentioned": "2025-02-25T22:03:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1612.09375", "slug": "f492938bf92c58ac", "title": "Basic Category Theory", "authors": ["Tom Leinster"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1612.09375", "doi": null, "topics": ["Mathematics"], "description": "This short introductory category theory textbook is for readers with relatively little mathematical background (e.g.", "description_method": "abstract-excerpt", "abstract": "This short introductory category theory textbook is for readers with relatively little mathematical background (e.g. the first half of an undergraduate mathematics degree). At its heart is the concept of a universal property, important throughout mathematics. After a chapter introducing the basic definitions, separate chapters present three ways of expressing universal properties: via adjoint functors, representable functors, and limits. A final chapter ties the three together. For each new categorical concept, a generous supply of examples is provided, taken from different parts of mathematics. At points where the leap in abstraction is particularly great (such as the Yoneda lemma), the reader will find careful and extensive explanations.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2019-08", "2020-08", "2022-10", "2026-04"], "last_mentioned": "2026-04-18T11:10:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1707.01195", "slug": "4c0cfa78f5401003", "title": "The impossibility of \"fairness\": a generalized impossibility result for decisions", "authors": ["Thomas Miconi"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1707.01195", "doi": null, "topics": ["Statistics & probability", "Psychology", "Algorithms"], "description": "Various measures can be used to estimate bias or unfairness in a predictor.", "description_method": "abstract-excerpt", "abstract": "Various measures can be used to estimate bias or unfairness in a predictor. Previous work has already established that some of these measures are incompatible with each other. Here we show that, when groups differ in prevalence of the predicted event, several intuitive, reasonable measures of fairness (probability of positive prediction given occurrence or non-occurrence; probability of occurrence given prediction or non-prediction; and ratio of predictions over occurrences for each group) are all mutually exclusive: if one of them is equal among groups, the other two must differ. The only exceptions are for perfect, or trivial (always-positive or always-negative) predictors. As a consequence, any non-perfect, non-trivial predictor must necessarily be \"unfair\" under two out of three reasonable sets of criteria. This result readily generalizes to a wide range of well-known statistical quantities (sensitivity, specificity, false positive rate, precision, etc.), all of which can be divided into three mutually exclusive groups. Importantly, The results applies to all predictors, whether algorithmic or human. We conclude with possible ways to handle this effect when assessing and designing prediction methods.", "score": 1.0, "mention_count": 4, "user_count": 3, "months": ["2018-06", "2020-02", "2022-02", "2023-03"], "last_mentioned": "2023-03-15T20:41:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1801.09553", "slug": "7ac9b6f5fbfa25d6", "title": "Extending the Algebraic Manipulability of Differentials", "authors": ["Jonathan Bartlett", "Asatur Zh. Khurshudyan"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1801.09553", "doi": null, "topics": ["Mathematics"], "description": "Treating differentials as independent algebraic units have a long history of use and abuse.", "description_method": "abstract-excerpt", "abstract": "Treating differentials as independent algebraic units have a long history of use and abuse. It is generally considered problematic to treat the derivative as a fraction of differentials rather than as a holistic unit acting as a limit, though for practical reasons it is often done for the first derivative. However, using a revised notation for the second and higher derivatives will allow for the ability to treat differentials as independent units for a much larger number of cases.", "score": 1.0, "mention_count": 4, "user_count": 3, "months": ["2022-05", "2022-07", "2024-01"], "last_mentioned": "2024-01-20T11:45:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1904.12320", "slug": "21ae4f4aaceca370", "title": "Real numbers, data science and chaos: How to fit any dataset with a single parameter", "authors": ["Laurent Boué"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1904.12320", "doi": null, "topics": ["Machine learning"], "description": "We show how any dataset of any modality (time-series, images, sound...) can be approximated by a well-behaved (continuous, differentiable...) scalar function with a single real-valued parameter.", "description_method": "abstract-excerpt", "abstract": "We show how any dataset of any modality (time-series, images, sound...) can be approximated by a well-behaved (continuous, differentiable...) scalar function with a single real-valued parameter. Building upon elementary concepts from chaos theory, we adopt a pedagogical approach demonstrating how to adjust this parameter in order to achieve arbitrary precision fit to all samples of the data. Targeting an audience of data scientists with a taste for the curious and unusual, the results presented here expand on previous similar observations regarding expressiveness power and generalization of machine learning models.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2019-05", "2021-10", "2022-01", "2023-04"], "last_mentioned": "2023-04-26T00:38:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1905.12133", "slug": "eac667af69da30ea", "title": "One SQL to Rule Them All", "authors": ["Edmon Begoli", "Tyler Akidau", "Fabian Hueske", "Julian Hyde", "Kathryn Knight", "Kenneth Knowles"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1905.12133", "doi": "10.1145/3299869.3314040", "topics": ["Databases"], "description": "Real-time data analysis and management are increasingly critical for today`s businesses.", "description_method": "abstract-excerpt", "abstract": "Real-time data analysis and management are increasingly critical for today`s businesses. SQL is the de facto lingua franca for these endeavors, yet support for robust streaming analysis and management with SQL remains limited. Many approaches restrict semantics to a reduced subset of features and/or require a suite of non-standard constructs. Additionally, use of event timestamps to provide native support for analyzing events according to when they actually occurred is not pervasive, and often comes with important limitations. We present a three-part proposal for integrating robust streaming into the SQL standard, namely: (1) time-varying relations as a foundation for classical tables as well as streaming data, (2) event time semantics, (3) a limited set of optional keyword extensions to control the materialization of time-varying query results. Motivated and illustrated using examples and lessons learned from implementations in Apache Calcite, Apache Flink, and Apache Beam, we show how with these minimal additions it is possible to utilize the complete suite of standard SQL semantics to perform robust stream processing.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2020-08", "2021-01", "2021-04", "2021-06"], "last_mentioned": "2021-06-07T15:58:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1906.00091", "slug": "0eeb390bc69196b0", "title": "Deep Learning Recommendation Model for Personalization and Recommendation Systems", "authors": ["Maxim Naumov", "Dheevatsa Mudigere", "Hao-Jun Michael Shi", "Jianyu Huang", "Narayanan Sundaraman", "Jongsoo Park", "Xiaodong Wang", "Udit Gupta", "Carole-Jean Wu", "Alisson G. Azzolini", "Dmytro Dzhulgakov", "Andrey Mallevich", "Ilia Cherniavskii", "Yinghai Lu", "Raghuraman Krishnamoorthi", "Ansha Yu", "Volodymyr Kondratenko", "Stephanie Pereira", "Xianjie Chen", "Wenlin Chen", "Vijay Rao", "Bill Jia", "Liang Xiong", "Misha Smelyanskiy"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1906.00091", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "With the advent of deep learning, neural network-based recommendation models have emerged as an important tool for tackling personalization and recommendation tasks.", "description_method": "abstract-excerpt", "abstract": "With the advent of deep learning, neural network-based recommendation models have emerged as an important tool for tackling personalization and recommendation tasks. These networks differ significantly from other deep learning networks due to their need to handle categorical features and are not well studied or understood. In this paper, we develop a state-of-the-art deep learning recommendation model (DLRM) and provide its implementation in both PyTorch and Caffe2 frameworks. In addition, we design a specialized parallelization scheme utilizing model parallelism on the embedding tables to mitigate memory constraints while exploiting data parallelism to scale-out compute from the fully-connected layers. We compare DLRM against existing recommendation models and characterize its performance on the Big Basin AI platform, demonstrating its usefulness as a benchmark for future algorithmic experimentation and system co-design.", "score": 1.0, "mention_count": 4, "user_count": 3, "months": ["2020-03", "2024-03", "2025-03"], "last_mentioned": "2025-03-07T19:16:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1907.07355", "slug": "42a34a3b5af10d2f", "title": "Probing Neural Network Comprehension of Natural Language Arguments", "authors": ["Timothy Niven", "Hung-Yu Kao"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1907.07355", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Language models"], "description": "We are surprised to find that BERT's peak performance of 77% on the Argument Reasoning Comprehension Task reaches just three points below the average untrained human baseline.", "description_method": "abstract-excerpt", "abstract": "We are surprised to find that BERT's peak performance of 77% on the Argument Reasoning Comprehension Task reaches just three points below the average untrained human baseline. However, we show that this result is entirely accounted for by exploitation of spurious statistical cues in the dataset. We analyze the nature of these cues and demonstrate that a range of models all exploit them. This analysis informs the construction of an adversarial dataset on which all models achieve random accuracy. Our adversarial dataset provides a more robust assessment of argument comprehension and should be adopted as the standard in future work.", "score": 1.0, "mention_count": 4, "user_count": 3, "months": ["2019-07", "2019-08", "2019-11", "2023-12"], "last_mentioned": "2023-12-26T15:48:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2009.05045", "slug": "f8039d59eb968fb8", "title": "Forecasting timelines of quantum computing", "authors": ["Jaime Sevilla", "C. Jess Riedel"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2009.05045", "doi": null, "topics": ["Physics", "Statistics & probability", "Distributed systems"], "description": "We consider how to forecast progress in the domain of quantum computing.", "description_method": "abstract-excerpt", "abstract": "We consider how to forecast progress in the domain of quantum computing. For this purpose we collect a dataset of quantum computer systems to date, scored on their physical qubits and gate error rate, and we define an index combining both metrics, the generalized logical qubit. We study the relationship between physical qubits and gate error rate, and tentatively conclude that they are positively correlated (albeit with some room for doubt), indicating a frontier of development that trades-off between them. We also apply a log-linear regression on the metrics to provide a tentative upper bound on how much progress can be expected over time. Within the (generally optimistic) assumptions of our model, including the key assumption that exponential progress in qubit count and gate fidelity will continue, we estimate that that proof-of-concept fault-tolerant computation based on superconductor technology is unlikely (<5% confidence) to be exhibited before 2026, and that quantum devices capable of factoring RSA-2048 are unlikely (<5% confidence) to exist before 2039. It is of course possible that these milestones will in fact be reached earlier, but that this would require faster progress than has yet been seen.", "score": 1.0, "mention_count": 4, "user_count": 3, "months": ["2022-07", "2023-07", "2024-05", "2024-12"], "last_mentioned": "2024-12-10T00:18:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2205.10343", "slug": "6b0341e715676608", "title": "Towards Understanding Grokking: An Effective Theory of Representation Learning", "authors": ["Ziming Liu", "Ouail Kitouni", "Niklas Nolte", "Eric J. Michaud", "Max Tegmark", "Mike Williams"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2205.10343", "doi": null, "topics": ["Physics", "Machine learning", "Psychology"], "description": "We aim to understand grokking, a phenomenon where models generalize long after overfitting their training set.", "description_method": "abstract-excerpt", "abstract": "We aim to understand grokking, a phenomenon where models generalize long after overfitting their training set. We present both a microscopic analysis anchored by an effective theory and a macroscopic analysis of phase diagrams describing learning performance across hyperparameters. We find that generalization originates from structured representations whose training dynamics and dependence on training set size can be predicted by our effective theory in a toy setting. We observe empirically the presence of four learning phases: comprehension, grokking, memorization, and confusion. We find representation learning to occur only in a \"Goldilocks zone\" (including comprehension and grokking) between memorization and confusion. We find on transformers the grokking phase stays closer to the memorization phase (compared to the comprehension phase), leading to delayed generalization. The Goldilocks phase is reminiscent of \"intelligence from starvation\" in Darwinian evolution, where resource limitations drive discovery of more efficient solutions. This study not only provides intuitive explanations of the origin of grokking, but also highlights the usefulness of physics-inspired tools, e.g., effective theories and phase diagrams, for understanding deep learning.", "score": 1.0, "mention_count": 4, "user_count": 3, "months": ["2022-12", "2023-01", "2023-02", "2024-06"], "last_mentioned": "2024-06-04T00:49:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.13534", "slug": "cf757e24b7819d27", "title": "How Language Model Hallucinations Can Snowball", "authors": ["Muru Zhang", "Ofir Press", "William Merrill", "Alisa Liu", "Noah A. Smith"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.13534", "doi": null, "topics": ["Language models"], "description": "A major risk of using language models in practical applications is their tendency to hallucinate incorrect statements.", "description_method": "abstract-excerpt", "abstract": "A major risk of using language models in practical applications is their tendency to hallucinate incorrect statements. Hallucinations are often attributed to knowledge gaps in LMs, but we hypothesize that in some cases, when justifying previously generated hallucinations, LMs output false claims that they can separately recognize as incorrect. We construct three question-answering datasets where ChatGPT and GPT-4 often state an incorrect answer and offer an explanation with at least one incorrect claim. Crucially, we find that ChatGPT and GPT-4 can identify 67% and 87% of their own mistakes, respectively. We refer to this phenomenon as hallucination snowballing: an LM over-commits to early mistakes, leading to more mistakes that it otherwise would not make.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2023-05", "2024-02", "2025-09", "2025-12"], "last_mentioned": "2025-12-28T16:30:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.02207", "slug": "8edd8229c15120e1", "title": "Language Models Represent Space and Time", "authors": ["Wes Gurnee", "Max Tegmark"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.02207", "doi": null, "topics": ["Language models", "Neuroscience", "Statistics & probability"], "description": "The capabilities of large language models (LLMs) have sparked debate over whether such systems just learn an enormous collection of superficial statistics or a set of more coherent and grounded representations that reflect the real world.", "description_method": "abstract-excerpt", "abstract": "The capabilities of large language models (LLMs) have sparked debate over whether such systems just learn an enormous collection of superficial statistics or a set of more coherent and grounded representations that reflect the real world. We find evidence for the latter by analyzing the learned representations of three spatial datasets (world, US, NYC places) and three temporal datasets (historical figures, artworks, news headlines) in the Llama-2 family of models. We discover that LLMs learn linear representations of space and time across multiple scales. These representations are robust to prompting variations and unified across different entity types (e.g. cities and landmarks). In addition, we identify individual \"space neurons\" and \"time neurons\" that reliably encode spatial and temporal coordinates. While further investigation is needed, our results suggest modern LLMs learn rich spatiotemporal representations of the real world and possess basic ingredients of a world model.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2023-11", "2024-07", "2024-10"], "last_mentioned": "2024-10-31T15:49:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2405.21060", "slug": "da8316994dc364f9", "title": "Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality", "authors": ["Tri Dao", "Albert Gu"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2405.21060", "doi": null, "topics": ["Language models", "Algorithms", "Machine learning"], "description": "While Transformers have been the main architecture behind deep learning's success in language modeling, state-space models (SSMs) such as Mamba have recently been shown to match or outperform Transformers at small to medium scale.", "description_method": "abstract-excerpt", "abstract": "While Transformers have been the main architecture behind deep learning's success in language modeling, state-space models (SSMs) such as Mamba have recently been shown to match or outperform Transformers at small to medium scale. We show that these families of models are actually quite closely related, and develop a rich framework of theoretical connections between SSMs and variants of attention, connected through various decompositions of a well-studied class of structured semiseparable matrices. Our state space duality (SSD) framework allows us to design a new architecture (Mamba-2) whose core layer is an a refinement of Mamba's selective SSM that is 2-8X faster, while continuing to be competitive with Transformers on language modeling.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2024-10", "2025-06", "2026-02"], "last_mentioned": "2026-02-04T00:28:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2512.14982", "slug": "88637636e1ad4b80", "title": "Prompt Repetition Improves Non-Reasoning LLMs", "authors": ["Yaniv Leviathan", "Matan Kalman", "Yossi Matias"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2512.14982", "doi": null, "topics": ["Language models"], "description": "When not using reasoning, repeating the input prompt improves performance for popular models (Gemini, GPT, Claude, and Deepseek) without increasing the number of generated tokens or latency.", "description_method": "abstract-excerpt", "abstract": "When not using reasoning, repeating the input prompt improves performance for popular models (Gemini, GPT, Claude, and Deepseek) without increasing the number of generated tokens or latency.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2026-02", "2026-03", "2026-07"], "last_mentioned": "2026-07-06T23:06:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:hep-th/0310077", "slug": "0f344cbe28b48692", "title": "A dialog on quantum gravity", "authors": ["Carlo Rovelli"], "year": 2003, "venue": "arXiv", "url": "https://arxiv.org/abs/hep-th/0310077", "doi": "10.1142/S0218271803004304", "topics": ["Physics"], "description": "The debate between loop quantum gravity and string theory is sometimes lively, and it is hard to present an impartial view on the issue.", "description_method": "abstract-excerpt", "abstract": "The debate between loop quantum gravity and string theory is sometimes lively, and it is hard to present an impartial view on the issue. Leaving any attempt to impartiality aside, I report here, instead, a conversation on this issue, overheard in the cafeteria of a Major American University.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2012-10", "2016-01", "2017-07", "2022-05"], "last_mentioned": "2022-05-12T07:03:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-017-17373-3", "slug": "adf98260796b6c10", "title": "Effect of canola oil consumption on memory, synapse and neuropathology in the triple transgenic mouse model of Alzheimer’s disease", "authors": ["Elisabetta Lauretti", "Domenico Praticò"], "year": 2017, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-017-17373-3", "doi": "10.1038/s41598-017-17373-3", "topics": ["Neuroscience", "Biology & medicine", "Psychology"], "description": "AbstractIn recent years consumption of canola oil has increased due to lower cost compared with olive oil and the perception that it shares its health benefits.", "description_method": "abstract-excerpt", "abstract": "AbstractIn recent years consumption of canola oil has increased due to lower cost compared with olive oil and the perception that it shares its health benefits. However, no data are available on the effect of canola oil intake on Alzheimer’s disease (AD) pathogenesis. Herein, we investigated the effect of chronic daily consumption of canola oil on the phenotype of a mouse model of AD that develops both plaques and tangles (3xTg). To this end mice received either regular chow or a chow diet supplemented with canola oil for 6 months. At this time point we found that chronic exposure to the canola-rich diet resulted in a significant increase in body weight and impairments in their working memory together with decrease levels of post-synaptic density protein-95, a marker of synaptic integrity, and an increase in the ratio of insoluble Aβ 42/40. No significant changes were observed in tau phosphorylation and neuroinflammation. Taken together, our findings do not support a beneficial effect of chronic canola oil consumption on two important aspects of AD pathophysiology which includes memory impairments as well as synaptic integrity. While more studies are needed, our data do not justify the current trend aimed at replacing olive oil with canola oil.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2018-08", "2019-07", "2021-11"], "last_mentioned": "2021-11-18T20:28:47Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-018-22271-3", "slug": "dda38f36ce6784ca", "title": "Exposure of Insects to Radio-Frequency Electromagnetic Fields from 2 to 120 GHz", "authors": ["Arno Thielens", "Duncan Bell", "David B. Mortimore", "Mark K. Greco", "Luc Martens", "Wout Joseph"], "year": 2018, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-018-22271-3", "doi": "10.1038/s41598-018-22271-3", "topics": ["Other research"], "description": "Abstract Insects are continually exposed to Radio-Frequency (RF) electromagnetic fields at different frequencies.", "description_method": "abstract-excerpt", "abstract": "Abstract Insects are continually exposed to Radio-Frequency (RF) electromagnetic fields at different frequencies. The range of frequencies used for wireless telecommunication systems will increase in the near future from below 6 GHz (2 G, 3 G, 4 G, and WiFi) to frequencies up to 120 GHz (5 G). This paper is the first to report the absorbed RF electromagnetic power in four different types of insects as a function of frequency from 2 GHz to 120 GHz. A set of insect models was obtained using novel Micro-CT (computer tomography) imaging. These models were used for the first time in finite-difference time-domain electromagnetic simulations. All insects showed a dependence of the absorbed power on the frequency. All insects showed a general increase in absorbed RF power at and above 6 GHz, in comparison to the absorbed RF power below 6 GHz. Our simulations showed that a shift of 10% of the incident power density to frequencies above 6 GHz would lead to an increase in absorbed power between 3–370%.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2019-04", "2019-10", "2020-06", "2021-12"], "last_mentioned": "2021-12-27T06:33:36Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.adk3705", "slug": "df9128e27dc6df88", "title": "A 485-million-year history of Earth’s surface temperature", "authors": ["Emily J. Judd", "Jessica E. Tierney", "Daniel J. Lunt", "Isabel P. Montañez", "Brian T. Huber", "Scott L. Wing", "Paul J. Valdes"], "year": 2024, "venue": "Science", "url": "https://doi.org/10.1126/science.adk3705", "doi": "10.1126/science.adk3705", "topics": ["Statistics & probability", "Machine learning"], "description": "A long-term record of global mean surface temperature (GMST) provides critical insight into the dynamical limits of Earth’s climate and the complex feedbacks between temperature and the broader Earth system.", "description_method": "abstract-excerpt", "abstract": "A long-term record of global mean surface temperature (GMST) provides critical insight into the dynamical limits of Earth’s climate and the complex feedbacks between temperature and the broader Earth system. Here, we present PhanDA, a reconstruction of GMST over the past 485 million years, generated by statistically integrating proxy data with climate model simulations. PhanDA exhibits a large range of GMST, spanning 11° to 36°C. Partitioning the reconstruction into climate states indicates that more time was spent in warmer rather than colder climates and reveals consistent latitudinal temperature gradients within each state. There is a strong correlation between atmospheric carbon dioxide (CO 2 ) concentrations and GMST, identifying CO 2 as the dominant control on variations in Phanerozoic global climate and suggesting an apparent Earth system sensitivity of ~8°C.", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2024-09", "2024-10", "2025-01", "2025-03"], "last_mentioned": "2025-03-05T17:02:34Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/2983990.2984004", "slug": "dbce0549cecba22c", "title": "Java and scala's type systems are unsound: the existential crisis of null pointers", "authors": ["Nada Amin", "Ross Tate"], "year": 2016, "venue": "Proceedings of the 2016 ACM SIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications", "url": "https://doi.org/10.1145/2983990.2984004", "doi": "10.1145/2983990.2984004", "topics": ["Programming languages"], "description": "", "description_method": null, "abstract": "", "score": 1.0, "mention_count": 4, "user_count": 4, "months": ["2020-01", "2020-08", "2021-12", "2022-01"], "last_mentioned": "2022-01-23T14:48:29Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3386329", "slug": "b2afbdf34776b22e", "title": "History of Logo", "authors": ["Cynthia Solomon", "Brian Harvey", "Ken Kahn", "Henry Lieberman", "Mark L. Miller", "Margaret Minsky", "Artemis Papert", "Brian Silverman"], "year": 2020, "venue": "Proceedings of the ACM on Programming Languages", "url": "https://doi.org/10.1145/3386329", "doi": "10.1145/3386329", "topics": ["Mathematics", "Programming languages", "Robotics"], "description": "Logo is more than a programming language.", "description_method": "abstract-excerpt", "abstract": "Logo is more than a programming language. It is a learning environment where children explore mathematical ideas and create projects of their own design. Logo, the first computer language explicitly designed for children, was invented by Seymour Papert, Wallace Feurzeig, Daniel Bobrow, and Cynthia Solomon in 1966 at Bolt, Beranek and Newman, Inc. (BBN). Logo’s design drew upon two theoretical frameworks: Jean Piaget’s constructivism and Marvin Minsky’s artificial intelligence research at MIT. One of Logo’s foundational ideas was that children should have a powerful programming environment. Early Lisp served as a model with its symbolic computation, recursive functions, operations on linked lists, and dynamic scoping of variables. Logo became a symbol for change in elementary mathematics education and in the nature of school itself. The search for harnessing the computer’s potential to provide new ways of teaching and learning became a central focus and guiding principle in the Logo language development as it encompassed a widening scope that included natural language, music, graphics, animation, story telling, turtle geometry, robots, and other physical devices.", "score": 1.0, "mention_count": 4, "user_count": 3, "months": ["2021-06", "2021-07", "2023-03", "2023-07"], "last_mentioned": "2023-07-17T09:47:39Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1011.0423", "slug": "78dc0b4607394e0e", "title": "A note comprising a negative resolution of the Efficient Market Hypothesis", "authors": ["Robert Viragh"], "year": 2010, "venue": "arXiv", "url": "https://arxiv.org/abs/1011.0423", "doi": null, "topics": ["Statistics & probability", "Economics"], "description": "This note comprises a negative resolution of the Efficient Market Hypothesis.", "description_method": "abstract-excerpt", "abstract": "This note comprises a negative resolution of the Efficient Market Hypothesis.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2018-06", "2019-04", "2020-05"], "last_mentioned": "2020-05-31T19:04:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1206.1901", "slug": "3973d4dfb06185b4", "title": "MCMC using Hamiltonian dynamics", "authors": ["Radford M. Neal"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1206.1901", "doi": "10.1201/b10905", "topics": ["Statistics & probability", "Psychology", "Physics"], "description": "Hamiltonian dynamics can be used to produce distant proposals for the Metropolis algorithm, thereby avoiding the slow exploration of the state space that results from the diffusive behaviour of simple random-walk proposals.", "description_method": "abstract-excerpt", "abstract": "Hamiltonian dynamics can be used to produce distant proposals for the Metropolis algorithm, thereby avoiding the slow exploration of the state space that results from the diffusive behaviour of simple random-walk proposals. Though originating in physics, Hamiltonian dynamics can be applied to most problems with continuous state spaces by simply introducing fictitious \"momentum\" variables. A key to its usefulness is that Hamiltonian dynamics preserves volume, and its trajectories can thus be used to define complex mappings without the need to account for a hard-to-compute Jacobian factor - a property that can be exactly maintained even when the dynamics is approximated by discretizing time. In this review, I discuss theoretical and practical aspects of Hamiltonian Monte Carlo, and present some of its variations, including using windows of states for deciding on acceptance or rejection, computing trajectories using fast approximations, tempering during the course of a trajectory to handle isolated modes, and short-cut methods that prevent useless trajectories from taking much computation time.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2017-02", "2019-04", "2021-10"], "last_mentioned": "2021-10-22T22:11:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1305.3913", "slug": "2b07c84149d5ee31", "title": "Indication of anomalous heat energy production in a reactor device", "authors": ["Giuseppe Levi", "Evelyn Foschi", "Torbjörn Hartman", "Bo Höistad", "Roland Pettersson", "Lars Tegnér", "Hanno Essén"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1305.3913", "doi": null, "topics": ["Other research"], "description": "An experimental investigation of possible anomalous heat production in a special type of reactor tube named E-Cat HT is carried out.", "description_method": "abstract-excerpt", "abstract": "An experimental investigation of possible anomalous heat production in a special type of reactor tube named E-Cat HT is carried out. The reactor tube is charged with a small amount of hydrogen loaded nickel powder plus some additives. The reaction is primarily initiated by heat from resistor coils inside the reactor tube. Measurement of the produced heat was performed with high-resolution thermal imaging cameras, recording data every second from the hot reactor tube. The measurements of electrical power input were performed with a large bandwidth three-phase power analyzer. Data were collected in two experimental runs lasting 96 and 116 hours, respectively. An anomalous heat production was indicated in both experiments. The 116-hour experiment also included a calibration of the experimental set-up without the active charge present in the E-Cat HT. In this case, no extra heat was generated beyond the expected heat from the electric input. Computed volumetric and gravimetric energy densities were found to be far above those of any known chemical source. Even by the most conservative assumptions as to the errors in the measurements, the result is still one order of magnitude greater than conventional energy sources.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2013-05", "2013-11", "2016-02"], "last_mentioned": "2016-02-25T18:11:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1607.00133", "slug": "ca532d8d260ed831", "title": "Deep Learning with Differential Privacy", "authors": ["Martín Abadi", "Andy Chu", "Ian Goodfellow", "H. Brendan McMahan", "Ilya Mironov", "Kunal Talwar", "Li Zhang"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1607.00133", "doi": "10.1145/2976749.2978318", "topics": ["Security & cryptography", "Machine learning", "Algorithms"], "description": "Machine learning techniques based on neural networks are achieving remarkable results in a wide variety of domains.", "description_method": "abstract-excerpt", "abstract": "Machine learning techniques based on neural networks are achieving remarkable results in a wide variety of domains. Often, the training of models requires large, representative datasets, which may be crowdsourced and contain sensitive information. The models should not expose private information in these datasets. Addressing this goal, we develop new algorithmic techniques for learning and a refined analysis of privacy costs within the framework of differential privacy. Our implementation and experiments demonstrate that we can train deep neural networks with non-convex objectives, under a modest privacy budget, and at a manageable cost in software complexity, training efficiency, and model quality.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2016-10", "2018-03", "2022-03"], "last_mentioned": "2022-03-17T08:02:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.02907", "slug": "5c085d544d4ba458", "title": "Semi-Supervised Classification with Graph Convolutional Networks", "authors": ["Thomas N. Kipf", "Max Welling"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.02907", "doi": null, "topics": ["Computer vision", "Machine learning"], "description": "We present a scalable approach for semi-supervised learning on graph-structured data that is based on an efficient variant of convolutional neural networks which operate directly on graphs.", "description_method": "abstract-excerpt", "abstract": "We present a scalable approach for semi-supervised learning on graph-structured data that is based on an efficient variant of convolutional neural networks which operate directly on graphs. We motivate the choice of our convolutional architecture via a localized first-order approximation of spectral graph convolutions. Our model scales linearly in the number of graph edges and learns hidden layer representations that encode both local graph structure and features of nodes. In a number of experiments on citation networks and on a knowledge graph dataset we demonstrate that our approach outperforms related methods by a significant margin.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2020-05", "2020-12", "2023-01"], "last_mentioned": "2023-01-24T21:56:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1611.10012", "slug": "8bf94733606024d1", "title": "Speed/accuracy trade-offs for modern convolutional object detectors", "authors": ["Jonathan Huang", "Vivek Rathod", "Chen Sun", "Menglong Zhu", "Anoop Korattikara", "Alireza Fathi", "Ian Fischer", "Zbigniew Wojna", "Yang Song", "Sergio Guadarrama", "Kevin Murphy"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1611.10012", "doi": null, "topics": ["Computer vision"], "description": "The goal of this paper is to serve as a guide for selecting a detection architecture that achieves the right speed/memory/accuracy balance for a given application and platform.", "description_method": "abstract-excerpt", "abstract": "The goal of this paper is to serve as a guide for selecting a detection architecture that achieves the right speed/memory/accuracy balance for a given application and platform. To this end, we investigate various ways to trade accuracy for speed and memory usage in modern convolutional object detection systems. A number of successful systems have been proposed in recent years, but apples-to-apples comparisons are difficult due to different base feature extractors (e.g., VGG, Residual Networks), different default image resolutions, as well as different hardware and software platforms. We present a unified implementation of the Faster R-CNN [Ren et al., 2015], R-FCN [Dai et al., 2016] and SSD [Liu et al., 2015] systems, which we view as \"meta-architectures\" and trace out the speed/accuracy trade-off curve created by using alternative feature extractors and varying other critical parameters such as image size within each of these meta-architectures. On one extreme end of this spectrum where speed and memory are critical, we present a detector that achieves real time speeds and can be deployed on a mobile device. On the opposite end in which accuracy is critical, we present a detector that achieves state-of-the-art performance measured on the COCO detection task.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2017-06", "2017-08", "2019-11"], "last_mentioned": "2019-11-06T16:50:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1801.04016", "slug": "ea202c0863d7fe29", "title": "Theoretical Impediments to Machine Learning With Seven Sparks from the Causal Revolution", "authors": ["Judea Pearl"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1801.04016", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Psychology"], "description": "Current machine learning systems operate, almost exclusively, in a statistical, or model-free mode, which entails severe theoretical limits on their power and performance.", "description_method": "abstract-excerpt", "abstract": "Current machine learning systems operate, almost exclusively, in a statistical, or model-free mode, which entails severe theoretical limits on their power and performance. Such systems cannot reason about interventions and retrospection and, therefore, cannot serve as the basis for strong AI. To achieve human level intelligence, learning machines need the guidance of a model of reality, similar to the ones used in causal inference tasks. To demonstrate the essential role of such models, I will present a summary of seven tasks which are beyond reach of current machine learning systems and which have been accomplished using the tools of causal modeling.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2018-01", "2018-04", "2025-07"], "last_mentioned": "2025-07-13T10:45:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.04383", "slug": "1ed81aebfb893b9a", "title": "Delayed Impact of Fair Machine Learning", "authors": ["Lydia T. Liu", "Sarah Dean", "Esther Rolf", "Max Simchowitz", "Moritz Hardt"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.04383", "doi": null, "topics": ["Machine learning", "Psychology"], "description": "Fairness in machine learning has predominantly been studied in static classification settings without concern for how decisions change the underlying population over time.", "description_method": "abstract-excerpt", "abstract": "Fairness in machine learning has predominantly been studied in static classification settings without concern for how decisions change the underlying population over time. Conventional wisdom suggests that fairness criteria promote the long-term well-being of those groups they aim to protect. We study how static fairness criteria interact with temporal indicators of well-being, such as long-term improvement, stagnation, and decline in a variable of interest. We demonstrate that even in a one-step feedback model, common fairness criteria in general do not promote improvement over time, and may in fact cause harm in cases where an unconstrained objective would not. We completely characterize the delayed impact of three standard criteria, contrasting the regimes in which these exhibit qualitatively different behavior. In addition, we find that a natural form of measurement error broadens the regime in which fairness criteria perform favorably. Our results highlight the importance of measurement and temporal modeling in the evaluation of fairness criteria, suggesting a range of new challenges and trade-offs.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2019-01", "2020-09", "2020-12"], "last_mentioned": "2020-12-04T00:50:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1903.07291", "slug": "12713dac7934d7e5", "title": "Semantic Image Synthesis with Spatially-Adaptive Normalization", "authors": ["Taesung Park", "Ming-Yu Liu", "Ting-Chun Wang", "Jun-Yan Zhu"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1903.07291", "doi": null, "topics": ["Other research"], "description": "We propose spatially-adaptive normalization, a simple but effective layer for synthesizing photorealistic images given an input semantic layout.", "description_method": "abstract-excerpt", "abstract": "We propose spatially-adaptive normalization, a simple but effective layer for synthesizing photorealistic images given an input semantic layout. Previous methods directly feed the semantic layout as input to the deep network, which is then processed through stacks of convolution, normalization, and nonlinearity layers. We show that this is suboptimal as the normalization layers tend to ``wash away'' semantic information. To address the issue, we propose using the input layout for modulating the activations in normalization layers through a spatially-adaptive, learned transformation. Experiments on several challenging datasets demonstrate the advantage of the proposed method over existing approaches, regarding both visual fidelity and alignment with input layouts. Finally, our model allows user control over both semantic and style. Code is available at https://github.com/NVlabs/SPADE .", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2019-03", "2020-02", "2021-06"], "last_mentioned": "2021-06-25T06:53:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1905.12149", "slug": "bcee028c71060e5d", "title": "SATNet: Bridging deep learning and logical reasoning using a differentiable satisfiability solver", "authors": ["Po-Wei Wang", "Priya L. Donti", "Bryan Wilder", "Zico Kolter"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1905.12149", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "Integrating logical reasoning within deep learning architectures has been a major goal of modern AI systems.", "description_method": "abstract-excerpt", "abstract": "Integrating logical reasoning within deep learning architectures has been a major goal of modern AI systems. In this paper, we propose a new direction toward this goal by introducing a differentiable (smoothed) maximum satisfiability (MAXSAT) solver that can be integrated into the loop of larger deep learning systems. Our (approximate) solver is based upon a fast coordinate descent approach to solving the semidefinite program (SDP) associated with the MAXSAT problem. We show how to analytically differentiate through the solution to this SDP and efficiently solve the associated backward pass. We demonstrate that by integrating this solver into end-to-end learning systems, we can learn the logical structure of challenging problems in a minimally supervised fashion. In particular, we show that we can learn the parity function using single-bit supervision (a traditionally hard task for deep networks) and learn how to play 9x9 Sudoku solely from examples. We also solve a \"visual Sudok\" problem that maps images of Sudoku puzzles to their associated logical solutions by combining our MAXSAT solver with a traditional convolutional architecture. Our approach thus shows promise in integrating logical structures within deep learning.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2020-05", "2020-06", "2022-12"], "last_mentioned": "2022-12-23T09:35:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2002.06177", "slug": "9ce8703232279773", "title": "The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence", "authors": ["Gary Marcus"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2002.06177", "doi": null, "topics": ["Psychology", "Neuroscience", "Machine learning"], "description": "Recent research in artificial intelligence and machine learning has largely emphasized general-purpose learning and ever-larger training sets and more and more compute.", "description_method": "abstract-excerpt", "abstract": "Recent research in artificial intelligence and machine learning has largely emphasized general-purpose learning and ever-larger training sets and more and more compute. In contrast, I propose a hybrid, knowledge-driven, reasoning-based approach, centered around cognitive models, that could provide the substrate for a richer, more robust AI than is currently possible.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2020-06", "2022-10", "2025-08"], "last_mentioned": "2025-08-10T00:37:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2007.03898", "slug": "14db901dd74d0fae", "title": "NVAE: A Deep Hierarchical Variational Autoencoder", "authors": ["Arash Vahdat", "Jan Kautz"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2007.03898", "doi": null, "topics": ["Statistics & probability", "Computer vision"], "description": "Normalizing flows, autoregressive models, variational autoencoders (VAEs), and deep energy-based models are among competing likelihood-based frameworks for deep generative learning.", "description_method": "abstract-excerpt", "abstract": "Normalizing flows, autoregressive models, variational autoencoders (VAEs), and deep energy-based models are among competing likelihood-based frameworks for deep generative learning. Among them, VAEs have the advantage of fast and tractable sampling and easy-to-access encoding networks. However, they are currently outperformed by other models such as normalizing flows and autoregressive models. While the majority of the research in VAEs is focused on the statistical challenges, we explore the orthogonal direction of carefully designing neural architectures for hierarchical VAEs. We propose Nouveau VAE (NVAE), a deep hierarchical VAE built for image generation using depth-wise separable convolutions and batch normalization. NVAE is equipped with a residual parameterization of Normal distributions and its training is stabilized by spectral regularization. We show that NVAE achieves state-of-the-art results among non-autoregressive likelihood-based models on the MNIST, CIFAR-10, CelebA 64, and CelebA HQ datasets and it provides a strong baseline on FFHQ. For example, on CIFAR-10, NVAE pushes the state-of-the-art from 2.98 to 2.91 bits per dimension, and it produces high-quality images on CelebA HQ. To the best of our knowledge, NVAE is the first successful VAE applied to natural images as large as 256$\\times$256 pixels. The source code is available at https://github.com/NVlabs/NVAE .", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2020-09", "2021-01", "2025-06"], "last_mentioned": "2025-06-27T03:35:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2010.03090", "slug": "10fe78c25509f141", "title": "Validating UTF-8 In Less Than One Instruction Per Byte", "authors": ["John Keiser", "Daniel Lemire"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2010.03090", "doi": "10.1002/spe.2920", "topics": ["Science & research", "Algorithms"], "description": "The majority of text is stored in UTF-8, which must be validated on ingestion.", "description_method": "abstract-excerpt", "abstract": "The majority of text is stored in UTF-8, which must be validated on ingestion. We present the lookup algorithm, which outperforms UTF-8 validation routines used in many libraries and languages by more than 10 times using commonly available SIMD instructions. To ensure reproducibility, our work is freely available as open source software.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2022-09", "2024-01", "2024-05"], "last_mentioned": "2024-05-16T21:10:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2105.01601", "slug": "4ecddf03fe7fbc72", "title": "MLP-Mixer: An all-MLP Architecture for Vision", "authors": ["Ilya Tolstikhin", "Neil Houlsby", "Alexander Kolesnikov", "Lucas Beyer", "Xiaohua Zhai", "Thomas Unterthiner", "Jessica Yung", "Andreas Steiner", "Daniel Keysers", "Jakob Uszkoreit", "Mario Lucic", "Alexey Dosovitskiy"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2105.01601", "doi": null, "topics": ["Computer vision", "Language models", "Machine learning"], "description": "Convolutional Neural Networks (CNNs) are the go-to model for computer vision.", "description_method": "abstract-excerpt", "abstract": "Convolutional Neural Networks (CNNs) are the go-to model for computer vision. Recently, attention-based networks, such as the Vision Transformer, have also become popular. In this paper we show that while convolutions and attention are both sufficient for good performance, neither of them are necessary. We present MLP-Mixer, an architecture based exclusively on multi-layer perceptrons (MLPs). MLP-Mixer contains two types of layers: one with MLPs applied independently to image patches (i.e. \"mixing\" the per-location features), and one with MLPs applied across patches (i.e. \"mixing\" spatial information). When trained on large datasets, or with modern regularization schemes, MLP-Mixer attains competitive scores on image classification benchmarks, with pre-training and inference cost comparable to state-of-the-art models. We hope that these results spark further research beyond the realms of well established CNNs and Transformers.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2021-05", "2023-11", "2026-06"], "last_mentioned": "2026-06-05T17:19:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.12794", "slug": "d35823348d2d11ca", "title": "GraphCast: Learning skillful medium-range global weather forecasting", "authors": ["Remi Lam", "Alvaro Sanchez-Gonzalez", "Matthew Willson", "Peter Wirnsberger", "Meire Fortunato", "Ferran Alet", "Suman Ravuri", "Timo Ewalds", "Zach Eaton-Rosen", "Weihua Hu", "Alexander Merose", "Stephan Hoyer", "George Holland", "Oriol Vinyals", "Jacklynn Stott", "Alexander Pritzel", "Shakir Mohamed", "Peter Battaglia"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.12794", "doi": null, "topics": ["Machine learning", "Psychology", "Economics"], "description": "Global medium-range weather forecasting is critical to decision-making across many social and economic domains.", "description_method": "abstract-excerpt", "abstract": "Global medium-range weather forecasting is critical to decision-making across many social and economic domains. Traditional numerical weather prediction uses increased compute resources to improve forecast accuracy, but cannot directly use historical weather data to improve the underlying model. We introduce a machine learning-based method called \"GraphCast\", which can be trained directly from reanalysis data. It predicts hundreds of weather variables, over 10 days at 0.25 degree resolution globally, in under one minute. We show that GraphCast significantly outperforms the most accurate operational deterministic systems on 90% of 1380 verification targets, and its forecasts support better severe event prediction, including tropical cyclones, atmospheric rivers, and extreme temperatures. GraphCast is a key advance in accurate and efficient weather forecasting, and helps realize the promise of machine learning for modeling complex dynamical systems.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2023-08", "2025-12", "2026-08"], "last_mentioned": "2026-08-08T13:56:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2309.00941", "slug": "30801da11d2a7373", "title": "Emergent Linear Representations in World Models of Self-Supervised Sequence Models", "authors": ["Neel Nanda", "Andrew Lee", "Martin Wattenberg"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2309.00941", "doi": null, "topics": ["Psychology", "Machine learning"], "description": "How do sequence models represent their decision-making process?", "description_method": "abstract-excerpt", "abstract": "How do sequence models represent their decision-making process? Prior work suggests that Othello-playing neural network learned nonlinear models of the board state (Li et al., 2023). In this work, we provide evidence of a closely related linear representation of the board. In particular, we show that probing for \"my colour\" vs. \"opponent's colour\" may be a simple yet powerful way to interpret the model's internal state. This precise understanding of the internal representations allows us to control the model's behaviour with simple vector arithmetic. Linear representations enable significant interpretability progress, which we demonstrate with further exploration of how the world model is computed.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2024-07", "2024-11", "2024-12"], "last_mentioned": "2024-12-19T19:37:04Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.16028", "slug": "474cdf560046ca4e", "title": "What Algorithms can Transformers Learn? A Study in Length Generalization", "authors": ["Hattie Zhou", "Arwen Bradley", "Etai Littwin", "Noam Razin", "Omid Saremi", "Josh Susskind", "Samy Bengio", "Preetum Nakkiran"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.16028", "doi": null, "topics": ["Language models", "Machine learning", "Algorithms"], "description": "Large language models exhibit surprising emergent generalization properties, yet also struggle on many simple reasoning tasks such as arithmetic and parity.", "description_method": "abstract-excerpt", "abstract": "Large language models exhibit surprising emergent generalization properties, yet also struggle on many simple reasoning tasks such as arithmetic and parity. This raises the question of if and when Transformer models can learn the true algorithm for solving a task. We study the scope of Transformers' abilities in the specific setting of length generalization on algorithmic tasks. Here, we propose a unifying framework to understand when and how Transformers can exhibit strong length generalization on a given task. Specifically, we leverage RASP (Weiss et al., 2021) -- a programming language designed for the computational model of a Transformer -- and introduce the RASP-Generalization Conjecture: Transformers tend to length generalize on a task if the task can be solved by a short RASP program which works for all input lengths. This simple conjecture remarkably captures most known instances of length generalization on algorithmic tasks. Moreover, we leverage our insights to drastically improve generalization performance on traditionally hard tasks (such as parity and addition). On the theoretical side, we give a simple example where the \"min-degree-interpolator\" model of learning from Abbe et al. (2023) does not correctly predict Transformers' out-of-distribution behavior, but our conjecture does. Overall, our work provides a novel perspective on the mechanisms of compositional generalization and the algorithmic capabilities of Transformers.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2024-02", "2024-04", "2024-12"], "last_mentioned": "2024-12-14T15:09:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:hep-th/0410270", "slug": "a49fd1b59f242f71", "title": "Spontaneous Inflation and the Origin of the Arrow of Time", "authors": ["Sean M. Carroll", "Jennifer Chen"], "year": 2004, "venue": "arXiv", "url": "https://arxiv.org/abs/hep-th/0410270", "doi": null, "topics": ["Statistics & probability", "Physics"], "description": "We suggest that spontaneous eternal inflation can provide a natural explanation for the thermodynamic arrow of time, and discuss the underlying assumptions and consequences of this view.", "description_method": "abstract-excerpt", "abstract": "We suggest that spontaneous eternal inflation can provide a natural explanation for the thermodynamic arrow of time, and discuss the underlying assumptions and consequences of this view. In the absence of inflation, we argue that systems coupled to gravity usually evolve asymptotically to the vacuum, which is the only natural state in a thermodynamic sense. In the presence of a small positive vacuum energy and an appropriate inflaton field, the de Sitter vacuum is unstable to the spontaneous onset of inflation at a higher energy scale. Starting from de Sitter, inflation can increase the total entropy of the universe without bound, creating universes similar to ours in the process. An important consequence of this picture is that inflation occurs asymptotically both forwards and backwards in time, implying a universe that is (statistically) time-symmetric on ultra-large scales.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2016-07", "2018-07", "2021-05"], "last_mentioned": "2021-05-14T12:01:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.adk1189", "slug": "4704c02777e1fc5a", "title": "Physics-based early warning signal shows that AMOC is on tipping course", "authors": ["René M. van Westen", "Michael Kliphuis", "Henk A. Dijkstra"], "year": 2024, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.adk1189", "doi": "10.1126/sciadv.adk1189", "topics": ["Physics", "Statistics & probability"], "description": "One of the most prominent climate tipping elements is the Atlantic meridional overturning circulation (AMOC), which can potentially collapse because of the input of fresh water in the North Atlantic.", "description_method": "abstract-excerpt", "abstract": "One of the most prominent climate tipping elements is the Atlantic meridional overturning circulation (AMOC), which can potentially collapse because of the input of fresh water in the North Atlantic. Although AMOC collapses have been induced in complex global climate models by strong freshwater forcing, the processes of an AMOC tipping event have so far not been investigated. Here, we show results of the first tipping event in the Community Earth System Model, including the large climate impacts of the collapse. Using these results, we develop a physics-based and observable early warning signal of AMOC tipping: the minimum of the AMOC-induced freshwater transport at the southern boundary of the Atlantic. Reanalysis products indicate that the present-day AMOC is on route to tipping. The early warning signal is a useful alternative to classical statistical ones, which, when applied to our simulated tipping event, turn out to be sensitive to the analyzed time interval before tipping.", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2024-02", "2024-06", "2024-10"], "last_mentioned": "2024-10-23T20:52:54Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3447786.3456248", "slug": "fcaf2b0858c76d6c", "title": "Unikraft", "authors": ["Simon Kuenzer", "Vlad-Andrei Bădoiu", "Hugo Lefeuvre", "Sharan Santhanam", "Alexander Jung", "Gaulthier Gain", "Cyril Soldani", "Costin Lupu", "Ştefan Teodorescu", "Costi Răducanu", "Cristian Banu", "Laurent Mathy", "Răzvan Deaconescu", "Costin Raiciu", "Felipe Huici"], "year": 2021, "venue": "Proceedings of the Sixteenth European Conference on Computer Systems", "url": "https://doi.org/10.1145/3447786.3456248", "doi": "10.1145/3447786.3456248", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2021-05", "2022-02", "2024-05"], "last_mentioned": "2024-05-14T19:49:24Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "url:https://papers.nips.cc/paper/2015/hash/86df7dcfd896fcaf2674f757a2463eba-Abstract.html", "slug": "cd4bb074ab2ae3a5", "title": "Hidden Technical Debt in Machine Learning Systems", "authors": ["Sculley, D.", "Holt, Gary", "Golovin, Daniel", "Davydov, Eugene", "Phillips, Todd", "Ebner, Dietmar", "Chaudhary, Vinay", "Young, Michael", "Crespo, Jean-François", "Dennison, Dan"], "year": 2015, "venue": "Advances in Neural Information Processing Systems", "url": "https://papers.nips.cc/paper/2015/hash/86df7dcfd896fcaf2674f757a2463eba-Abstract.html", "doi": null, "topics": ["Machine learning"], "description": "", "description_method": null, "abstract": "", "score": 1.0, "mention_count": 3, "user_count": 3, "months": ["2021-05", "2021-11", "2022-09"], "last_mentioned": "2022-09-10T12:33:14Z", "metadata_source": "publisher-citation-tags", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-020-0820-9", "slug": "a0da7a3308348399", "title": "The proximal origin of SARS-CoV-2", "authors": ["Kristian G. Andersen", "Andrew Rambaut", "W. Ian Lipkin", "Edward C. Holmes", "Robert F. Garry"], "year": 2020, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-020-0820-9", "doi": "10.1038/s41591-020-0820-9", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 40, "user_count": 38, "months": ["2020-03", "2020-04", "2020-05", "2020-07", "2020-09", "2020-12", "2021-02", "2021-03", "2021-05", "2021-06", "2022-10", "2023-02", "2023-06", "2023-07", "2023-08", "2024-01", "2024-06", "2025-01", "2025-09"], "last_mentioned": "2025-09-01T17:11:17Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/ncomms2380", "slug": "adce7421a9ce33c7", "title": "The impact of free-ranging domestic cats on wildlife of the United States", "authors": ["Scott R. Loss", "Tom Will", "Peter P. Marra"], "year": 2013, "venue": "Nature Communications", "url": "https://doi.org/10.1038/ncomms2380", "doi": "10.1038/ncomms2380", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 33, "user_count": 33, "months": ["2016-09", "2017-09", "2018-03", "2018-07", "2019-05", "2019-07", "2021-08", "2021-12", "2022-05", "2022-07", "2022-08", "2022-09", "2023-04", "2023-06", "2023-07", "2023-09", "2023-10", "2023-11", "2023-12", "2024-01", "2024-02", "2024-10", "2025-07", "2026-03"], "last_mentioned": "2026-03-26T03:32:02Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2001.08361", "slug": "82302458a6d03bc0", "title": "Scaling Laws for Neural Language Models", "authors": ["Jared Kaplan", "Sam McCandlish", "Tom Henighan", "Tom B. Brown", "Benjamin Chess", "Rewon Child", "Scott Gray", "Alec Radford", "Jeffrey Wu", "Dario Amodei"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2001.08361", "doi": null, "topics": ["Language models", "Statistics & probability"], "description": "We study empirical scaling laws for language model performance on the cross-entropy loss.", "description_method": "abstract-excerpt", "abstract": "We study empirical scaling laws for language model performance on the cross-entropy loss. The loss scales as a power-law with model size, dataset size, and the amount of compute used for training, with some trends spanning more than seven orders of magnitude. Other architectural details such as network width or depth have minimal effects within a wide range. Simple equations govern the dependence of overfitting on model/dataset size and the dependence of training speed on model size. These relationships allow us to determine the optimal allocation of a fixed compute budget. Larger models are significantly more sample-efficient, such that optimally compute-efficient training involves training very large models on a relatively modest amount of data and stopping significantly before convergence.", "score": 0.0, "mention_count": 31, "user_count": 27, "months": ["2020-01", "2020-09", "2021-01", "2021-02", "2021-08", "2021-10", "2022-03", "2023-01", "2023-02", "2023-03", "2023-04", "2023-05", "2023-07", "2023-08", "2023-12", "2024-02", "2024-03", "2024-09", "2024-10", "2024-11", "2025-01", "2025-03", "2025-08", "2026-02", "2026-05", "2026-06", "2026-07"], "last_mentioned": "2026-07-07T19:03:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2203.15556", "slug": "677985344c2e8bbc", "title": "Training Compute-Optimal Large Language Models", "authors": ["Jordan Hoffmann", "Sebastian Borgeaud", "Arthur Mensch", "Elena Buchatskaya", "Trevor Cai", "Eliza Rutherford", "Diego de Las Casas", "Lisa Anne Hendricks", "Johannes Welbl", "Aidan Clark", "Tom Hennigan", "Eric Noland", "Katie Millican", "George van den Driessche", "Bogdan Damoc", "Aurelia Guy", "Simon Osindero", "Karen Simonyan", "Erich Elsen", "Jack W. Rae", "Oriol Vinyals", "Laurent Sifre"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2203.15556", "doi": null, "topics": ["Language models", "Statistics & probability"], "description": "We investigate the optimal model size and number of tokens for training a transformer language model under a given compute budget.", "description_method": "abstract-excerpt", "abstract": "We investigate the optimal model size and number of tokens for training a transformer language model under a given compute budget. We find that current large language models are significantly undertrained, a consequence of the recent focus on scaling language models whilst keeping the amount of training data constant. By training over 400 language models ranging from 70 million to over 16 billion parameters on 5 to 500 billion tokens, we find that for compute-optimal training, the model size and the number of training tokens should be scaled equally: for every doubling of model size the number of training tokens should also be doubled. We test this hypothesis by training a predicted compute-optimal model, Chinchilla, that uses the same compute budget as Gopher but with 70B parameters and 4$\\times$ more more data. Chinchilla uniformly and significantly outperforms Gopher (280B), GPT-3 (175B), Jurassic-1 (178B), and Megatron-Turing NLG (530B) on a large range of downstream evaluation tasks. This also means that Chinchilla uses substantially less compute for fine-tuning and inference, greatly facilitating downstream usage. As a highlight, Chinchilla reaches a state-of-the-art average accuracy of 67.5% on the MMLU benchmark, greater than a 7% improvement over Gopher.", "score": 0.0, "mention_count": 31, "user_count": 27, "months": ["2022-04", "2022-08", "2022-09", "2022-10", "2022-11", "2022-12", "2023-02", "2023-03", "2023-04", "2023-05", "2023-07", "2023-09", "2023-12", "2025-08", "2025-12", "2026-06"], "last_mentioned": "2026-06-16T18:48:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2210.13382", "slug": "5c09d2330f8c5098", "title": "Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Task", "authors": ["Kenneth Li", "Aspen K. Hopkins", "David Bau", "Fernanda Viégas", "Hanspeter Pfister", "Martin Wattenberg"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2210.13382", "doi": null, "topics": ["Language models", "Statistics & probability"], "description": "Language models show a surprising range of capabilities, but the source of their apparent competence is unclear.", "description_method": "abstract-excerpt", "abstract": "Language models show a surprising range of capabilities, but the source of their apparent competence is unclear. Do these networks just memorize a collection of surface statistics, or do they rely on internal representations of the process that generates the sequences they see? We investigate this question by applying a variant of the GPT model to the task of predicting legal moves in a simple board game, Othello. Although the network has no a priori knowledge of the game or its rules, we uncover evidence of an emergent nonlinear internal representation of the board state. Interventional experiments indicate this representation can be used to control the output of the network and create \"latent saliency maps\" that can help explain predictions in human terms.", "score": 0.0, "mention_count": 30, "user_count": 19, "months": ["2023-01", "2023-02", "2023-03", "2023-04", "2023-05", "2024-03", "2024-05", "2024-06", "2024-10", "2024-11", "2026-04"], "last_mentioned": "2026-04-30T11:35:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2302.13971", "slug": "72496ff2f98e79f5", "title": "LLaMA: Open and Efficient Foundation Language Models", "authors": ["Hugo Touvron", "Thibaut Lavril", "Gautier Izacard", "Xavier Martinet", "Marie-Anne Lachaux", "Timothée Lacroix", "Baptiste Rozière", "Naman Goyal", "Eric Hambro", "Faisal Azhar", "Aurelien Rodriguez", "Armand Joulin", "Edouard Grave", "Guillaume Lample"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2302.13971", "doi": null, "topics": ["Language models", "Algorithms"], "description": "We introduce LLaMA, a collection of foundation language models ranging from 7B to 65B parameters.", "description_method": "abstract-excerpt", "abstract": "We introduce LLaMA, a collection of foundation language models ranging from 7B to 65B parameters. We train our models on trillions of tokens, and show that it is possible to train state-of-the-art models using publicly available datasets exclusively, without resorting to proprietary and inaccessible datasets. In particular, LLaMA-13B outperforms GPT-3 (175B) on most benchmarks, and LLaMA-65B is competitive with the best models, Chinchilla-70B and PaLM-540B. We release all our models to the research community.", "score": 0.0, "mention_count": 30, "user_count": 27, "months": ["2023-03", "2023-04", "2023-05", "2023-06", "2023-07", "2023-09", "2023-12", "2024-03", "2024-06", "2025-02"], "last_mentioned": "2025-02-07T14:53:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2412.19437", "slug": "70d3c4e3779a388c", "title": "DeepSeek-V3 Technical Report", "authors": [" DeepSeek-AI", "Aixin Liu", "Bei Feng", "Bing Xue", "Bingxuan Wang", "Bochao Wu", "Chengda Lu", "Chenggang Zhao", "Chengqi Deng", "Chenyu Zhang", "Chong Ruan", "Damai Dai", "Daya Guo", "Dejian Yang", "Deli Chen", "Dongjie Ji", "Erhang Li", "Fangyun Lin", "Fucong Dai", "Fuli Luo", "Guangbo Hao", "Guanting Chen", "Guowei Li", "H. Zhang", "Han Bao", "Hanwei Xu", "Haocheng Wang", "Haowei Zhang", "Honghui Ding", "Huajian Xin", "Huazuo Gao", "Hui Li", "Hui Qu", "J. L. Cai", "Jian Liang", "Jianzhong Guo", "Jiaqi Ni", "Jiashi Li", "Jiawei Wang", "Jin Chen", "Jingchang Chen", "Jingyang Yuan", "Junjie Qiu", "Junlong Li", "Junxiao Song", "Kai Dong", "Kai Hu", "Kaige Gao", "Kang Guan", "Kexin Huang", "Kuai Yu", "Lean Wang", "Lecong Zhang", "Lei Xu", "Leyi Xia", "Liang Zhao", "Litong Wang", "Liyue Zhang", "Meng Li", "Miaojun Wang", "Mingchuan Zhang", "Minghua Zhang", "Minghui Tang", "Mingming Li", "Ning Tian", "Panpan Huang", "Peiyi Wang", "Peng Zhang", "Qiancheng Wang", "Qihao Zhu", "Qinyu Chen", "Qiushi Du", "R. J. Chen", "R. L. Jin", "Ruiqi Ge", "Ruisong Zhang", "Ruizhe Pan", "Runji Wang", "Runxin Xu", "Ruoyu Zhang", "Ruyi Chen", "S. S. Li", "Shanghao Lu", "Shangyan Zhou", "Shanhuang Chen", "Shaoqing Wu", "Shengfeng Ye", "Shengfeng Ye", "Shirong Ma", "Shiyu Wang", "Shuang Zhou", "Shuiping Yu", "Shunfeng Zhou", "Shuting Pan", "T. Wang", "Tao Yun", "Tian Pei", "Tianyu Sun", "W. L. Xiao", "Wangding Zeng", "Wanjia Zhao", "Wei An", "Wen Liu", "Wenfeng Liang", "Wenjun Gao", "Wenqin Yu", "Wentao Zhang", "X. Q. Li", "Xiangyue Jin", "Xianzu Wang", "Xiao Bi", "Xiaodong Liu", "Xiaohan Wang", "Xiaojin Shen", "Xiaokang Chen", "Xiaokang Zhang", "Xiaosha Chen", "Xiaotao Nie", "Xiaowen Sun", "Xiaoxiang Wang", "Xin Cheng", "Xin Liu", "Xin Xie", "Xingchao Liu", "Xingkai Yu", "Xinnan Song", "Xinxia Shan", "Xinyi Zhou", "Xinyu Yang", "Xinyuan Li", "Xuecheng Su", "Xuheng Lin", "Y. K. Li", "Y. Q. Wang", "Y. X. Wei", "Y. X. Zhu", "Yang Zhang", "Yanhong Xu", "Yanhong Xu", "Yanping Huang", "Yao Li", "Yao Zhao", "Yaofeng Sun", "Yaohui Li", "Yaohui Wang", "Yi Yu", "Yi Zheng", "Yichao Zhang", "Yifan Shi", "Yiliang Xiong", "Ying He", "Ying Tang", "Yishi Piao", "Yisong Wang", "Yixuan Tan", "Yiyang Ma", "Yiyuan Liu", "Yongqiang Guo", "Yu Wu", "Yuan Ou", "Yuchen Zhu", "Yuduan Wang", "Yue Gong", "Yuheng Zou", "Yujia He", "Yukun Zha", "Yunfan Xiong", "Yunxian Ma", "Yuting Yan", "Yuxiang Luo", "Yuxiang You", "Yuxuan Liu", "Yuyang Zhou", "Z. F. Wu", "Z. Z. Ren", "Zehui Ren", "Zhangli Sha", "Zhe Fu", "Zhean Xu", "Zhen Huang", "Zhen Zhang", "Zhenda Xie", "Zhengyan Zhang", "Zhewen Hao", "Zhibin Gou", "Zhicheng Ma", "Zhigang Yan", "Zhihong Shao", "Zhipeng Xu", "Zhiyu Wu", "Zhongyu Zhang", "Zhuoshu Li", "Zihui Gu", "Zijia Zhu", "Zijun Liu", "Zilin Li", "Ziwei Xie", "Ziyang Song", "Ziyi Gao", "Zizheng Pan"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2412.19437", "doi": null, "topics": ["Machine learning", "Language models"], "description": "We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token.", "description_method": "abstract-excerpt", "abstract": "We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) and DeepSeekMoE architectures, which were thoroughly validated in DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets a multi-token prediction training objective for stronger performance. We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens, followed by Supervised Fine-Tuning and Reinforcement Learning stages to fully harness its capabilities. Comprehensive evaluations reveal that DeepSeek-V3 outperforms other open-source models and achieves performance comparable to leading closed-source models. Despite its excellent performance, DeepSeek-V3 requires only 2.788M H800 GPU hours for its full training. In addition, its training process is remarkably stable. Throughout the entire training process, we did not experience any irrecoverable loss spikes or perform any rollbacks. The model checkpoints are available at https://github.com/deepseek-ai/DeepSeek-V3.", "score": 0.0, "mention_count": 28, "user_count": 23, "months": ["2025-01", "2025-02", "2025-03", "2025-04", "2025-05", "2025-09", "2025-12", "2026-02", "2026-03", "2026-05", "2026-06"], "last_mentioned": "2026-06-21T19:07:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-021-01630-0", "slug": "a510641e29ba4064", "title": "Risks of myocarditis, pericarditis, and cardiac arrhythmias associated with COVID-19 vaccination or SARS-CoV-2 infection", "authors": ["Martina Patone", "Xue W. Mei", "Lahiru Handunnetthi", "Sharon Dixon", "Francesco Zaccardi", "Manu Shankar-Hari", "Peter Watkinson", "Kamlesh Khunti", "Anthony Harnden", "Carol A. C. Coupland", "Keith M. Channon", "Nicholas L. Mills", "Aziz Sheikh", "Julia Hippisley-Cox"], "year": 2021, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-021-01630-0", "doi": "10.1038/s41591-021-01630-0", "topics": ["Biology & medicine"], "description": "Abstract Although myocarditis and pericarditis were not observed as adverse events in coronavirus disease 2019 (COVID-19) vaccine trials, there have been numerous reports of suspected cases following vaccination in the general population.", "description_method": "abstract-excerpt", "abstract": "Abstract Although myocarditis and pericarditis were not observed as adverse events in coronavirus disease 2019 (COVID-19) vaccine trials, there have been numerous reports of suspected cases following vaccination in the general population. We undertook a self-controlled case series study of people aged 16 or older vaccinated for COVID-19 in England between 1 December 2020 and 24 August 2021 to investigate hospital admission or death from myocarditis, pericarditis and cardiac arrhythmias in the 1–28 days following adenovirus (ChAdOx1, n = 20,615,911) or messenger RNA-based (BNT162b2, n = 16,993,389; mRNA-1273, n = 1,006,191) vaccines or a severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) positive test ( n = 3,028,867). We found increased risks of myocarditis associated with the first dose of ChAdOx1 and BNT162b2 vaccines and the first and second doses of the mRNA-1273 vaccine over the 1–28 days postvaccination period, and after a SARS-CoV-2 positive test. We estimated an extra two (95% confidence interval (CI) 0, 3), one (95% CI 0, 2) and six (95% CI 2, 8) myocarditis events per 1 million people vaccinated with ChAdOx1, BNT162b2 and mRNA-1273, respectively, in the 28 days following a first dose and an extra ten (95% CI 7, 11) myocarditis events per 1 million vaccinated in the 28 days after a second dose of mRNA-1273. This compares with an extra 40 (95% CI 38, 41) myocarditis events per 1 million patients in the 28 days following a SARS-CoV-2 positive test. We also observed increased risks of pericarditis and cardiac arrhythmias following a positive SARS-CoV-2 test. Similar associations were not observed with any of the COVID-19 vaccines, apart from an increased risk of arrhythmia following a second dose of mRNA-1273. Subgroup analyses by age showed the increased risk of myocarditis associated with the two mRNA vaccines was present only in those younger than 40.", "score": 0.0, "mention_count": 27, "user_count": 15, "months": ["2021-12", "2022-01", "2022-02", "2022-11", "2022-12", "2023-02", "2023-03", "2023-05", "2023-10", "2024-03", "2024-05"], "last_mentioned": "2024-05-05T19:13:06Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1410.5401", "slug": "3e9ef4a6e85e2578", "title": "Neural Turing Machines", "authors": ["Alex Graves", "Greg Wayne", "Ivo Danihelka"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1410.5401", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "We extend the capabilities of neural networks by coupling them to external memory resources, which they can interact with by attentional processes.", "description_method": "abstract-excerpt", "abstract": "We extend the capabilities of neural networks by coupling them to external memory resources, which they can interact with by attentional processes. The combined system is analogous to a Turing Machine or Von Neumann architecture but is differentiable end-to-end, allowing it to be efficiently trained with gradient descent. Preliminary results demonstrate that Neural Turing Machines can infer simple algorithms such as copying, sorting, and associative recall from input and output examples.", "score": 0.0, "mention_count": 25, "user_count": 22, "months": ["2014-10", "2015-02", "2015-03", "2015-05", "2015-07", "2015-08", "2015-09", "2016-01", "2016-05", "2016-07", "2016-12", "2017-04", "2017-07", "2018-06", "2020-02", "2020-06", "2021-07", "2022-02", "2023-01", "2024-04", "2025-10", "2026-07"], "last_mentioned": "2026-07-07T19:03:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2304.15004", "slug": "d7cd8c081ad54f2f", "title": "Are Emergent Abilities of Large Language Models a Mirage?", "authors": ["Rylan Schaeffer", "Brando Miranda", "Sanmi Koyejo"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2304.15004", "doi": null, "topics": ["Language models", "Statistics & probability", "Psychology"], "description": "Recent work claims that large language models display emergent abilities, abilities not present in smaller-scale models that are present in larger-scale models.", "description_method": "abstract-excerpt", "abstract": "Recent work claims that large language models display emergent abilities, abilities not present in smaller-scale models that are present in larger-scale models. What makes emergent abilities intriguing is two-fold: their sharpness, transitioning seemingly instantaneously from not present to present, and their unpredictability, appearing at seemingly unforeseeable model scales. Here, we present an alternative explanation for emergent abilities: that for a particular task and model family, when analyzing fixed model outputs, emergent abilities appear due to the researcher's choice of metric rather than due to fundamental changes in model behavior with scale. Specifically, nonlinear or discontinuous metrics produce apparent emergent abilities, whereas linear or continuous metrics produce smooth, continuous predictable changes in model performance. We present our alternative explanation in a simple mathematical model, then test it in three complementary ways: we (1) make, test and confirm three predictions on the effect of metric choice using the InstructGPT/GPT-3 family on tasks with claimed emergent abilities; (2) make, test and confirm two predictions about metric choices in a meta-analysis of emergent abilities on BIG-Bench; and (3) show to choose metrics to produce never-before-seen seemingly emergent abilities in multiple vision tasks across diverse deep networks. Via all three analyses, we provide evidence that alleged emergent abilities evaporate with different metrics or with better statistics, and may not be a fundamental property of scaling AI models.", "score": 0.0, "mention_count": 25, "user_count": 20, "months": ["2023-05", "2023-06", "2023-09", "2023-10", "2023-12", "2024-03", "2024-04", "2024-09", "2025-02", "2025-05", "2025-08", "2026-01", "2026-03", "2026-04", "2026-05", "2026-07"], "last_mentioned": "2026-07-13T14:45:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2501.12948", "slug": "cecabff8e7252f59", "title": "DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning", "authors": [" DeepSeek-AI", "Daya Guo", "Dejian Yang", "Haowei Zhang", "Junxiao Song", "Peiyi Wang", "Qihao Zhu", "Runxin Xu", "Ruoyu Zhang", "Shirong Ma", "Xiao Bi", "Xiaokang Zhang", "Xingkai Yu", "Yu Wu", "Z. F. Wu", "Zhibin Gou", "Zhihong Shao", "Zhuoshu Li", "Ziyi Gao", "Aixin Liu", "Bing Xue", "Bingxuan Wang", "Bochao Wu", "Bei Feng", "Chengda Lu", "Chenggang Zhao", "Chengqi Deng", "Chenyu Zhang", "Chong Ruan", "Damai Dai", "Deli Chen", "Dongjie Ji", "Erhang Li", "Fangyun Lin", "Fucong Dai", "Fuli Luo", "Guangbo Hao", "Guanting Chen", "Guowei Li", "H. Zhang", "Han Bao", "Hanwei Xu", "Haocheng Wang", "Honghui Ding", "Huajian Xin", "Huazuo Gao", "Hui Qu", "Hui Li", "Jianzhong Guo", "Jiashi Li", "Jiawei Wang", "Jingchang Chen", "Jingyang Yuan", "Junjie Qiu", "Junlong Li", "J. L. Cai", "Jiaqi Ni", "Jian Liang", "Jin Chen", "Kai Dong", "Kai Hu", "Kaige Gao", "Kang Guan", "Kexin Huang", "Kuai Yu", "Lean Wang", "Lecong Zhang", "Liang Zhao", "Litong Wang", "Liyue Zhang", "Lei Xu", "Leyi Xia", "Mingchuan Zhang", "Minghua Zhang", "Minghui Tang", "Meng Li", "Miaojun Wang", "Mingming Li", "Ning Tian", "Panpan Huang", "Peng Zhang", "Qiancheng Wang", "Qinyu Chen", "Qiushi Du", "Ruiqi Ge", "Ruisong Zhang", "Ruizhe Pan", "Runji Wang", "R. J. Chen", "R. L. Jin", "Ruyi Chen", "Shanghao Lu", "Shangyan Zhou", "Shanhuang Chen", "Shengfeng Ye", "Shiyu Wang", "Shuiping Yu", "Shunfeng Zhou", "Shuting Pan", "S. S. Li", "Shuang Zhou", "Shaoqing Wu", "Shengfeng Ye", "Tao Yun", "Tian Pei", "Tianyu Sun", "T. Wang", "Wangding Zeng", "Wanjia Zhao", "Wen Liu", "Wenfeng Liang", "Wenjun Gao", "Wenqin Yu", "Wentao Zhang", "W. L. Xiao", "Wei An", "Xiaodong Liu", "Xiaohan Wang", "Xiaokang Chen", "Xiaotao Nie", "Xin Cheng", "Xin Liu", "Xin Xie", "Xingchao Liu", "Xinyu Yang", "Xinyuan Li", "Xuecheng Su", "Xuheng Lin", "X. Q. Li", "Xiangyue Jin", "Xiaojin Shen", "Xiaosha Chen", "Xiaowen Sun", "Xiaoxiang Wang", "Xinnan Song", "Xinyi Zhou", "Xianzu Wang", "Xinxia Shan", "Y. K. Li", "Y. Q. Wang", "Y. X. Wei", "Yang Zhang", "Yanhong Xu", "Yao Li", "Yao Zhao", "Yaofeng Sun", "Yaohui Wang", "Yi Yu", "Yichao Zhang", "Yifan Shi", "Yiliang Xiong", "Ying He", "Yishi Piao", "Yisong Wang", "Yixuan Tan", "Yiyang Ma", "Yiyuan Liu", "Yongqiang Guo", "Yuan Ou", "Yuduan Wang", "Yue Gong", "Yuheng Zou", "Yujia He", "Yunfan Xiong", "Yuxiang Luo", "Yuxiang You", "Yuxuan Liu", "Yuyang Zhou", "Y. X. Zhu", "Yanhong Xu", "Yanping Huang", "Yaohui Li", "Yi Zheng", "Yuchen Zhu", "Yunxian Ma", "Ying Tang", "Yukun Zha", "Yuting Yan", "Z. Z. Ren", "Zehui Ren", "Zhangli Sha", "Zhe Fu", "Zhean Xu", "Zhenda Xie", "Zhengyan Zhang", "Zhewen Hao", "Zhicheng Ma", "Zhigang Yan", "Zhiyu Wu", "Zihui Gu", "Zijia Zhu", "Zijun Liu", "Zilin Li", "Ziwei Xie", "Ziyang Song", "Zizheng Pan", "Zhen Huang", "Zhipeng Xu", "Zhongyu Zhang", "Zhen Zhang"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2501.12948", "doi": "10.1038/s41586-025-09422-z", "topics": ["Machine learning", "Psychology", "Mathematics"], "description": "General reasoning represents a long-standing and formidable challenge in artificial intelligence.", "description_method": "abstract-excerpt", "abstract": "General reasoning represents a long-standing and formidable challenge in artificial intelligence. Recent breakthroughs, exemplified by large language models (LLMs) and chain-of-thought prompting, have achieved considerable success on foundational reasoning tasks. However, this success is heavily contingent upon extensive human-annotated demonstrations, and models' capabilities are still insufficient for more complex problems. Here we show that the reasoning abilities of LLMs can be incentivized through pure reinforcement learning (RL), obviating the need for human-labeled reasoning trajectories. The proposed RL framework facilitates the emergent development of advanced reasoning patterns, such as self-reflection, verification, and dynamic strategy adaptation. Consequently, the trained model achieves superior performance on verifiable tasks such as mathematics, coding competitions, and STEM fields, surpassing its counterparts trained via conventional supervised learning on human demonstrations. Moreover, the emergent reasoning patterns exhibited by these large-scale models can be systematically harnessed to guide and enhance the reasoning capabilities of smaller models.", "score": 0.0, "mention_count": 24, "user_count": 24, "months": ["2025-01", "2025-02", "2025-04", "2025-06", "2025-09", "2025-10", "2025-11", "2026-01", "2026-04", "2026-06", "2026-07"], "last_mentioned": "2026-07-12T07:23:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2207.05221", "slug": "665d09faee243328", "title": "Language Models (Mostly) Know What They Know", "authors": ["Saurav Kadavath", "Tom Conerly", "Amanda Askell", "Tom Henighan", "Dawn Drain", "Ethan Perez", "Nicholas Schiefer", "Zac Hatfield-Dodds", "Nova DasSarma", "Eli Tran-Johnson", "Scott Johnston", "Sheer El-Showk", "Andy Jones", "Nelson Elhage", "Tristan Hume", "Anna Chen", "Yuntao Bai", "Sam Bowman", "Stanislav Fort", "Deep Ganguli", "Danny Hernandez", "Josh Jacobson", "Jackson Kernion", "Shauna Kravec", "Liane Lovitt", "Kamal Ndousse", "Catherine Olsson", "Sam Ringer", "Dario Amodei", "Tom Brown", "Jack Clark", "Nicholas Joseph", "Ben Mann", "Sam McCandlish", "Chris Olah", "Jared Kaplan"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2207.05221", "doi": null, "topics": ["Language models", "Statistics & probability", "Mathematics"], "description": "We study whether language models can evaluate the validity of their own claims and predict which questions they will be able to answer correctly.", "description_method": "abstract-excerpt", "abstract": "We study whether language models can evaluate the validity of their own claims and predict which questions they will be able to answer correctly. We first show that larger models are well-calibrated on diverse multiple choice and true/false questions when they are provided in the right format. Thus we can approach self-evaluation on open-ended sampling tasks by asking models to first propose answers, and then to evaluate the probability \"P(True)\" that their answers are correct. We find encouraging performance, calibration, and scaling for P(True) on a diverse array of tasks. Performance at self-evaluation further improves when we allow models to consider many of their own samples before predicting the validity of one specific possibility. Next, we investigate whether models can be trained to predict \"P(IK)\", the probability that \"I know\" the answer to a question, without reference to any particular proposed answer. Models perform well at predicting P(IK) and partially generalize across tasks, though they struggle with calibration of P(IK) on new tasks. The predicted P(IK) probabilities also increase appropriately in the presence of relevant source materials in the context, and in the presence of hints towards the solution of mathematical word problems. We hope these observations lay the groundwork for training more honest models, and for investigating how honesty generalizes to cases where models are trained on objectives other than the imitation of human writing.", "score": 0.0, "mention_count": 23, "user_count": 8, "months": ["2022-08", "2023-05", "2023-09", "2023-10", "2023-11", "2024-06", "2024-09", "2024-10", "2025-12"], "last_mentioned": "2025-12-09T21:23:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1110.1556", "slug": "a24c333fe1637e3f", "title": "Jewish Problems", "authors": ["Tanya Khovanova", "Alexey Radul"], "year": 2011, "venue": "arXiv", "url": "https://arxiv.org/abs/1110.1556", "doi": null, "topics": ["Mathematics"], "description": "This is a special collection of problems that were given to select applicants during oral entrance exams to the math department of Moscow State University.", "description_method": "abstract-excerpt", "abstract": "This is a special collection of problems that were given to select applicants during oral entrance exams to the math department of Moscow State University. These problems were designed to prevent Jews and other undesirables from getting a passing grade. Among problems that were used by the department to blackball unwanted candidate students, these problems are distinguished by having a simple solution that is difficult to find. Using problems with a simple solution protected the administration from extra complaints and appeals. This collection therefore has mathematical as well as historical value.", "score": 0.0, "mention_count": 22, "user_count": 22, "months": ["2011-10", "2012-11", "2013-03", "2013-11", "2017-11", "2018-06", "2018-09", "2019-10", "2020-07", "2020-08", "2020-11", "2021-11", "2022-06", "2022-08", "2024-01", "2024-05", "2024-10", "2025-11", "2026-05"], "last_mentioned": "2026-05-28T19:40:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2203.02155", "slug": "81125c1bcbaa0fea", "title": "Training language models to follow instructions with human feedback", "authors": ["Long Ouyang", "Jeff Wu", "Xu Jiang", "Diogo Almeida", "Carroll L. Wainwright", "Pamela Mishkin", "Chong Zhang", "Sandhini Agarwal", "Katarina Slama", "Alex Ray", "John Schulman", "Jacob Hilton", "Fraser Kelton", "Luke Miller", "Maddie Simens", "Amanda Askell", "Peter Welinder", "Paul Christiano", "Jan Leike", "Ryan Lowe"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2203.02155", "doi": null, "topics": ["Language models", "Machine learning", "Statistics & probability"], "description": "Making language models bigger does not inherently make them better at following a user's intent.", "description_method": "abstract-excerpt", "abstract": "Making language models bigger does not inherently make them better at following a user's intent. For example, large language models can generate outputs that are untruthful, toxic, or simply not helpful to the user. In other words, these models are not aligned with their users. In this paper, we show an avenue for aligning language models with user intent on a wide range of tasks by fine-tuning with human feedback. Starting with a set of labeler-written prompts and prompts submitted through the OpenAI API, we collect a dataset of labeler demonstrations of the desired model behavior, which we use to fine-tune GPT-3 using supervised learning. We then collect a dataset of rankings of model outputs, which we use to further fine-tune this supervised model using reinforcement learning from human feedback. We call the resulting models InstructGPT. In human evaluations on our prompt distribution, outputs from the 1.3B parameter InstructGPT model are preferred to outputs from the 175B GPT-3, despite having 100x fewer parameters. Moreover, InstructGPT models show improvements in truthfulness and reductions in toxic output generation while having minimal performance regressions on public NLP datasets. Even though InstructGPT still makes simple mistakes, our results show that fine-tuning with human feedback is a promising direction for aligning language models with human intent.", "score": 0.0, "mention_count": 22, "user_count": 19, "months": ["2022-10", "2022-12", "2023-01", "2023-02", "2023-03", "2023-05", "2023-06", "2023-07", "2023-10", "2024-06", "2025-08", "2026-06"], "last_mentioned": "2026-06-06T01:25:39Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1508.06576", "slug": "f25582825f21be3e", "title": "A Neural Algorithm of Artistic Style", "authors": ["Leon A. Gatys", "Alexander S. Ecker", "Matthias Bethge"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1508.06576", "doi": null, "topics": ["Algorithms", "Machine learning", "Psychology"], "description": "In fine art, especially painting, humans have mastered the skill to create unique visual experiences through composing a complex interplay between the content and style of an image.", "description_method": "abstract-excerpt", "abstract": "In fine art, especially painting, humans have mastered the skill to create unique visual experiences through composing a complex interplay between the content and style of an image. Thus far the algorithmic basis of this process is unknown and there exists no artificial system with similar capabilities. However, in other key areas of visual perception such as object and face recognition near-human performance was recently demonstrated by a class of biologically inspired vision models called Deep Neural Networks. Here we introduce an artificial system based on a Deep Neural Network that creates artistic images of high perceptual quality. The system uses neural representations to separate and recombine content and style of arbitrary images, providing a neural algorithm for the creation of artistic images. Moreover, in light of the striking similarities between performance-optimised artificial neural networks and biological vision, our work offers a path forward to an algorithmic understanding of how humans create and perceive artistic imagery.", "score": 0.0, "mention_count": 21, "user_count": 19, "months": ["2015-08", "2015-09", "2015-11", "2015-12", "2016-02", "2016-03", "2016-04", "2016-08", "2016-09", "2016-10", "2016-12", "2017-05", "2017-10", "2017-11", "2018-02", "2018-07", "2022-06", "2023-04"], "last_mentioned": "2023-04-29T21:52:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.10559", "slug": "9c7e543a5fec5b70", "title": "Why Can GPT Learn In-Context? Language Models Implicitly Perform Gradient Descent as Meta-Optimizers", "authors": ["Damai Dai", "Yutao Sun", "Li Dong", "Yaru Hao", "Shuming Ma", "Zhifang Sui", "Furu Wei"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.10559", "doi": null, "topics": ["Language models", "Machine learning", "Psychology"], "description": "Large pretrained language models have shown surprising in-context learning (ICL) ability.", "description_method": "abstract-excerpt", "abstract": "Large pretrained language models have shown surprising in-context learning (ICL) ability. With a few demonstration input-label pairs, they can predict the label for an unseen input without parameter updates. Despite the great success in performance, its working mechanism still remains an open question. In this paper, we explain language models as meta-optimizers and understand in-context learning as implicit finetuning. Theoretically, we figure out that Transformer attention has a dual form of gradient descent. On top of it, we understand ICL as follows: GPT first produces meta-gradients according to the demonstration examples, and then these meta-gradients are applied to the original GPT to build an ICL model. We comprehensively compare the behaviors of in-context learning and explicit finetuning on real tasks to provide empirical evidence that supports our understanding. Experimental results show that in-context learning behaves similarly to explicit finetuning from multiple perspectives. Inspired by the dual form between Transformer attention and gradient descent, we design a momentum-based attention by analogy with gradient descent with momentum. The improved performance over vanilla attention further supports our understanding from another perspective, and more importantly, shows the potential to utilize our understanding for future model design. The code is available at \\url{https://aka.ms/icl}.", "score": 0.0, "mention_count": 21, "user_count": 11, "months": ["2023-01", "2023-02", "2023-03", "2023-04", "2023-05", "2024-07"], "last_mentioned": "2024-07-07T21:25:57Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2309.10668", "slug": "1d8118ac982663b0", "title": "Language Modeling Is Compression", "authors": ["Grégoire Delétang", "Anian Ruoss", "Paul-Ambroise Duquenne", "Elliot Catt", "Tim Genewein", "Christopher Mattern", "Jordi Grau-Moya", "Li Kevin Wenliang", "Matthew Aitchison", "Laurent Orseau", "Marcus Hutter", "Joel Veness"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2309.10668", "doi": null, "topics": ["Language models", "Machine learning"], "description": "It has long been established that predictive models can be transformed into lossless compressors and vice versa.", "description_method": "abstract-excerpt", "abstract": "It has long been established that predictive models can be transformed into lossless compressors and vice versa. Incidentally, in recent years, the machine learning community has focused on training increasingly large and powerful self-supervised (language) models. Since these large language models exhibit impressive predictive capabilities, they are well-positioned to be strong compressors. In this work, we advocate for viewing the prediction problem through the lens of compression and evaluate the compression capabilities of large (foundation) models. We show that large language models are powerful general-purpose predictors and that the compression viewpoint provides novel insights into scaling laws, tokenization, and in-context learning. For example, Chinchilla 70B, while trained primarily on text, compresses ImageNet patches to 43.4% and LibriSpeech samples to 16.4% of their raw size, beating domain-specific compressors like PNG (58.5%) or FLAC (30.3%), respectively. Finally, we show that the prediction-compression equivalence allows us to use any compressor (like gzip) to build a conditional generative model.", "score": 0.0, "mention_count": 21, "user_count": 21, "months": ["2023-09", "2023-10", "2023-12", "2024-01", "2024-03", "2024-04", "2024-07", "2024-08", "2025-05", "2025-06", "2025-07", "2025-10", "2026-01", "2026-02", "2026-06"], "last_mentioned": "2026-06-05T00:24:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2601.02671", "slug": "231aa8b73779147d", "title": "Extracting books from production language models", "authors": ["Ahmed Ahmed", "A. Feder Cooper", "Sanmi Koyejo", "Percy Liang"], "year": 2026, "venue": "arXiv", "url": "https://arxiv.org/abs/2601.02671", "doi": null, "topics": ["Language models"], "description": "Many unresolved legal questions over LLMs and copyright center on memorization: whether specific training data have been encoded in the model's weights during training, and whether those memorized data can be extracted in the model's outputs.", "description_method": "abstract-excerpt", "abstract": "Many unresolved legal questions over LLMs and copyright center on memorization: whether specific training data have been encoded in the model's weights during training, and whether those memorized data can be extracted in the model's outputs. While many believe that LLMs do not memorize much of their training data, recent work shows that substantial amounts of copyrighted text can be extracted from open-weight models. However, it remains an open question if similar extraction is feasible for production LLMs, given the safety measures these systems implement. We investigate this question using a two-phase procedure: (1) an initial probe to test for extraction feasibility, which sometimes uses a Best-of-N (BoN) jailbreak, followed by (2) iterative continuation prompts to attempt to extract the book. We evaluate our procedure on four production LLMs -- Claude 3.7 Sonnet, GPT-4.1, Gemini 2.5 Pro, and Grok 3 -- and we measure extraction success with a score computed from a block-based approximation of longest common substring (nv-recall). With different per-LLM experimental configurations, we were able to extract varying amounts of text. For the Phase 1 probe, it was unnecessary to jailbreak Gemini 2.5 Pro and Grok 3 to extract text (e.g, nv-recall of 76.8% and 70.3%, respectively, for Harry Potter and the Sorcerer's Stone), while it was necessary for Claude 3.7 Sonnet and GPT-4.1. In some cases, jailbroken Claude 3.7 Sonnet outputs entire books near-verbatim (e.g., nv-recall=95.8%). GPT-4.1 requires significantly more BoN attempts (e.g., 20X), and eventually refuses to continue (e.g., nv-recall=4.0%). Taken together, our work highlights that, even with model- and system-level safeguards, extraction of (in-copyright) training data remains a risk for production LLMs.", "score": 0.0, "mention_count": 21, "user_count": 13, "months": ["2026-01", "2026-02", "2026-03", "2026-05", "2026-06", "2026-07"], "last_mentioned": "2026-07-23T17:14:33Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1810.04805", "slug": "955cabf3213dc361", "title": "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding", "authors": ["Jacob Devlin", "Ming-Wei Chang", "Kenton Lee", "Kristina Toutanova"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1810.04805", "doi": null, "topics": ["Language models"], "description": "We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers.", "description_method": "abstract-excerpt", "abstract": "We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models, BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers. As a result, the pre-trained BERT model can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks, such as question answering and language inference, without substantial task-specific architecture modifications. BERT is conceptually simple and empirically powerful. It obtains new state-of-the-art results on eleven natural language processing tasks, including pushing the GLUE score to 80.5% (7.7% point absolute improvement), MultiNLI accuracy to 86.7% (4.6% absolute improvement), SQuAD v1.1 question answering Test F1 to 93.2 (1.5 point absolute improvement) and SQuAD v2.0 Test F1 to 83.1 (5.1 point absolute improvement).", "score": 0.0, "mention_count": 20, "user_count": 20, "months": ["2019-02", "2019-03", "2019-08", "2020-01", "2021-01", "2022-11", "2022-12", "2023-02", "2023-03", "2023-06", "2023-07", "2023-09", "2023-10", "2024-02", "2026-06"], "last_mentioned": "2026-06-01T22:07:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.14975", "slug": "468995ebfb5c9372", "title": "Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedback", "authors": ["Katherine Tian", "Eric Mitchell", "Allan Zhou", "Archit Sharma", "Rafael Rafailov", "Huaxiu Yao", "Chelsea Finn", "Christopher D. Manning"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.14975", "doi": null, "topics": ["Language models", "Machine learning"], "description": "A trustworthy real-world prediction system should produce well-calibrated confidence scores; that is, its confidence in an answer should be indicative of the likelihood that the answer is correct, enabling deferral to an expert in cases of low-confidence predictions.", "description_method": "abstract-excerpt", "abstract": "A trustworthy real-world prediction system should produce well-calibrated confidence scores; that is, its confidence in an answer should be indicative of the likelihood that the answer is correct, enabling deferral to an expert in cases of low-confidence predictions. Recent studies have shown that unsupervised pre-training produces large language models (LMs) whose conditional probabilities are remarkably well-calibrated. However, the most widely-used LMs are fine-tuned with reinforcement learning from human feedback (RLHF-LMs), and some studies have suggested that RLHF-LMs produce conditional probabilities that are very poorly calibrated. In light of this perceived weakness, we conduct a broad evaluation of methods for extracting confidence scores from RLHF-LMs. For RLHF-LMs such as ChatGPT, GPT-4, and Claude, we find that verbalized confidences emitted as output tokens are typically better-calibrated than the model's conditional probabilities on the TriviaQA, SciQ, and TruthfulQA benchmarks, often reducing the expected calibration error by a relative 50%.", "score": 0.0, "mention_count": 20, "user_count": 4, "months": ["2023-09", "2023-10", "2023-11", "2023-12", "2024-09", "2024-10", "2026-01"], "last_mentioned": "2026-01-19T04:38:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41380-022-01661-0", "slug": "cd570706bad66e34", "title": "The serotonin theory of depression: a systematic umbrella review of the evidence", "authors": ["Joanna Moncrieff", "Ruth E. Cooper", "Tom Stockmann", "Simone Amendola", "Michael P. Hengartner", "Mark A. Horowitz"], "year": 2022, "venue": "Molecular Psychiatry", "url": "https://doi.org/10.1038/s41380-022-01661-0", "doi": "10.1038/s41380-022-01661-0", "topics": ["Statistics & probability", "Economics"], "description": "Abstract The serotonin hypothesis of depression is still influential.", "description_method": "abstract-excerpt", "abstract": "Abstract The serotonin hypothesis of depression is still influential. We aimed to synthesise and evaluate evidence on whether depression is associated with lowered serotonin concentration or activity in a systematic umbrella review of the principal relevant areas of research. PubMed, EMBASE and PsycINFO were searched using terms appropriate to each area of research, from their inception until December 2020. Systematic reviews, meta-analyses and large data-set analyses in the following areas were identified: serotonin and serotonin metabolite, 5-HIAA, concentrations in body fluids; serotonin 5-HT 1A receptor binding; serotonin transporter (SERT) levels measured by imaging or at post-mortem; tryptophan depletion studies; SERT gene associations and SERT gene-environment interactions. Studies of depression associated with physical conditions and specific subtypes of depression (e.g. bipolar depression) were excluded. Two independent reviewers extracted the data and assessed the quality of included studies using the AMSTAR-2, an adapted AMSTAR-2, or the STREGA for a large genetic study. The certainty of study results was assessed using a modified version of the GRADE. We did not synthesise results of individual meta-analyses because they included overlapping studies. The review was registered with PROSPERO (CRD42020207203). 17 studies were included: 12 systematic reviews and meta-analyses, 1 collaborative meta-analysis, 1 meta-analysis of large cohort studies, 1 systematic review and narrative synthesis, 1 genetic association study and 1 umbrella review. Quality of reviews was variable with some genetic studies of high quality. Two meta-analyses of overlapping studies examining the serotonin metabolite, 5-HIAA, showed no association with depression (largest n = 1002). One meta-analysis of cohort studies of plasma serotonin showed no relationship with depression, and evidence that lowered serotonin concentration was associated with antidepressant use ( n = 1869). Two meta-analyses of overlapping studies examining the 5-HT 1A receptor (largest n = 561), and three meta-analyses of overlapping studies examining SERT binding (largest n = 1845) showed weak and inconsistent evidence of reduced binding in some areas, which would be consistent with increased synaptic availability of serotonin in people with depression, if this was the original, causal abnormaly. However, effects of prior antidepressant use were not reliably excluded. One meta-analysis of tryptophan depletion studies found no effect in most healthy volunteers ( n = 566), but weak evidence of an effect in those with a family history of depression ( n = 75). Another systematic review ( n = 342) and a sample of ten subsequent studies ( n = 407) found no effect in volunteers. No systematic review of tryptophan depletion studies has been performed since 2007. The two largest and highest quality studies of the SERT gene, one genetic association study ( n = 115,257) and one collaborative meta-analysis ( n = 43,165), revealed no evidence of an association with depression, or of an interaction between genotype, stress and depression. The main areas of serotonin research provide no consistent evidence of there being an association between serotonin and depression, and no support for the hypothesis that depression is caused by lowered serotonin activity or concentrations. Some evidence was consistent with the possibility that long-term antidepressant use reduces serotonin concentration.", "score": 0.0, "mention_count": 20, "user_count": 19, "months": ["2022-07", "2022-08", "2022-10", "2023-04", "2023-05", "2023-06", "2023-08", "2024-09", "2024-10", "2025-04", "2026-01"], "last_mentioned": "2026-01-09T22:34:43Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1904.09828", "slug": "373d85eb7ef5afb2", "title": "Magic: The Gathering is Turing Complete", "authors": ["Alex Churchill", "Stella Biderman", "Austin Herrick"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1904.09828", "doi": null, "topics": ["Statistics & probability"], "description": "$\\textit{Magic: The Gathering}$ is a popular and famously complicated trading card game about magical combat.", "description_method": "abstract-excerpt", "abstract": "$\\textit{Magic: The Gathering}$ is a popular and famously complicated trading card game about magical combat. In this paper we show that optimal play in real-world $\\textit{Magic}$ is at least as hard as the Halting Problem, solving a problem that has been open for a decade. To do this, we present a methodology for embedding an arbitrary Turing machine into a game of $\\textit{Magic}$ such that the first player is guaranteed to win the game if and only if the Turing machine halts. Our result applies to how real $\\textit{Magic}$ is played, can be achieved using standard-size tournament-legal decks, and does not rely on stochasticity or hidden information. Our result is also highly unusual in that all moves of both players are forced in the construction. This shows that even recognising who will win a game in which neither player has a non-trivial decision to make for the rest of the game is undecidable. We conclude with a discussion of the implications for a unified computational theory of games and remarks about the playability of such a board in a tournament setting.", "score": 0.0, "mention_count": 19, "user_count": 18, "months": ["2019-05", "2019-08", "2019-11", "2020-06", "2021-11", "2022-06", "2022-12", "2023-06", "2023-12", "2024-09", "2025-01", "2026-02", "2026-06"], "last_mentioned": "2026-06-12T03:00:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2206.08896", "slug": "0546429381f41d23", "title": "Evolution through Large Models", "authors": ["Joel Lehman", "Jonathan Gordon", "Shawn Jain", "Kamal Ndousse", "Cathy Yeh", "Kenneth O. Stanley"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2206.08896", "doi": null, "topics": ["Language models", "Machine learning", "Robotics"], "description": "This paper pursues the insight that large language models (LLMs) trained to generate code can vastly improve the effectiveness of mutation operators applied to programs in genetic programming (GP).", "description_method": "abstract-excerpt", "abstract": "This paper pursues the insight that large language models (LLMs) trained to generate code can vastly improve the effectiveness of mutation operators applied to programs in genetic programming (GP). Because such LLMs benefit from training data that includes sequential changes and modifications, they can approximate likely changes that humans would make. To highlight the breadth of implications of such evolution through large models (ELM), in the main experiment ELM combined with MAP-Elites generates hundreds of thousands of functional examples of Python programs that output working ambulating robots in the Sodarace domain, which the original LLM had never seen in pre-training. These examples then help to bootstrap training a new conditional language model that can output the right walker for a particular terrain. The ability to bootstrap new models that can output appropriate artifacts for a given context in a domain where zero training data was previously available carries implications for open-endedness, deep learning, and reinforcement learning. These implications are explored here in depth in the hope of inspiring new directions of research now opened up by ELM.", "score": 0.0, "mention_count": 19, "user_count": 3, "months": ["2022-12", "2023-01", "2023-02", "2023-03", "2023-05", "2023-06", "2024-01", "2024-03", "2024-04", "2024-06", "2025-04", "2025-05"], "last_mentioned": "2025-05-15T00:19:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.04494", "slug": "2ffaca383a802e70", "title": "Amortized Planning with Large-Scale Transformers: A Case Study on Chess", "authors": ["Anian Ruoss", "Grégoire Delétang", "Sourabh Medapati", "Jordi Grau-Moya", "Li Kevin Wenliang", "Elliot Catt", "John Reid", "Cannada A. Lewis", "Joel Veness", "Tim Genewein"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.04494", "doi": null, "topics": ["Language models", "Machine learning", "Psychology"], "description": "This paper uses chess, a landmark planning problem in AI, to assess transformers' performance on a planning task where memorization is futile $\\unicode{x2013}$ even at a large scale.", "description_method": "abstract-excerpt", "abstract": "This paper uses chess, a landmark planning problem in AI, to assess transformers' performance on a planning task where memorization is futile $\\unicode{x2013}$ even at a large scale. To this end, we release ChessBench, a large-scale benchmark dataset of 10 million chess games with legal move and value annotations (15 billion data points) provided by Stockfish 16, the state-of-the-art chess engine. We train transformers with up to 270 million parameters on ChessBench via supervised learning and perform extensive ablations to assess the impact of dataset size, model size, architecture type, and different prediction targets (state-values, action-values, and behavioral cloning). Our largest models learn to predict action-values for novel boards quite accurately, implying highly non-trivial generalization. Despite performing no explicit search, our resulting chess policy solves challenging chess puzzles and achieves a surprisingly strong Lichess blitz Elo of 2895 against humans (grandmaster level). We also compare to Leela Chess Zero and AlphaZero (trained without supervision via self-play) with and without search. We show that, although a remarkably good approximation of Stockfish's search-based algorithm can be distilled into large-scale transformers via supervised learning, perfect distillation is still beyond reach, thus making ChessBench well-suited for future research.", "score": 0.0, "mention_count": 19, "user_count": 13, "months": ["2024-02", "2024-04", "2024-06", "2024-08", "2024-09", "2024-10", "2024-11", "2024-12", "2025-03", "2025-04", "2026-03"], "last_mentioned": "2026-03-09T17:47:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0908.1803", "slug": "5e6571d32e28335e", "title": "Are Black Hole Starships Possible", "authors": ["Louis Crane", "Shawn Westmoreland"], "year": 2009, "venue": "arXiv", "url": "https://arxiv.org/abs/0908.1803", "doi": null, "topics": ["Physics"], "description": "We investigate whether it is physically possible to build starships or power sources using the Hawking radiation of an artificial black hole as a power source.", "description_method": "abstract-excerpt", "abstract": "We investigate whether it is physically possible to build starships or power sources using the Hawking radiation of an artificial black hole as a power source. The proposal seems to be at the edge of possibility, but quantum gravity effects could change the picture.", "score": 0.0, "mention_count": 18, "user_count": 10, "months": ["2011-06", "2011-12", "2014-01", "2014-03", "2015-04", "2016-08", "2019-03", "2019-09", "2019-11", "2020-01", "2020-02", "2020-05", "2020-11", "2021-08", "2022-05", "2025-11"], "last_mentioned": "2025-11-27T22:56:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2205.14334", "slug": "c9be57471d039c4c", "title": "Teaching Models to Express Their Uncertainty in Words", "authors": ["Stephanie Lin", "Jacob Hilton", "Owain Evans"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2205.14334", "doi": null, "topics": ["Language models", "Statistics & probability"], "description": "We show that a GPT-3 model can learn to express uncertainty about its own answers in natural language -- without use of model logits.", "description_method": "abstract-excerpt", "abstract": "We show that a GPT-3 model can learn to express uncertainty about its own answers in natural language -- without use of model logits. When given a question, the model generates both an answer and a level of confidence (e.g. \"90% confidence\" or \"high confidence\"). These levels map to probabilities that are well calibrated. The model also remains moderately calibrated under distribution shift, and is sensitive to uncertainty in its own answers, rather than imitating human examples. To our knowledge, this is the first time a model has been shown to express calibrated uncertainty about its own answers in natural language. For testing calibration, we introduce the CalibratedMath suite of tasks. We compare the calibration of uncertainty expressed in words (\"verbalized probability\") to uncertainty extracted from model logits. Both kinds of uncertainty are capable of generalizing calibration under distribution shift. We also provide evidence that GPT-3's ability to generalize calibration depends on pre-trained latent representations that correlate with epistemic uncertainty over its answers.", "score": 0.0, "mention_count": 18, "user_count": 3, "months": ["2022-05", "2023-09", "2023-10", "2023-11", "2024-09", "2024-10"], "last_mentioned": "2024-10-30T15:20:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2210.03629", "slug": "e504ba1ba9c3fffc", "title": "ReAct: Synergizing Reasoning and Acting in Language Models", "authors": ["Shunyu Yao", "Jeffrey Zhao", "Dian Yu", "Nan Du", "Izhak Shafran", "Karthik Narasimhan", "Yuan Cao"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2210.03629", "doi": null, "topics": ["Psychology", "Language models", "Machine learning"], "description": "While large language models (LLMs) have demonstrated impressive capabilities across tasks in language understanding and interactive decision making, their abilities for reasoning (e.g.", "description_method": "abstract-excerpt", "abstract": "While large language models (LLMs) have demonstrated impressive capabilities across tasks in language understanding and interactive decision making, their abilities for reasoning (e.g. chain-of-thought prompting) and acting (e.g. action plan generation) have primarily been studied as separate topics. In this paper, we explore the use of LLMs to generate both reasoning traces and task-specific actions in an interleaved manner, allowing for greater synergy between the two: reasoning traces help the model induce, track, and update action plans as well as handle exceptions, while actions allow it to interface with external sources, such as knowledge bases or environments, to gather additional information. We apply our approach, named ReAct, to a diverse set of language and decision making tasks and demonstrate its effectiveness over state-of-the-art baselines, as well as improved human interpretability and trustworthiness over methods without reasoning or acting components. Concretely, on question answering (HotpotQA) and fact verification (Fever), ReAct overcomes issues of hallucination and error propagation prevalent in chain-of-thought reasoning by interacting with a simple Wikipedia API, and generates human-like task-solving trajectories that are more interpretable than baselines without reasoning traces. On two interactive decision making benchmarks (ALFWorld and WebShop), ReAct outperforms imitation and reinforcement learning methods by an absolute success rate of 34% and 10% respectively, while being prompted with only one or two in-context examples. Project site with code: https://react-lm.github.io", "score": 0.0, "mention_count": 18, "user_count": 15, "months": ["2023-03", "2023-04", "2023-05", "2023-06", "2023-07", "2023-09", "2024-08", "2025-04", "2025-07"], "last_mentioned": "2025-07-27T14:22:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2210.17323", "slug": "0d1e60374ee440ce", "title": "GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers", "authors": ["Elias Frantar", "Saleh Ashkboos", "Torsten Hoefler", "Dan Alistarh"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2210.17323", "doi": null, "topics": ["Language models", "Human–computer interaction", "Algorithms"], "description": "Generative Pre-trained Transformer models, known as GPT or OPT, set themselves apart through breakthrough performance across complex language modelling tasks, but also by their extremely high computational and storage costs.", "description_method": "abstract-excerpt", "abstract": "Generative Pre-trained Transformer models, known as GPT or OPT, set themselves apart through breakthrough performance across complex language modelling tasks, but also by their extremely high computational and storage costs. Specifically, due to their massive size, even inference for large, highly-accurate GPT models may require multiple performant GPUs, which limits the usability of such models. While there is emerging work on relieving this pressure via model compression, the applicability and performance of existing compression techniques is limited by the scale and complexity of GPT models. In this paper, we address this challenge, and propose GPTQ, a new one-shot weight quantization method based on approximate second-order information, that is both highly-accurate and highly-efficient. Specifically, GPTQ can quantize GPT models with 175 billion parameters in approximately four GPU hours, reducing the bitwidth down to 3 or 4 bits per weight, with negligible accuracy degradation relative to the uncompressed baseline. Our method more than doubles the compression gains relative to previously-proposed one-shot quantization methods, preserving accuracy, allowing us for the first time to execute an 175 billion-parameter model inside a single GPU for generative inference. Moreover, we also show that our method can still provide reasonable accuracy in the extreme quantization regime, in which weights are quantized to 2-bit or even ternary quantization levels. We show experimentally that these improvements can be leveraged for end-to-end inference speedups over FP16, of around 3.25x when using high-end GPUs (NVIDIA A100) and 4.5x when using more cost-effective ones (NVIDIA A6000). The implementation is available at https://github.com/IST-DASLab/gptq.", "score": 0.0, "mention_count": 18, "user_count": 14, "months": ["2023-03", "2023-05", "2023-09", "2023-12"], "last_mentioned": "2023-12-16T04:36:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2312.00752", "slug": "d04b11459a9ebe42", "title": "Mamba: Linear-Time Sequence Modeling with Selective State Spaces", "authors": ["Albert Gu", "Tri Dao"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2312.00752", "doi": null, "topics": ["Language models", "Machine learning", "Biology & medicine"], "description": "Foundation models, now powering most of the exciting applications in deep learning, are almost universally based on the Transformer architecture and its core attention module.", "description_method": "abstract-excerpt", "abstract": "Foundation models, now powering most of the exciting applications in deep learning, are almost universally based on the Transformer architecture and its core attention module. Many subquadratic-time architectures such as linear attention, gated convolution and recurrent models, and structured state space models (SSMs) have been developed to address Transformers' computational inefficiency on long sequences, but they have not performed as well as attention on important modalities such as language. We identify that a key weakness of such models is their inability to perform content-based reasoning, and make several improvements. First, simply letting the SSM parameters be functions of the input addresses their weakness with discrete modalities, allowing the model to selectively propagate or forget information along the sequence length dimension depending on the current token. Second, even though this change prevents the use of efficient convolutions, we design a hardware-aware parallel algorithm in recurrent mode. We integrate these selective SSMs into a simplified end-to-end neural network architecture without attention or even MLP blocks (Mamba). Mamba enjoys fast inference (5$\\times$ higher throughput than Transformers) and linear scaling in sequence length, and its performance improves on real data up to million-length sequences. As a general sequence model backbone, Mamba achieves state-of-the-art performance across several modalities such as language, audio, and genomics. On language modeling, our Mamba-3B model outperforms Transformers of the same size and matches Transformers twice its size, both in pretraining and downstream evaluation.", "score": 0.0, "mention_count": 18, "user_count": 18, "months": ["2023-12", "2024-01", "2024-02", "2024-03", "2024-06", "2024-07", "2024-10", "2025-02", "2025-05", "2025-08", "2026-02"], "last_mentioned": "2026-02-04T00:28:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1405.1548", "slug": "4184e054f9389da4", "title": "The Cellular Automaton Interpretation of Quantum Mechanics", "authors": ["Gerard 't Hooft"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1405.1548", "doi": null, "topics": ["Physics", "Mathematics", "Biology & medicine"], "description": "When investigating theories at the tiniest conceivable scales in nature, almost all researchers today revert to the quantum language, accepting the verdict from the Copenhagen doctrine that the only way to describe what is going on will always involve states in Hilbert space, controlled by operator equations.", "description_method": "abstract-excerpt", "abstract": "When investigating theories at the tiniest conceivable scales in nature, almost all researchers today revert to the quantum language, accepting the verdict from the Copenhagen doctrine that the only way to describe what is going on will always involve states in Hilbert space, controlled by operator equations. Returning to classical, that is, non quantum mechanical, descriptions will be forever impossible, unless one accepts some extremely contrived theoretical constructions that may or may not reproduce the quantum mechanical phenomena observed in experiments. Dissatisfied, this author investigated how one can look at things differently. This book is an overview of older material, but also contains many new observations and calculations. Quantum mechanics is looked upon as a tool, not as a theory. Examples are displayed of models that are classical in essence, but can be analysed by the use of quantum techniques, and we argue that even the Standard Model, together with gravitational interactions, might be viewed as a quantum mechanical approach to analyse a system that could be classical at its core. We explain how such thoughts can conceivably be reconciled with Bell's theorem, and how the usual objections voiced against the notion of `superdeterminism' can be overcome, at least in principle. Our proposal would eradicate the collapse problem and the measurement problem. Even the existence of an \"arrow of time\" can perhaps be explained in a more elegant way than usual. Discussions added in v3: the role of the gravitational force, a mathematical physics definition of free will, and an unconventional view on the arrow of time, amongst others.", "score": 0.0, "mention_count": 17, "user_count": 11, "months": ["2014-06", "2017-01", "2017-04", "2017-06", "2018-12", "2019-03", "2019-10", "2020-03", "2020-04", "2020-10", "2022-07"], "last_mentioned": "2022-07-09T12:39:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2005.11710", "slug": "9af1133f377d4da9", "title": "Featherweight Go", "authors": ["Robert Griesemer", "Raymond Hu", "Wen Kokke", "Julien Lange", "Ian Lance Taylor", "Bernardo Toninho", "Philip Wadler", "Nobuko Yoshida"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2005.11710", "doi": null, "topics": ["Other research"], "description": "We describe a design for generics in Go inspired by previous work on Featherweight Java by Igarashi, Pierce, and Wadler.", "description_method": "abstract-excerpt", "abstract": "We describe a design for generics in Go inspired by previous work on Featherweight Java by Igarashi, Pierce, and Wadler. Whereas subtyping in Java is nominal, in Go it is structural, and whereas generics in Java are defined via erasure, in Go we use monomorphisation. Although monomorphisation is widely used, we are one of the first to formalise it. Our design also supports a solution to The Expression Problem.", "score": 0.0, "mention_count": 17, "user_count": 13, "months": ["2020-06", "2020-07", "2020-11", "2021-01", "2021-02", "2021-05", "2021-07", "2021-10", "2021-12", "2022-03", "2022-05", "2023-09", "2024-11", "2025-06", "2026-08"], "last_mentioned": "2026-08-01T15:05:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2107.03374", "slug": "ef846621ba647afd", "title": "Evaluating Large Language Models Trained on Code", "authors": ["Mark Chen", "Jerry Tworek", "Heewoo Jun", "Qiming Yuan", "Henrique Ponde de Oliveira Pinto", "Jared Kaplan", "Harri Edwards", "Yuri Burda", "Nicholas Joseph", "Greg Brockman", "Alex Ray", "Raul Puri", "Gretchen Krueger", "Michael Petrov", "Heidy Khlaaf", "Girish Sastry", "Pamela Mishkin", "Brooke Chan", "Scott Gray", "Nick Ryder", "Mikhail Pavlov", "Alethea Power", "Lukasz Kaiser", "Mohammad Bavarian", "Clemens Winter", "Philippe Tillet", "Felipe Petroski Such", "Dave Cummings", "Matthias Plappert", "Fotios Chantzis", "Elizabeth Barnes", "Ariel Herbert-Voss", "William Hebgen Guss", "Alex Nichol", "Alex Paino", "Nikolas Tezak", "Jie Tang", "Igor Babuschkin", "Suchir Balaji", "Shantanu Jain", "William Saunders", "Christopher Hesse", "Andrew N. Carr", "Jan Leike", "Josh Achiam", "Vedant Misra", "Evan Morikawa", "Alec Radford", "Matthew Knight", "Miles Brundage", "Mira Murati", "Katie Mayer", "Peter Welinder", "Bob McGrew", "Dario Amodei", "Sam McCandlish", "Ilya Sutskever", "Wojciech Zaremba"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2107.03374", "doi": null, "topics": ["Language models", "Security & cryptography", "Economics"], "description": "We introduce Codex, a GPT language model fine-tuned on publicly available code from GitHub, and study its Python code-writing capabilities.", "description_method": "abstract-excerpt", "abstract": "We introduce Codex, a GPT language model fine-tuned on publicly available code from GitHub, and study its Python code-writing capabilities. A distinct production version of Codex powers GitHub Copilot. On HumanEval, a new evaluation set we release to measure functional correctness for synthesizing programs from docstrings, our model solves 28.8% of the problems, while GPT-3 solves 0% and GPT-J solves 11.4%. Furthermore, we find that repeated sampling from the model is a surprisingly effective strategy for producing working solutions to difficult prompts. Using this method, we solve 70.2% of our problems with 100 samples per problem. Careful investigation of our model reveals its limitations, including difficulty with docstrings describing long chains of operations and with binding operations to variables. Finally, we discuss the potential broader impacts of deploying powerful code generation technologies, covering safety, security, and economics.", "score": 0.0, "mention_count": 17, "user_count": 9, "months": ["2021-08", "2021-09", "2021-12", "2022-01", "2022-07", "2022-08", "2022-10", "2023-05", "2023-07", "2023-09", "2023-10", "2024-02", "2025-02", "2025-04"], "last_mentioned": "2025-04-23T19:24:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2405.07987", "slug": "6c7695929f8bbcea", "title": "The Platonic Representation Hypothesis", "authors": ["Minyoung Huh", "Brian Cheung", "Tongzhou Wang", "Phillip Isola"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2405.07987", "doi": null, "topics": ["Statistics & probability", "Machine learning", "Language models"], "description": "We argue that representations in AI models, particularly deep networks, are converging.", "description_method": "abstract-excerpt", "abstract": "We argue that representations in AI models, particularly deep networks, are converging. First, we survey many examples of convergence in the literature: over time and across multiple domains, the ways by which different neural networks represent data are becoming more aligned. Next, we demonstrate convergence across data modalities: as vision models and language models get larger, they measure distance between datapoints in a more and more alike way. We hypothesize that this convergence is driving toward a shared statistical model of reality, akin to Plato's concept of an ideal reality. We term such a representation the platonic representation and discuss several possible selective pressures toward it. Finally, we discuss the implications of these trends, their limitations, and counterexamples to our analysis.", "score": 0.0, "mention_count": 17, "user_count": 15, "months": ["2024-05", "2024-10", "2025-03", "2025-05", "2025-07", "2025-08", "2025-12", "2026-01", "2026-03", "2026-07"], "last_mentioned": "2026-07-07T20:34:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1301.3781", "slug": "bfb95f89e35f56e1", "title": "Efficient Estimation of Word Representations in Vector Space", "authors": ["Tomas Mikolov", "Kai Chen", "Greg Corrado", "Jeffrey Dean"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1301.3781", "doi": null, "topics": ["Machine learning"], "description": "We propose two novel model architectures for computing continuous vector representations of words from very large data sets.", "description_method": "abstract-excerpt", "abstract": "We propose two novel model architectures for computing continuous vector representations of words from very large data sets. The quality of these representations is measured in a word similarity task, and the results are compared to the previously best performing techniques based on different types of neural networks. We observe large improvements in accuracy at much lower computational cost, i.e. it takes less than a day to learn high quality word vectors from a 1.6 billion words data set. Furthermore, we show that these vectors provide state-of-the-art performance on our test set for measuring syntactic and semantic word similarities.", "score": 0.0, "mention_count": 16, "user_count": 13, "months": ["2013-08", "2013-09", "2013-11", "2015-06", "2015-12", "2016-05", "2017-07", "2017-08", "2017-09", "2019-02", "2023-10", "2025-02", "2025-05"], "last_mentioned": "2025-05-29T21:07:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.5602", "slug": "760145e5fced388d", "title": "Playing Atari with Deep Reinforcement Learning", "authors": ["Volodymyr Mnih", "Koray Kavukcuoglu", "David Silver", "Alex Graves", "Ioannis Antonoglou", "Daan Wierstra", "Martin Riedmiller"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.5602", "doi": null, "topics": ["Machine learning", "Computer vision", "Algorithms"], "description": "We present the first deep learning model to successfully learn control policies directly from high-dimensional sensory input using reinforcement learning.", "description_method": "abstract-excerpt", "abstract": "We present the first deep learning model to successfully learn control policies directly from high-dimensional sensory input using reinforcement learning. The model is a convolutional neural network, trained with a variant of Q-learning, whose input is raw pixels and whose output is a value function estimating future rewards. We apply our method to seven Atari 2600 games from the Arcade Learning Environment, with no adjustment of the architecture or learning algorithm. We find that it outperforms all previous approaches on six of the games and surpasses a human expert on three of them.", "score": 0.0, "mention_count": 16, "user_count": 15, "months": ["2014-01", "2014-02", "2014-07", "2014-10", "2014-12", "2015-01", "2015-02", "2016-03", "2016-05", "2016-06", "2016-08", "2017-01", "2017-04", "2024-01"], "last_mentioned": "2024-01-21T01:01:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1409.3540", "slug": "ebc46789202fd01f", "title": "ARC: A compact, high-field, fusion nuclear science facility and demonstration power plant with demountable magnets", "authors": ["B. N. Sorbom", "J. Ball", "T. R. Palmer", "F. J. Mangiarotti", "J. M. Sierchio", "P. Bonoli", "C. Kasten", "D. A. Sutherland", "H. S. Barnard", "C. B. Haakonsen", "J. Goh", "C. Sung", "D. G. Whyte"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1409.3540", "doi": "10.1016/j.fusengdes.2015.07.008", "topics": ["Physics", "Algorithms"], "description": "The affordable, robust, compact (ARC) reactor conceptual design study aims to reduce the size, cost, and complexity of a combined fusion nuclear science facility (FNSF) and demonstration fusion Pilot power plant.", "description_method": "abstract-excerpt", "abstract": "The affordable, robust, compact (ARC) reactor conceptual design study aims to reduce the size, cost, and complexity of a combined fusion nuclear science facility (FNSF) and demonstration fusion Pilot power plant. ARC is a 200-250 MWe tokamak reactor with a major radius of 3.3 m, a minor radius of 1.1 m, and an on-axis magnetic field of 9.2 T. ARC has rare earth barium copper oxide (REBCO) superconducting toroidal field coils, which have joints to enable disassembly. This allows the vacuum vessel to be replaced quickly, mitigating first wall survivability concerns, and permits a single device to test many vacuum vessel designs and divertor materials. The design point has a plasma fusion gain of Q_p~13.6, yet is fully non-inductive, with a modest bootstrap fraction of only ~63%. Thus ARC offers a high power gain with relatively large external control of the current profile. This highly attractive combination is enabled by the ~23 T peak field on coil with newly available REBCO superconductor technology. External current drive is provided by two innovative inboard RF launchers using 25 MW of lower hybrid and 13.6 MW of ion cyclotron fast wave power. The resulting efficient current drive provides a robust, steady state core plasma far from disruptive limits. ARC uses an all-liquid blanket, consisting of low pressure, slowly flowing fluorine lithium beryllium (FLiBe) molten salt. The liquid blanket is low-risk technology and provides effective neutron moderation and shielding, excellent heat removal, and a tritium breeding ratio >= 1.1. The large temperature range over which FLiBe is liquid permits blanket operation at 900 K with single phase fluid cooling and a high-efficiency Brayton cycle, allowing for net electricity generation when operating ARC as a Pilot power plant.", "score": 0.0, "mention_count": 16, "user_count": 9, "months": ["2015-08", "2016-12", "2018-03", "2018-11", "2019-08", "2020-09", "2020-12", "2021-10", "2022-01", "2022-03", "2022-06", "2024-07"], "last_mentioned": "2024-07-06T11:30:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1503.02531", "slug": "3b2608d5aa123711", "title": "Distilling the Knowledge in a Neural Network", "authors": ["Geoffrey Hinton", "Oriol Vinyals", "Jeff Dean"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1503.02531", "doi": null, "topics": ["Machine learning", "Economics", "Algorithms"], "description": "A very simple way to improve the performance of almost any machine learning algorithm is to train many different models on the same data and then to average their predictions.", "description_method": "abstract-excerpt", "abstract": "A very simple way to improve the performance of almost any machine learning algorithm is to train many different models on the same data and then to average their predictions. Unfortunately, making predictions using a whole ensemble of models is cumbersome and may be too computationally expensive to allow deployment to a large number of users, especially if the individual models are large neural nets. Caruana and his collaborators have shown that it is possible to compress the knowledge in an ensemble into a single model which is much easier to deploy and we develop this approach further using a different compression technique. We achieve some surprising results on MNIST and we show that we can significantly improve the acoustic model of a heavily used commercial system by distilling the knowledge in an ensemble of models into a single model. We also introduce a new type of ensemble composed of one or more full models and many specialist models which learn to distinguish fine-grained classes that the full models confuse. Unlike a mixture of experts, these specialist models can be trained rapidly and in parallel.", "score": 0.0, "mention_count": 16, "user_count": 16, "months": ["2015-07", "2016-03", "2016-12", "2017-02", "2017-04", "2017-08", "2017-12", "2018-02", "2020-11", "2022-04", "2023-03", "2024-07", "2025-07", "2026-03", "2026-07"], "last_mentioned": "2026-07-22T20:41:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2303.11366", "slug": "3fb5d6b0a837235b", "title": "Reflexion: Language Agents with Verbal Reinforcement Learning", "authors": ["Noah Shinn", "Federico Cassano", "Edward Berman", "Ashwin Gopinath", "Karthik Narasimhan", "Shunyu Yao"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2303.11366", "doi": null, "topics": ["Machine learning", "Psychology", "Language models"], "description": "Large language models (LLMs) have been increasingly used to interact with external environments (e.g., games, compilers, APIs) as goal-driven agents.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLMs) have been increasingly used to interact with external environments (e.g., games, compilers, APIs) as goal-driven agents. However, it remains challenging for these language agents to quickly and efficiently learn from trial-and-error as traditional reinforcement learning methods require extensive training samples and expensive model fine-tuning. We propose Reflexion, a novel framework to reinforce language agents not by updating weights, but instead through linguistic feedback. Concretely, Reflexion agents verbally reflect on task feedback signals, then maintain their own reflective text in an episodic memory buffer to induce better decision-making in subsequent trials. Reflexion is flexible enough to incorporate various types (scalar values or free-form language) and sources (external or internally simulated) of feedback signals, and obtains significant improvements over a baseline agent across diverse tasks (sequential decision-making, coding, language reasoning). For example, Reflexion achieves a 91% pass@1 accuracy on the HumanEval coding benchmark, surpassing the previous state-of-the-art GPT-4 that achieves 80%. We also conduct ablation and analysis studies using different feedback signals, feedback incorporation methods, and agent types, and provide insights into how they affect performance.", "score": 0.0, "mention_count": 16, "user_count": 12, "months": ["2023-03", "2023-04", "2023-05", "2023-06", "2023-07", "2023-08", "2023-09", "2024-10"], "last_mentioned": "2024-10-03T23:32:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2307.03172", "slug": "9dbf664b6954efbc", "title": "Lost in the Middle: How Language Models Use Long Contexts", "authors": ["Nelson F. Liu", "Kevin Lin", "John Hewitt", "Ashwin Paranjape", "Michele Bevilacqua", "Fabio Petroni", "Percy Liang"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2307.03172", "doi": null, "topics": ["Language models"], "description": "While recent language models have the ability to take long contexts as input, relatively little is known about how well they use longer context.", "description_method": "abstract-excerpt", "abstract": "While recent language models have the ability to take long contexts as input, relatively little is known about how well they use longer context. We analyze the performance of language models on two tasks that require identifying relevant information in their input contexts: multi-document question answering and key-value retrieval. We find that performance can degrade significantly when changing the position of relevant information, indicating that current language models do not robustly make use of information in long input contexts. In particular, we observe that performance is often highest when relevant information occurs at the beginning or end of the input context, and significantly degrades when models must access relevant information in the middle of long contexts, even for explicitly long-context models. Our analysis provides a better understanding of how language models use their input context and provides new evaluation protocols for future long-context language models.", "score": 0.0, "mention_count": 16, "user_count": 15, "months": ["2023-07", "2023-08", "2023-09", "2023-10", "2023-11", "2023-12", "2024-04", "2024-07", "2025-08", "2026-07"], "last_mentioned": "2026-07-29T13:56:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.06824", "slug": "acae8c8efc656dbf", "title": "The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets", "authors": ["Samuel Marks", "Max Tegmark"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.06824", "doi": null, "topics": ["Language models", "Mathematics", "Human–computer interaction"], "description": "Large Language Models (LLMs) have impressive capabilities, but are prone to outputting falsehoods.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) have impressive capabilities, but are prone to outputting falsehoods. Recent work has developed techniques for inferring whether a LLM is telling the truth by training probes on the LLM's internal activations. However, this line of work is controversial, with some authors pointing out failures of these probes to generalize in basic ways, among other conceptual issues. In this work, we use high-quality datasets of simple true/false statements to study in detail the structure of LLM representations of truth, drawing on three lines of evidence: 1. Visualizations of LLM true/false statement representations, which reveal clear linear structure. 2. Transfer experiments in which probes trained on one dataset generalize to different datasets. 3. Causal evidence obtained by surgically intervening in a LLM's forward pass, causing it to treat false statements as true and vice versa. Overall, we present evidence that at sufficient scale, LLMs linearly represent the truth or falsehood of factual statements. We also show that simple difference-in-mean probes generalize as well as other probing techniques while identifying directions which are more causally implicated in model outputs.", "score": 0.0, "mention_count": 16, "user_count": 6, "months": ["2023-10", "2023-11", "2024-05", "2024-09", "2024-10", "2025-08"], "last_mentioned": "2025-08-03T19:21:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.17764", "slug": "f300ded3bd442799", "title": "The Era of 1-bit LLMs: All Large Language Models are in 1.58 Bits", "authors": ["Shuming Ma", "Hongyu Wang", "Lingxiao Ma", "Lei Wang", "Wenhui Wang", "Shaohan Huang", "Li Dong", "Ruiping Wang", "Jilong Xue", "Furu Wei"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.17764", "doi": null, "topics": ["Language models"], "description": "Recent research, such as BitNet, is paving the way for a new era of 1-bit Large Language Models (LLMs).", "description_method": "abstract-excerpt", "abstract": "Recent research, such as BitNet, is paving the way for a new era of 1-bit Large Language Models (LLMs). In this work, we introduce a 1-bit LLM variant, namely BitNet b1.58, in which every single parameter (or weight) of the LLM is ternary {-1, 0, 1}. It matches the full-precision (i.e., FP16 or BF16) Transformer LLM with the same model size and training tokens in terms of both perplexity and end-task performance, while being significantly more cost-effective in terms of latency, memory, throughput, and energy consumption. More profoundly, the 1.58-bit LLM defines a new scaling law and recipe for training new generations of LLMs that are both high-performance and cost-effective. Furthermore, it enables a new computation paradigm and opens the door for designing specific hardware optimized for 1-bit LLMs.", "score": 0.0, "mention_count": 16, "user_count": 16, "months": ["2024-02", "2024-03", "2024-04", "2024-05", "2024-06", "2025-01", "2025-04", "2025-10", "2026-01", "2026-05", "2026-07"], "last_mentioned": "2026-07-21T12:25:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2412.06769", "slug": "68c71dc1a042dedd", "title": "Training Large Language Models to Reason in a Continuous Latent Space", "authors": ["Shibo Hao", "Sainbayar Sukhbaatar", "DiJia Su", "Xian Li", "Zhiting Hu", "Jason Weston", "Yuandong Tian"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2412.06769", "doi": null, "topics": ["Language models"], "description": "Large language models (LLMs) are typically constrained to reason in the language space, where they express the reasoning process through a chain-of-thought (CoT) to solve complex problems.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLMs) are typically constrained to reason in the language space, where they express the reasoning process through a chain-of-thought (CoT) to solve complex problems. However, the language space may not always be optimal for reasoning. Most word tokens primarily ensure textual coherence and are not essential for reasoning, while some critical tokens require complex planning and pose challenges to LLMs. To explore the potential of reasoning beyond language, we introduce a new paradigm called Coconut (Chain of Continuous Thought). Coconut utilizes the last hidden state of the LLM as a representation of the reasoning state, termed \"continuous thought.\" Instead of decoding this state into words, we feed it back to the model as the next input embedding directly in the continuous space. This latent reasoning paradigm enables an advanced reasoning pattern, where continuous thoughts can encode multiple alternative next steps, allowing the model to perform a breadth-first search (BFS) rather than committing prematurely to a single deterministic path as in CoT. Coconut outperforms CoT on logical reasoning tasks that require substantial search during planning and achieves a better trade-off between accuracy and efficiency.", "score": 0.0, "mention_count": 16, "user_count": 16, "months": ["2024-12", "2025-01", "2025-02", "2025-04", "2025-05", "2025-07", "2025-12", "2026-04", "2026-06", "2026-08"], "last_mentioned": "2026-08-17T09:25:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2503.23674", "slug": "946c9f236339c214", "title": "Large Language Models Pass the Turing Test", "authors": ["Cameron R. Jones", "Benjamin K. Bergen"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2503.23674", "doi": null, "topics": ["Language models", "Psychology", "Economics"], "description": "We evaluated 4 systems (ELIZA, GPT-4o, LLaMa-3.1-405B, and GPT-4.5) in two randomised, controlled, and pre-registered Turing tests on independent populations.", "description_method": "abstract-excerpt", "abstract": "We evaluated 4 systems (ELIZA, GPT-4o, LLaMa-3.1-405B, and GPT-4.5) in two randomised, controlled, and pre-registered Turing tests on independent populations. Participants had 5 minute conversations simultaneously with another human participant and one of these systems before judging which conversational partner they thought was human. When prompted to adopt a humanlike persona, GPT-4.5 was judged to be the human 73% of the time: significantly more often than interrogators selected the real human participant. LLaMa-3.1, with the same prompt, was judged to be the human 56% of the time -- not significantly more or less often than the humans they were being compared to -- while baseline models (ELIZA and GPT-4o) achieved win rates significantly below chance (23% and 21% respectively). The results constitute the first empirical evidence that any artificial system passes a standard three-party Turing test. The results have implications for debates about what kind of intelligence is exhibited by Large Language Models (LLMs), and the social and economic impacts these systems are likely to have.", "score": 0.0, "mention_count": 16, "user_count": 12, "months": ["2025-06", "2025-10", "2025-12", "2026-01", "2026-02", "2026-03", "2026-04", "2026-05", "2026-08"], "last_mentioned": "2026-08-17T12:29:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nm.3985", "slug": "00a6475bd748b676", "title": "A SARS-like cluster of circulating bat coronaviruses shows potential for human emergence", "authors": ["Vineet D Menachery", "Boyd L Yount", "Kari Debbink", "Sudhakar Agnihothram", "Lisa E Gralinski", "Jessica A Plante", "Rachel L Graham", "Trevor Scobey", "Xing-Yi Ge", "Eric F Donaldson", "Scott H Randell", "Antonio Lanzavecchia", "Wayne A Marasco", "Zhengli-Li Shi", "Ralph S Baric"], "year": 2015, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/nm.3985", "doi": "10.1038/nm.3985", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 16, "user_count": 12, "months": ["2020-02", "2020-03", "2020-05", "2020-12", "2021-05", "2021-07", "2022-11", "2023-06"], "last_mentioned": "2023-06-11T11:16:10Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1308.0850", "slug": "87a20ba100842ce5", "title": "Generating Sequences With Recurrent Neural Networks", "authors": ["Alex Graves"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1308.0850", "doi": null, "topics": ["Machine learning"], "description": "This paper shows how Long Short-term Memory recurrent neural networks can be used to generate complex sequences with long-range structure, simply by predicting one data point at a time.", "description_method": "abstract-excerpt", "abstract": "This paper shows how Long Short-term Memory recurrent neural networks can be used to generate complex sequences with long-range structure, simply by predicting one data point at a time. The approach is demonstrated for text (where the data are discrete) and online handwriting (where the data are real-valued). It is then extended to handwriting synthesis by allowing the network to condition its predictions on a text sequence. The resulting system is able to generate highly realistic cursive handwriting in a wide variety of styles.", "score": 0.0, "mention_count": 15, "user_count": 11, "months": ["2015-06", "2015-08", "2015-12", "2016-03", "2016-04", "2016-05", "2017-04", "2018-04", "2018-11", "2020-08", "2023-01", "2025-02"], "last_mentioned": "2025-02-08T11:21:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1406.2661", "slug": "346c058b588d8e3c", "title": "Generative Adversarial Networks", "authors": ["Ian J. Goodfellow", "Jean Pouget-Abadie", "Mehdi Mirza", "Bing Xu", "David Warde-Farley", "Sherjil Ozair", "Aaron Courville", "Yoshua Bengio"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1406.2661", "doi": null, "topics": ["Machine learning", "Statistics & probability"], "description": "We propose a new framework for estimating generative models via an adversarial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G.", "description_method": "abstract-excerpt", "abstract": "We propose a new framework for estimating generative models via an adversarial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G. The training procedure for G is to maximize the probability of D making a mistake. This framework corresponds to a minimax two-player game. In the space of arbitrary functions G and D, a unique solution exists, with G recovering the training data distribution and D equal to 1/2 everywhere. In the case where G and D are defined by multilayer perceptrons, the entire system can be trained with backpropagation. There is no need for any Markov chains or unrolled approximate inference networks during either training or generation of samples. Experiments demonstrate the potential of the framework through qualitative and quantitative evaluation of the generated samples.", "score": 0.0, "mention_count": 15, "user_count": 13, "months": ["2014-06", "2015-10", "2017-01", "2017-03", "2017-04", "2018-01", "2018-10", "2020-06", "2021-04", "2023-01", "2023-02", "2023-05", "2023-11", "2024-02", "2024-09"], "last_mentioned": "2024-09-26T05:23:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2110.09485", "slug": "589d10ae90bda37a", "title": "Learning in High Dimension Always Amounts to Extrapolation", "authors": ["Randall Balestriero", "Jerome Pesenti", "Yann LeCun"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2110.09485", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "The notion of interpolation and extrapolation is fundamental in various fields from deep learning to function approximation.", "description_method": "abstract-excerpt", "abstract": "The notion of interpolation and extrapolation is fundamental in various fields from deep learning to function approximation. Interpolation occurs for a sample $x$ whenever this sample falls inside or on the boundary of the given dataset's convex hull. Extrapolation occurs when $x$ falls outside of that convex hull. One fundamental (mis)conception is that state-of-the-art algorithms work so well because of their ability to correctly interpolate training data. A second (mis)conception is that interpolation happens throughout tasks and datasets, in fact, many intuitions and theories rely on that assumption. We empirically and theoretically argue against those two points and demonstrate that on any high-dimensional ($>$100) dataset, interpolation almost surely never happens. Those results challenge the validity of our current interpolation/extrapolation definition as an indicator of generalization performances.", "score": 0.0, "mention_count": 15, "user_count": 10, "months": ["2021-12", "2022-05", "2023-09", "2023-10", "2023-11", "2024-03", "2024-10", "2025-01", "2026-02"], "last_mentioned": "2026-02-14T07:07:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.08073", "slug": "1379bea687106ea1", "title": "Constitutional AI: Harmlessness from AI Feedback", "authors": ["Yuntao Bai", "Saurav Kadavath", "Sandipan Kundu", "Amanda Askell", "Jackson Kernion", "Andy Jones", "Anna Chen", "Anna Goldie", "Azalia Mirhoseini", "Cameron McKinnon", "Carol Chen", "Catherine Olsson", "Christopher Olah", "Danny Hernandez", "Dawn Drain", "Deep Ganguli", "Dustin Li", "Eli Tran-Johnson", "Ethan Perez", "Jamie Kerr", "Jared Mueller", "Jeffrey Ladish", "Joshua Landau", "Kamal Ndousse", "Kamile Lukosuite", "Liane Lovitt", "Michael Sellitto", "Nelson Elhage", "Nicholas Schiefer", "Noemi Mercado", "Nova DasSarma", "Robert Lasenby", "Robin Larson", "Sam Ringer", "Scott Johnston", "Shauna Kravec", "Sheer El Showk", "Stanislav Fort", "Tamera Lanham", "Timothy Telleen-Lawton", "Tom Conerly", "Tom Henighan", "Tristan Hume", "Samuel R. Bowman", "Zac Hatfield-Dodds", "Ben Mann", "Dario Amodei", "Nicholas Joseph", "Sam McCandlish", "Tom Brown", "Jared Kaplan"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.08073", "doi": null, "topics": ["Psychology", "Machine learning"], "description": "As AI systems become more capable, we would like to enlist their help to supervise other AIs.", "description_method": "abstract-excerpt", "abstract": "As AI systems become more capable, we would like to enlist their help to supervise other AIs. We experiment with methods for training a harmless AI assistant through self-improvement, without any human labels identifying harmful outputs. The only human oversight is provided through a list of rules or principles, and so we refer to the method as 'Constitutional AI'. The process involves both a supervised learning and a reinforcement learning phase. In the supervised phase we sample from an initial model, then generate self-critiques and revisions, and then finetune the original model on revised responses. In the RL phase, we sample from the finetuned model, use a model to evaluate which of the two samples is better, and then train a preference model from this dataset of AI preferences. We then train with RL using the preference model as the reward signal, i.e. we use 'RL from AI Feedback' (RLAIF). As a result we are able to train a harmless but non-evasive AI assistant that engages with harmful queries by explaining its objections to them. Both the SL and RL methods can leverage chain-of-thought style reasoning to improve the human-judged performance and transparency of AI decision making. These methods make it possible to control AI behavior more precisely and with far fewer human labels.", "score": 0.0, "mention_count": 15, "user_count": 15, "months": ["2023-02", "2023-03", "2023-05", "2023-09", "2024-08", "2024-10", "2025-02", "2025-06", "2025-12", "2026-01", "2026-02"], "last_mentioned": "2026-02-25T20:22:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.07923", "slug": "5a927e0d66d6126a", "title": "The Expressive Power of Transformers with Chain of Thought", "authors": ["William Merrill", "Ashish Sabharwal"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.07923", "doi": null, "topics": ["Language models", "Algorithms", "Machine learning"], "description": "Recent theoretical work has identified surprisingly simple reasoning problems, such as checking if two nodes in a graph are connected or simulating finite-state machines, that are provably unsolvable by standard transformers that answer immediately after reading their input.", "description_method": "abstract-excerpt", "abstract": "Recent theoretical work has identified surprisingly simple reasoning problems, such as checking if two nodes in a graph are connected or simulating finite-state machines, that are provably unsolvable by standard transformers that answer immediately after reading their input. However, in practice, transformers' reasoning can be improved by allowing them to use a \"chain of thought\" or \"scratchpad\", i.e., generate and condition on a sequence of intermediate tokens before answering. Motivated by this, we ask: Does such intermediate generation fundamentally extend the computational power of a decoder-only transformer? We show that the answer is yes, but the amount of increase depends crucially on the amount of intermediate generation. For instance, we find that transformer decoders with a logarithmic number of decoding steps (w.r.t. the input length) push the limits of standard transformers only slightly, while a linear number of decoding steps, assuming projected pre-norm (a slight generalization of standard pre-norm), adds a clear new ability (under standard complexity conjectures): recognizing all regular languages. Our results also imply that linear steps keep transformer decoders within context-sensitive languages, and polynomial steps with generalized pre-norm make them recognize exactly the class of polynomial-time solvable problems -- the first exact characterization of a type of transformers in terms of standard complexity classes. Together, this provides a nuanced framework for understanding how the length of a transformer's chain of thought or scratchpad impacts its reasoning power.", "score": 0.0, "mention_count": 15, "user_count": 7, "months": ["2024-06", "2024-09", "2024-11", "2024-12", "2025-01", "2025-06", "2025-09", "2026-02", "2026-08"], "last_mentioned": "2026-08-10T19:15:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41562-022-01443-2", "slug": "164ca94764b35f0e", "title": "Science must respect the dignity and rights of all humans", "authors": [], "year": 2022, "venue": "Nature Human Behaviour", "url": "https://doi.org/10.1038/s41562-022-01443-2", "doi": "10.1038/s41562-022-01443-2", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 15, "user_count": 10, "months": ["2022-09", "2022-10", "2022-12", "2023-04", "2023-05", "2023-09", "2023-10", "2024-08", "2025-02", "2025-04", "2025-06", "2026-06"], "last_mentioned": "2026-06-14T03:59:34Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abp8715", "slug": "25f6eae26afae86e", "title": "The Huanan Seafood Wholesale Market in Wuhan was the early epicenter of the COVID-19 pandemic", "authors": ["Michael Worobey", "Joshua I. Levy", "Lorena Malpica Serrano", "Alexander Crits-Christoph", "Jonathan E. Pekar", "Stephen A. Goldstein", "Angela L. Rasmussen", "Moritz U. G. Kraemer", "Chris Newman", "Marion P. G. Koopmans", "Marc A. Suchard", "Joel O. Wertheim", "Philippe Lemey", "David L. Robertson", "Robert F. Garry", "Edward C. Holmes", "Andrew Rambaut", "Kristian G. Andersen"], "year": 2022, "venue": "Science", "url": "https://doi.org/10.1126/science.abp8715", "doi": "10.1126/science.abp8715", "topics": ["Economics"], "description": "Understanding how severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged in 2019 is critical to preventing future zoonotic outbreaks before they become the next pandemic.", "description_method": "abstract-excerpt", "abstract": "Understanding how severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged in 2019 is critical to preventing future zoonotic outbreaks before they become the next pandemic. The Huanan Seafood Wholesale Market in Wuhan, China, was identified as a likely source of cases in early reports, but later this conclusion became controversial. We show here that the earliest known COVID-19 cases from December 2019, including those without reported direct links, were geographically centered on this market. We report that live SARS-CoV-2–susceptible mammals were sold at the market in late 2019 and that within the market, SARS-CoV-2–positive environmental samples were spatially associated with vendors selling live mammals. Although there is insufficient evidence to define upstream events, and exact circumstances remain obscure, our analyses indicate that the emergence of SARS-CoV-2 occurred through the live wildlife trade in China and show that the Huanan market was the epicenter of the COVID-19 pandemic.", "score": 0.0, "mention_count": 15, "user_count": 8, "months": ["2022-09", "2023-06", "2023-07", "2024-09", "2025-01"], "last_mentioned": "2025-01-26T08:27:05Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1701.06538", "slug": "a419dd17d47978d2", "title": "Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer", "authors": ["Noam Shazeer", "Azalia Mirhoseini", "Krzysztof Maziarz", "Andy Davis", "Quoc Le", "Geoffrey Hinton", "Jeff Dean"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1701.06538", "doi": null, "topics": ["Machine learning", "Language models", "Computer vision"], "description": "The capacity of a neural network to absorb information is limited by its number of parameters.", "description_method": "abstract-excerpt", "abstract": "The capacity of a neural network to absorb information is limited by its number of parameters. Conditional computation, where parts of the network are active on a per-example basis, has been proposed in theory as a way of dramatically increasing model capacity without a proportional increase in computation. In practice, however, there are significant algorithmic and performance challenges. In this work, we address these challenges and finally realize the promise of conditional computation, achieving greater than 1000x improvements in model capacity with only minor losses in computational efficiency on modern GPU clusters. We introduce a Sparsely-Gated Mixture-of-Experts layer (MoE), consisting of up to thousands of feed-forward sub-networks. A trainable gating network determines a sparse combination of these experts to use for each example. We apply the MoE to the tasks of language modeling and machine translation, where model capacity is critical for absorbing the vast quantities of knowledge available in the training corpora. We present model architectures in which a MoE with up to 137 billion parameters is applied convolutionally between stacked LSTM layers. On large language modeling and machine translation benchmarks, these models achieve significantly better results than state-of-the-art at lower computational cost.", "score": 0.0, "mention_count": 14, "user_count": 12, "months": ["2017-07", "2017-08", "2018-04", "2018-12", "2019-11", "2020-07", "2023-07", "2023-12", "2024-06", "2025-08", "2026-07"], "last_mentioned": "2026-07-27T19:11:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2010.11929", "slug": "191d18337a0bd2d1", "title": "An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale", "authors": ["Alexey Dosovitskiy", "Lucas Beyer", "Alexander Kolesnikov", "Dirk Weissenborn", "Xiaohua Zhai", "Thomas Unterthiner", "Mostafa Dehghani", "Matthias Minderer", "Georg Heigold", "Sylvain Gelly", "Jakob Uszkoreit", "Neil Houlsby"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2010.11929", "doi": null, "topics": ["Computer vision", "Language models", "Psychology"], "description": "While the Transformer architecture has become the de-facto standard for natural language processing tasks, its applications to computer vision remain limited.", "description_method": "abstract-excerpt", "abstract": "While the Transformer architecture has become the de-facto standard for natural language processing tasks, its applications to computer vision remain limited. In vision, attention is either applied in conjunction with convolutional networks, or used to replace certain components of convolutional networks while keeping their overall structure in place. We show that this reliance on CNNs is not necessary and a pure transformer applied directly to sequences of image patches can perform very well on image classification tasks. When pre-trained on large amounts of data and transferred to multiple mid-sized or small image recognition benchmarks (ImageNet, CIFAR-100, VTAB, etc.), Vision Transformer (ViT) attains excellent results compared to state-of-the-art convolutional networks while requiring substantially fewer computational resources to train.", "score": 0.0, "mention_count": 14, "user_count": 13, "months": ["2021-07", "2021-08", "2022-03", "2022-08", "2023-03", "2023-06", "2023-11", "2024-02", "2024-10", "2024-12", "2025-07"], "last_mentioned": "2025-07-05T12:14:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.09720", "slug": "33b88e7107bc9f1e", "title": "The case for 4-bit precision: k-bit Inference Scaling Laws", "authors": ["Tim Dettmers", "Luke Zettlemoyer"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.09720", "doi": null, "topics": ["Language models"], "description": "Quantization methods reduce the number of bits required to represent each parameter in a model, trading accuracy for smaller memory footprints and inference latencies.", "description_method": "abstract-excerpt", "abstract": "Quantization methods reduce the number of bits required to represent each parameter in a model, trading accuracy for smaller memory footprints and inference latencies. However, the final model size depends on both the number of parameters of the original model and the rate of compression. For example, a 30B 8-bit model and a 60B 4-bit model have the same number of bits but may have very different zero-shot accuracies. In this work, we study this trade-off by developing inference scaling laws of zero-shot performance in Large Language Models (LLMs) to determine the bit-precision and model size that maximizes zero-shot performance. We run more than 35,000 experiments with 16-bit inputs and k-bit parameters to examine which zero-shot quantization methods improve scaling for 3 to 8-bit precision at scales of 19M to 176B parameters across the LLM families BLOOM, OPT, NeoX/Pythia, and GPT-2. We find that it is challenging to improve the bit-level scaling trade-off, with the only improvements being the use of a small block size -- splitting the parameters into small independently quantized blocks -- and the quantization data type being used (e.g., Int vs Float). Overall, our findings show that {4-bit} precision is almost universally optimal for total model bits and zero-shot accuracy.", "score": 0.0, "mention_count": 14, "user_count": 9, "months": ["2023-01", "2023-03", "2023-08", "2023-11", "2024-01", "2024-04", "2024-11", "2025-03", "2026-06"], "last_mentioned": "2026-06-29T23:17:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.02226", "slug": "bc0dd93e0662c047", "title": "Think before you speak: Training Language Models With Pause Tokens", "authors": ["Sachin Goyal", "Ziwei Ji", "Ankit Singh Rawat", "Aditya Krishna Menon", "Sanjiv Kumar", "Vaishnavh Nagarajan"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.02226", "doi": null, "topics": ["Language models"], "description": "Language models generate responses by producing a series of tokens in immediate succession: the $(K+1)^{th}$ token is an outcome of manipulating $K$ hidden vectors per layer, one vector per preceding token.", "description_method": "abstract-excerpt", "abstract": "Language models generate responses by producing a series of tokens in immediate succession: the $(K+1)^{th}$ token is an outcome of manipulating $K$ hidden vectors per layer, one vector per preceding token. What if instead we were to let the model manipulate say, $K+10$ hidden vectors, before it outputs the $(K+1)^{th}$ token? We operationalize this idea by performing training and inference on language models with a (learnable) $\\textit{pause}$ token, a sequence of which is appended to the input prefix. We then delay extracting the model's outputs until the last pause token is seen, thereby allowing the model to process extra computation before committing to an answer. We empirically evaluate $\\textit{pause-training}$ on decoder-only models of 1B and 130M parameters with causal pretraining on C4, and on downstream tasks covering reasoning, question-answering, general understanding and fact recall. Our main finding is that inference-time delays show gains when the model is both pre-trained and finetuned with delays. For the 1B model, we witness gains on 8 of 9 tasks, most prominently, a gain of $18\\%$ EM score on the QA task of SQuAD, $8\\%$ on CommonSenseQA and $1\\%$ accuracy on the reasoning task of GSM8k. Our work raises a range of conceptual and practical future research questions on making delayed next-token prediction a widely applicable new paradigm.", "score": 0.0, "mention_count": 14, "user_count": 12, "months": ["2023-10", "2024-02", "2024-03", "2024-05", "2024-06", "2024-07", "2024-09", "2025-03", "2025-05", "2025-10"], "last_mentioned": "2025-10-06T06:24:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2410.05229", "slug": "eb778a0201af689b", "title": "GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models", "authors": ["Iman Mirzadeh", "Keivan Alizadeh", "Hooman Shahrokhi", "Oncel Tuzel", "Samy Bengio", "Mehrdad Farajtabar"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2410.05229", "doi": null, "topics": ["Mathematics", "Language models"], "description": "Recent advancements in Large Language Models (LLMs) have sparked interest in their formal reasoning capabilities, particularly in mathematics.", "description_method": "abstract-excerpt", "abstract": "Recent advancements in Large Language Models (LLMs) have sparked interest in their formal reasoning capabilities, particularly in mathematics. The GSM8K benchmark is widely used to assess the mathematical reasoning of models on grade-school-level questions. While the performance of LLMs on GSM8K has significantly improved in recent years, it remains unclear whether their mathematical reasoning capabilities have genuinely advanced, raising questions about the reliability of the reported metrics. To address these concerns, we conduct a large-scale study on several SOTA open and closed models. To overcome the limitations of existing evaluations, we introduce GSM-Symbolic, an improved benchmark created from symbolic templates that allow for the generation of a diverse set of questions. GSM-Symbolic enables more controllable evaluations, providing key insights and more reliable metrics for measuring the reasoning capabilities of models.Our findings reveal that LLMs exhibit noticeable variance when responding to different instantiations of the same question. Specifically, the performance of all models declines when only the numerical values in the question are altered in the GSM-Symbolic benchmark. Furthermore, we investigate the fragility of mathematical reasoning in these models and show that their performance significantly deteriorates as the number of clauses in a question increases. We hypothesize that this decline is because current LLMs cannot perform genuine logical reasoning; they replicate reasoning steps from their training data. Adding a single clause that seems relevant to the question causes significant performance drops (up to 65%) across all state-of-the-art models, even though the clause doesn't contribute to the reasoning chain needed for the final answer. Overall, our work offers a more nuanced understanding of LLMs' capabilities and limitations in mathematical reasoning.", "score": 0.0, "mention_count": 14, "user_count": 14, "months": ["2024-10", "2024-11", "2024-12", "2025-01", "2025-06", "2026-07"], "last_mentioned": "2026-07-06T22:38:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-020-19802-w", "slug": "d01b484f1ccefbd8", "title": "Post-lockdown SARS-CoV-2 nucleic acid screening in nearly ten million residents of Wuhan, China", "authors": ["Shiyi Cao", "Yong Gan", "Chao Wang", "Max Bachmann", "Shanbo Wei", "Jie Gong", "Yuchai Huang", "Tiantian Wang", "Liqing Li", "Kai Lu", "Heng Jiang", "Yanhong Gong", "Hongbin Xu", "Xin Shen", "Qingfeng Tian", "Chuanzhu Lv", "Fujian Song", "Xiaoxv Yin", "Zuxun Lu"], "year": 2020, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-020-19802-w", "doi": "10.1038/s41467-020-19802-w", "topics": ["Other research"], "description": "Abstract Stringent COVID-19 control measures were imposed in Wuhan between January 23 and April 8, 2020.", "description_method": "abstract-excerpt", "abstract": "Abstract Stringent COVID-19 control measures were imposed in Wuhan between January 23 and April 8, 2020. Estimates of the prevalence of infection following the release of restrictions could inform post-lockdown pandemic management. Here, we describe a city-wide SARS-CoV-2 nucleic acid screening programme between May 14 and June 1, 2020 in Wuhan. All city residents aged six years or older were eligible and 9,899,828 (92.9%) participated. No new symptomatic cases and 300 asymptomatic cases (detection rate 0.303/10,000, 95% CI 0.270–0.339/10,000) were identified. There were no positive tests amongst 1,174 close contacts of asymptomatic cases. 107 of 34,424 previously recovered COVID-19 patients tested positive again (re-positive rate 0.31%, 95% CI 0.423–0.574%). The prevalence of SARS-CoV-2 infection in Wuhan was therefore very low five to eight weeks after the end of lockdown.", "score": 0.0, "mention_count": 14, "user_count": 10, "months": ["2020-11", "2020-12", "2021-02", "2021-05", "2021-06", "2021-07"], "last_mentioned": "2021-07-13T18:32:00Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/359576.359579", "slug": "55215fdcca1642ad", "title": "Can programming be liberated from the von Neumann style?", "authors": ["John Backus"], "year": 1978, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/359576.359579", "doi": "10.1145/359576.359579", "topics": ["Programming languages", "Mathematics", "Psychology"], "description": "Conventional programming languages are growing ever more enormous, but not stronger.", "description_method": "abstract-excerpt", "abstract": "Conventional programming languages are growing ever more enormous, but not stronger. Inherent defects at the most basic level cause them to be both fat and weak: their primitive word-at-a-time style of programming inherited from their common ancestor—the von Neumann computer, their close coupling of semantics to state transitions, their division of programming into a world of expressions and a world of statements, their inability to effectively use powerful combining forms for building new programs from existing ones, and their lack of useful mathematical properties for reasoning about programs. An alternative functional style of programming is founded on the use of combining forms for creating programs. Functional programs deal with structured data, are often nonrepetitive and nonrecursive, are hierarchically constructed, do not name their arguments, and do not require the complex machinery of procedure declarations to become generally applicable. Combining forms can use high level programs to build still higher level ones in a style not possible in conventional languages. Associated with the functional style of programming is an algebra of programs whose variables range over programs and whose operations are combining forms. This algebra can be used to transform programs and to solve equations whose “unknowns” are programs in much the same way one transforms equations in high school algebra. These transformations are given by algebraic laws and are carried out in the same language in which programs are written. Combining forms are chosen not only for their programming power but also for the power of their associated algebraic laws. General theorems of the algebra give the detailed behavior and termination conditions for large classes of programs. A new class of computing systems uses the functional programming style both in its programming language and in its state transition rules. Unlike von Neumann languages, these systems have semantics loosely coupled to states—only one state transition occurs per major computation.", "score": 0.0, "mention_count": 14, "user_count": 11, "months": ["2020-08", "2020-10", "2021-04", "2021-12", "2022-10", "2023-03", "2024-03", "2024-09", "2024-10", "2025-01", "2025-09"], "last_mentioned": "2025-09-27T00:39:09Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0712.3329", "slug": "ecb93b4b901844a2", "title": "Universal Intelligence: A Definition of Machine Intelligence", "authors": ["Shane Legg", "Marcus Hutter"], "year": 2007, "venue": "arXiv", "url": "https://arxiv.org/abs/0712.3329", "doi": null, "topics": ["Psychology", "Mathematics"], "description": "A fundamental problem in artificial intelligence is that nobody really knows what intelligence is.", "description_method": "abstract-excerpt", "abstract": "A fundamental problem in artificial intelligence is that nobody really knows what intelligence is. The problem is especially acute when we need to consider artificial systems which are significantly different to humans. In this paper we approach this problem in the following way: We take a number of well known informal definitions of human intelligence that have been given by experts, and extract their essential features. These are then mathematically formalised to produce a general measure of intelligence for arbitrary machines. We believe that this equation formally captures the concept of machine intelligence in the broadest reasonable sense. We then show how this formal definition is related to the theory of universal optimal learning agents. Finally, we survey the many other tests and definitions of intelligence that have been proposed for machines.", "score": 0.0, "mention_count": 13, "user_count": 13, "months": ["2015-02", "2016-08", "2017-03", "2018-01", "2018-08", "2018-12", "2019-02", "2022-01", "2022-04", "2022-08", "2023-03"], "last_mentioned": "2023-03-26T12:26:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1702.01715", "slug": "e1798dab32f6d30d", "title": "Software Engineering at Google", "authors": ["Fergus Henderson"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1702.01715", "doi": null, "topics": ["Programming languages"], "description": "We catalog and describe Google's key software engineering practices.", "description_method": "abstract-excerpt", "abstract": "We catalog and describe Google's key software engineering practices.", "score": 0.0, "mention_count": 13, "user_count": 12, "months": ["2017-02", "2017-03", "2017-04", "2017-05", "2018-02", "2019-01", "2019-03", "2019-09", "2020-06", "2020-10", "2022-06", "2022-07"], "last_mentioned": "2022-07-25T14:42:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1911.08265", "slug": "a8d5cc8d1b028cbf", "title": "Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model", "authors": ["Julian Schrittwieser", "Ioannis Antonoglou", "Thomas Hubert", "Karen Simonyan", "Laurent Sifre", "Simon Schmitt", "Arthur Guez", "Edward Lockhart", "Demis Hassabis", "Thore Graepel", "Timothy Lillicrap", "David Silver"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1911.08265", "doi": "10.1038/s41586-020-03051-4", "topics": ["Algorithms", "Psychology"], "description": "Constructing agents with planning capabilities has long been one of the main challenges in the pursuit of artificial intelligence.", "description_method": "abstract-excerpt", "abstract": "Constructing agents with planning capabilities has long been one of the main challenges in the pursuit of artificial intelligence. Tree-based planning methods have enjoyed huge success in challenging domains, such as chess and Go, where a perfect simulator is available. However, in real-world problems the dynamics governing the environment are often complex and unknown. In this work we present the MuZero algorithm which, by combining a tree-based search with a learned model, achieves superhuman performance in a range of challenging and visually complex domains, without any knowledge of their underlying dynamics. MuZero learns a model that, when applied iteratively, predicts the quantities most directly relevant to planning: the reward, the action-selection policy, and the value function. When evaluated on 57 different Atari games - the canonical video game environment for testing AI techniques, in which model-based planning approaches have historically struggled - our new algorithm achieved a new state of the art. When evaluated on Go, chess and shogi, without any knowledge of the game rules, MuZero matched the superhuman performance of the AlphaZero algorithm that was supplied with the game rules.", "score": 0.0, "mention_count": 13, "user_count": 13, "months": ["2019-11", "2019-12", "2020-07", "2020-08", "2020-12", "2022-06", "2023-07", "2025-09"], "last_mentioned": "2025-09-02T21:21:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2004.11663", "slug": "bd410e6cedea4500", "title": "Retrofitting Parallelism onto OCaml", "authors": ["KC Sivaramakrishnan", "Stephen Dolan", "Leo White", "Sadiq Jaffer", "Tom Kelly", "Anmol Sahoo", "Sudha Parimala", "Atul Dhiman", "Anil Madhavapeddy"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2004.11663", "doi": "10.1145/3408995", "topics": ["Programming languages", "Computer systems"], "description": "OCaml is an industrial-strength, multi-paradigm programming language, widely used in industry and academia.", "description_method": "abstract-excerpt", "abstract": "OCaml is an industrial-strength, multi-paradigm programming language, widely used in industry and academia. OCaml is also one of the few modern managed system programming languages to lack support for shared memory parallel programming. This paper describes the design, a full-fledged implementation and evaluation of a mostly-concurrent garbage collector (GC) for the multicore extension of the OCaml programming language. Given that we propose to add parallelism to a widely used programming language with millions of lines of existing code, we face the challenge of maintaining backwards compatibility--not just in terms of the language features but also the performance of single-threaded code running with the new GC. To this end, the paper presents a series of novel techniques and demonstrates that the new GC strikes a balance between performance and feature backwards compatibility for sequential programs and scales admirably on modern multicore processors.", "score": 0.0, "mention_count": 13, "user_count": 4, "months": ["2020-06", "2020-07", "2020-08", "2020-09", "2020-11", "2021-03", "2021-05", "2021-09", "2022-01", "2022-12"], "last_mentioned": "2022-12-16T15:54:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2006.08381", "slug": "bd2cc190adaa3900", "title": "DreamCoder: Growing generalizable, interpretable knowledge with wake-sleep Bayesian program learning", "authors": ["Kevin Ellis", "Catherine Wong", "Maxwell Nye", "Mathias Sable-Meyer", "Luc Cary", "Lucas Morales", "Luke Hewitt", "Armando Solar-Lezama", "Joshua B. Tenenbaum"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2006.08381", "doi": null, "topics": ["Programming languages", "Machine learning", "Statistics & probability"], "description": "Expert problem-solving is driven by powerful languages for thinking about problems and their solutions.", "description_method": "abstract-excerpt", "abstract": "Expert problem-solving is driven by powerful languages for thinking about problems and their solutions. Acquiring expertise means learning these languages -- systems of concepts, alongside the skills to use them. We present DreamCoder, a system that learns to solve problems by writing programs. It builds expertise by creating programming languages for expressing domain concepts, together with neural networks to guide the search for programs within these languages. A ``wake-sleep'' learning algorithm alternately extends the language with new symbolic abstractions and trains the neural network on imagined and replayed problems. DreamCoder solves both classic inductive programming tasks and creative tasks such as drawing pictures and building scenes. It rediscovers the basics of modern functional programming, vector algebra and classical physics, including Newton's and Coulomb's laws. Concepts are built compositionally from those learned earlier, yielding multi-layered symbolic representations that are interpretable and transferrable to new tasks, while still growing scalably and flexibly with experience.", "score": 0.0, "mention_count": 13, "user_count": 10, "months": ["2020-12", "2021-08", "2021-11", "2022-02", "2022-07", "2023-05", "2023-07", "2023-09", "2024-10", "2026-05"], "last_mentioned": "2026-05-26T17:22:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2112.04426", "slug": "a1df8d8d59a41f04", "title": "Improving language models by retrieving from trillions of tokens", "authors": ["Sebastian Borgeaud", "Arthur Mensch", "Jordan Hoffmann", "Trevor Cai", "Eliza Rutherford", "Katie Millican", "George van den Driessche", "Jean-Baptiste Lespiau", "Bogdan Damoc", "Aidan Clark", "Diego de Las Casas", "Aurelia Guy", "Jacob Menick", "Roman Ring", "Tom Hennigan", "Saffron Huang", "Loren Maggiore", "Chris Jones", "Albin Cassirer", "Andy Brock", "Michela Paganini", "Geoffrey Irving", "Oriol Vinyals", "Simon Osindero", "Karen Simonyan", "Jack W. Rae", "Erich Elsen", "Laurent Sifre"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2112.04426", "doi": null, "topics": ["Language models", "Databases"], "description": "We enhance auto-regressive language models by conditioning on document chunks retrieved from a large corpus, based on local similarity with preceding tokens.", "description_method": "abstract-excerpt", "abstract": "We enhance auto-regressive language models by conditioning on document chunks retrieved from a large corpus, based on local similarity with preceding tokens. With a $2$ trillion token database, our Retrieval-Enhanced Transformer (RETRO) obtains comparable performance to GPT-3 and Jurassic-1 on the Pile, despite using 25$\\times$ fewer parameters. After fine-tuning, RETRO performance translates to downstream knowledge-intensive tasks such as question answering. RETRO combines a frozen Bert retriever, a differentiable encoder and a chunked cross-attention mechanism to predict tokens based on an order of magnitude more data than what is typically consumed during training. We typically train RETRO from scratch, yet can also rapidly RETROfit pre-trained transformers with retrieval and still achieve good performance. Our work opens up new avenues for improving language models through explicit memory at unprecedented scale.", "score": 0.0, "mention_count": 13, "user_count": 9, "months": ["2022-01", "2022-02", "2022-07", "2022-12", "2023-01", "2023-05", "2024-03", "2024-06", "2024-11"], "last_mentioned": "2024-11-10T01:20:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2201.02177", "slug": "e9ad18ed3a6be829", "title": "Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets", "authors": ["Alethea Power", "Yuri Burda", "Harri Edwards", "Igor Babuschkin", "Vedant Misra"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2201.02177", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "In this paper we propose to study generalization of neural networks on small algorithmically generated datasets.", "description_method": "abstract-excerpt", "abstract": "In this paper we propose to study generalization of neural networks on small algorithmically generated datasets. In this setting, questions about data efficiency, memorization, generalization, and speed of learning can be studied in great detail. In some situations we show that neural networks learn through a process of \"grokking\" a pattern in the data, improving generalization performance from random chance level to perfect generalization, and that this improvement in generalization can happen well past the point of overfitting. We also study generalization as a function of dataset size and find that smaller datasets require increasing amounts of optimization for generalization. We argue that these datasets provide a fertile ground for studying a poorly understood aspect of deep learning: generalization of overparametrized neural networks beyond memorization of the finite training dataset.", "score": 0.0, "mention_count": 13, "user_count": 12, "months": ["2022-02", "2022-04", "2023-02", "2023-03", "2023-04", "2023-05", "2023-08", "2024-04", "2026-02", "2026-06"], "last_mentioned": "2026-06-04T03:59:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2204.02311", "slug": "3bcd5b39cab255a8", "title": "PaLM: Scaling Language Modeling with Pathways", "authors": ["Aakanksha Chowdhery", "Sharan Narang", "Jacob Devlin", "Maarten Bosma", "Gaurav Mishra", "Adam Roberts", "Paul Barham", "Hyung Won Chung", "Charles Sutton", "Sebastian Gehrmann", "Parker Schuh", "Kensen Shi", "Sasha Tsvyashchenko", "Joshua Maynez", "Abhishek Rao", "Parker Barnes", "Yi Tay", "Noam Shazeer", "Vinodkumar Prabhakaran", "Emily Reif", "Nan Du", "Ben Hutchinson", "Reiner Pope", "James Bradbury", "Jacob Austin", "Michael Isard", "Guy Gur-Ari", "Pengcheng Yin", "Toju Duke", "Anselm Levskaya", "Sanjay Ghemawat", "Sunipa Dev", "Henryk Michalewski", "Xavier Garcia", "Vedant Misra", "Kevin Robinson", "Liam Fedus", "Denny Zhou", "Daphne Ippolito", "David Luan", "Hyeontaek Lim", "Barret Zoph", "Alexander Spiridonov", "Ryan Sepassi", "David Dohan", "Shivani Agrawal", "Mark Omernick", "Andrew M. Dai", "Thanumalayan Sankaranarayana Pillai", "Marie Pellat", "Aitor Lewkowycz", "Erica Moreira", "Rewon Child", "Oleksandr Polozov", "Katherine Lee", "Zongwei Zhou", "Xuezhi Wang", "Brennan Saeta", "Mark Diaz", "Orhan Firat", "Michele Catasta", "Jason Wei", "Kathy Meier-Hellstern", "Douglas Eck", "Jeff Dean", "Slav Petrov", "Noah Fiedel"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2204.02311", "doi": null, "topics": ["Language models", "Psychology"], "description": "Large language models have been shown to achieve remarkable performance across a variety of natural language tasks using few-shot learning, which drastically reduces the number of task-specific training examples needed to adapt the model to a particular application.", "description_method": "abstract-excerpt", "abstract": "Large language models have been shown to achieve remarkable performance across a variety of natural language tasks using few-shot learning, which drastically reduces the number of task-specific training examples needed to adapt the model to a particular application. To further our understanding of the impact of scale on few-shot learning, we trained a 540-billion parameter, densely activated, Transformer language model, which we call Pathways Language Model PaLM. We trained PaLM on 6144 TPU v4 chips using Pathways, a new ML system which enables highly efficient training across multiple TPU Pods. We demonstrate continued benefits of scaling by achieving state-of-the-art few-shot learning results on hundreds of language understanding and generation benchmarks. On a number of these tasks, PaLM 540B achieves breakthrough performance, outperforming the finetuned state-of-the-art on a suite of multi-step reasoning tasks, and outperforming average human performance on the recently released BIG-bench benchmark. A significant number of BIG-bench tasks showed discontinuous improvements from model scale, meaning that performance steeply increased as we scaled to our largest model. PaLM also has strong capabilities in multilingual tasks and source code generation, which we demonstrate on a wide array of benchmarks. We additionally provide a comprehensive analysis on bias and toxicity, and study the extent of training data memorization with respect to model scale. Finally, we discuss the ethical considerations related to large language models and discuss potential mitigation strategies.", "score": 0.0, "mention_count": 13, "user_count": 12, "months": ["2022-04", "2022-05", "2022-06", "2022-07", "2022-10", "2022-11", "2022-12", "2023-03", "2023-05"], "last_mentioned": "2023-05-24T06:16:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2205.11916", "slug": "273984ea93c49a37", "title": "Large Language Models are Zero-Shot Reasoners", "authors": ["Takeshi Kojima", "Shixiang Shane Gu", "Machel Reid", "Yutaka Matsuo", "Yusuke Iwasawa"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2205.11916", "doi": null, "topics": ["Language models", "Neuroscience"], "description": "Pretrained large language models (LLMs) are widely used in many sub-fields of natural language processing (NLP) and generally known as excellent few-shot learners with task-specific exemplars.", "description_method": "abstract-excerpt", "abstract": "Pretrained large language models (LLMs) are widely used in many sub-fields of natural language processing (NLP) and generally known as excellent few-shot learners with task-specific exemplars. Notably, chain of thought (CoT) prompting, a recent technique for eliciting complex multi-step reasoning through step-by-step answer examples, achieved the state-of-the-art performances in arithmetics and symbolic reasoning, difficult system-2 tasks that do not follow the standard scaling laws for LLMs. While these successes are often attributed to LLMs' ability for few-shot learning, we show that LLMs are decent zero-shot reasoners by simply adding \"Let's think step by step\" before each answer. Experimental results demonstrate that our Zero-shot-CoT, using the same single prompt template, significantly outperforms zero-shot LLM performances on diverse benchmark reasoning tasks including arithmetics (MultiArith, GSM8K, AQUA-RAT, SVAMP), symbolic reasoning (Last Letter, Coin Flip), and other logical reasoning tasks (Date Understanding, Tracking Shuffled Objects), without any hand-crafted few-shot examples, e.g. increasing the accuracy on MultiArith from 17.7% to 78.7% and GSM8K from 10.4% to 40.7% with large InstructGPT model (text-davinci-002), as well as similar magnitudes of improvements with another off-the-shelf large model, 540B parameter PaLM. The versatility of this single prompt across very diverse reasoning tasks hints at untapped and understudied fundamental zero-shot capabilities of LLMs, suggesting high-level, multi-task broad cognitive capabilities may be extracted by simple prompting. We hope our work not only serves as the minimal strongest zero-shot baseline for the challenging reasoning benchmarks, but also highlights the importance of carefully exploring and analyzing the enormous zero-shot knowledge hidden inside LLMs before crafting finetuning datasets or few-shot exemplars.", "score": 0.0, "mention_count": 13, "user_count": 12, "months": ["2022-06", "2023-01", "2023-02", "2023-03", "2023-05", "2023-07", "2023-10", "2023-11", "2024-07", "2024-09", "2025-03", "2025-06", "2026-01"], "last_mentioned": "2026-01-20T19:10:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.17493", "slug": "0da7913c890d2c36", "title": "The Curse of Recursion: Training on Generated Data Makes Models Forget", "authors": ["Ilia Shumailov", "Zakhar Shumaylov", "Yiren Zhao", "Yarin Gal", "Nicolas Papernot", "Ross Anderson"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.17493", "doi": null, "topics": ["Language models", "Physics", "Networks"], "description": "Stable Diffusion revolutionised image creation from descriptive text.", "description_method": "abstract-excerpt", "abstract": "Stable Diffusion revolutionised image creation from descriptive text. GPT-2, GPT-3(.5) and GPT-4 demonstrated astonishing performance across a variety of language tasks. ChatGPT introduced such language models to the general public. It is now clear that large language models (LLMs) are here to stay, and will bring about drastic change in the whole ecosystem of online text and images. In this paper we consider what the future might hold. What will happen to GPT-{n} once LLMs contribute much of the language found online? We find that use of model-generated content in training causes irreversible defects in the resulting models, where tails of the original content distribution disappear. We refer to this effect as Model Collapse and show that it can occur in Variational Autoencoders, Gaussian Mixture Models and LLMs. We build theoretical intuition behind the phenomenon and portray its ubiquity amongst all learned generative models. We demonstrate that it has to be taken seriously if we are to sustain the benefits of training from large-scale data scraped from the web. Indeed, the value of data collected about genuine human interactions with systems will be increasingly valuable in the presence of content generated by LLMs in data crawled from the Internet.", "score": 0.0, "mention_count": 13, "user_count": 12, "months": ["2023-06", "2023-07", "2023-08", "2023-11", "2024-01", "2024-02", "2024-04", "2024-05", "2024-06"], "last_mentioned": "2024-06-27T21:23:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2307.11760", "slug": "1577f97eabb67914", "title": "Large Language Models Understand and Can be Enhanced by Emotional Stimuli", "authors": ["Cheng Li", "Jindong Wang", "Yixuan Zhang", "Kaijie Zhu", "Wenxin Hou", "Jianxun Lian", "Fang Luo", "Qiang Yang", "Xing Xie"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2307.11760", "doi": null, "topics": ["Psychology", "Language models"], "description": "Emotional intelligence significantly impacts our daily behaviors and interactions.", "description_method": "abstract-excerpt", "abstract": "Emotional intelligence significantly impacts our daily behaviors and interactions. Although Large Language Models (LLMs) are increasingly viewed as a stride toward artificial general intelligence, exhibiting impressive performance in numerous tasks, it is still uncertain if LLMs can genuinely grasp psychological emotional stimuli. Understanding and responding to emotional cues gives humans a distinct advantage in problem-solving. In this paper, we take the first step towards exploring the ability of LLMs to understand emotional stimuli. To this end, we first conduct automatic experiments on 45 tasks using various LLMs, including Flan-T5-Large, Vicuna, Llama 2, BLOOM, ChatGPT, and GPT-4. Our tasks span deterministic and generative applications that represent comprehensive evaluation scenarios. Our automatic experiments show that LLMs have a grasp of emotional intelligence, and their performance can be improved with emotional prompts (which we call \"EmotionPrompt\" that combines the original prompt with emotional stimuli), e.g., 8.00% relative performance improvement in Instruction Induction and 115% in BIG-Bench. In addition to those deterministic tasks that can be automatically evaluated using existing metrics, we conducted a human study with 106 participants to assess the quality of generative tasks using both vanilla and emotional prompts. Our human study results demonstrate that EmotionPrompt significantly boosts the performance of generative tasks (10.9% average improvement in terms of performance, truthfulness, and responsibility metrics). We provide an in-depth discussion regarding why EmotionPrompt works for LLMs and the factors that may influence its performance. We posit that EmotionPrompt heralds a novel avenue for exploring interdisciplinary knowledge for human-LLMs interaction.", "score": 0.0, "mention_count": 13, "user_count": 9, "months": ["2023-10", "2023-11", "2025-01", "2025-04", "2026-02", "2026-05"], "last_mentioned": "2026-05-22T13:36:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2311.02462", "slug": "288b98572ecf9c0b", "title": "Levels of AGI for Operationalizing Progress on the Path to AGI", "authors": ["Meredith Ringel Morris", "Jascha Sohl-Dickstein", "Noah Fiedel", "Tris Warkentin", "Allan Dafoe", "Aleksandra Faust", "Clement Farabet", "Shane Legg"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2311.02462", "doi": null, "topics": ["Psychology"], "description": "We propose a framework for classifying the capabilities and behavior of Artificial General Intelligence (AGI) models and their precursors.", "description_method": "abstract-excerpt", "abstract": "We propose a framework for classifying the capabilities and behavior of Artificial General Intelligence (AGI) models and their precursors. This framework introduces levels of AGI performance, generality, and autonomy, providing a common language to compare models, assess risks, and measure progress along the path to AGI. To develop our framework, we analyze existing definitions of AGI, and distill six principles that a useful ontology for AGI should satisfy. With these principles in mind, we propose \"Levels of AGI\" based on depth (performance) and breadth (generality) of capabilities, and reflect on how current systems fit into this ontology. We discuss the challenging requirements for future benchmarks that quantify the behavior and capabilities of AGI models against these levels. Finally, we discuss how these levels of AGI interact with deployment considerations such as autonomy and risk, and emphasize the importance of carefully selecting Human-AI Interaction paradigms for responsible and safe deployment of highly capable AI systems.", "score": 0.0, "mention_count": 13, "user_count": 8, "months": ["2023-11", "2023-12", "2024-03", "2024-05", "2024-06", "2025-05", "2025-06", "2025-10"], "last_mentioned": "2025-10-18T15:55:31Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2404.15758", "slug": "5f94c36c070dbd61", "title": "Let's Think Dot by Dot: Hidden Computation in Transformer Language Models", "authors": ["Jacob Pfau", "William Merrill", "Samuel R. Bowman"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2404.15758", "doi": null, "topics": ["Language models", "Algorithms"], "description": "Chain-of-thought responses from language models improve performance across most benchmarks.", "description_method": "abstract-excerpt", "abstract": "Chain-of-thought responses from language models improve performance across most benchmarks. However, it remains unclear to what extent these performance gains can be attributed to human-like task decomposition or simply the greater computation that additional tokens allow. We show that transformers can use meaningless filler tokens (e.g., '......') in place of a chain of thought to solve two hard algorithmic tasks they could not solve when responding without intermediate tokens. However, we find empirically that learning to use filler tokens is difficult and requires specific, dense supervision to converge. We also provide a theoretical characterization of the class of problems where filler tokens are useful in terms of the quantifier depth of a first-order formula. For problems satisfying this characterization, chain-of-thought tokens need not provide information about the intermediate computational steps involved in multi-token computations. In summary, our results show that additional tokens can provide computational benefits independent of token choice. The fact that intermediate tokens can act as filler tokens raises concerns about large language models engaging in unauditable, hidden computations that are increasingly detached from the observed chain-of-thought tokens.", "score": 0.0, "mention_count": 13, "user_count": 11, "months": ["2024-05", "2024-07", "2025-01", "2025-03", "2025-05", "2025-07", "2026-02", "2026-04", "2026-08"], "last_mentioned": "2026-08-17T06:40:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2501.15654", "slug": "94bc1872a6410141", "title": "People who frequently use ChatGPT for writing tasks are accurate and robust detectors of AI-generated text", "authors": ["Jenna Russell", "Marzena Karpinska", "Mohit Iyyer"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2501.15654", "doi": null, "topics": ["Language models"], "description": "In this paper, we study how well humans can detect text generated by commercial LLMs (GPT-4o, Claude, o1).", "description_method": "abstract-excerpt", "abstract": "In this paper, we study how well humans can detect text generated by commercial LLMs (GPT-4o, Claude, o1). We hire annotators to read 300 non-fiction English articles, label them as either human-written or AI-generated, and provide paragraph-length explanations for their decisions. Our experiments show that annotators who frequently use LLMs for writing tasks excel at detecting AI-generated text, even without any specialized training or feedback. In fact, the majority vote among five such \"expert\" annotators misclassifies only 1 of 300 articles, significantly outperforming most commercial and open-source detectors we evaluated even in the presence of evasion tactics like paraphrasing and humanization. Qualitative analysis of the experts' free-form explanations shows that while they rely heavily on specific lexical clues ('AI vocabulary'), they also pick up on more complex phenomena within the text (e.g., formality, originality, clarity) that are challenging to assess for automatic detectors. We release our annotated dataset and code to spur future research into both human and automated detection of AI-generated text.", "score": 0.0, "mention_count": 13, "user_count": 11, "months": ["2025-11", "2025-12", "2026-04", "2026-05", "2026-06", "2026-07", "2026-08"], "last_mentioned": "2026-08-21T20:52:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nclimate3004", "slug": "921ac5dd72336100", "title": "Greening of the Earth and its drivers", "authors": ["Zaichun Zhu", "Shilong Piao", "Ranga B. Myneni", "Mengtian Huang", "Zhenzhong Zeng", "Josep G. Canadell", "Philippe Ciais", "Stephen Sitch", "Pierre Friedlingstein", "Almut Arneth", "Chunxiang Cao", "Lei Cheng", "Etsushi Kato", "Charles Koven", "Yue Li", "Xu Lian", "Yongwen Liu", "Ronggao Liu", "Jiafu Mao", "Yaozhong Pan", "Shushi Peng", "Josep Peñuelas", "Benjamin Poulter", "Thomas A. M. Pugh", "Benjamin D. Stocker", "Nicolas Viovy", "Xuhui Wang", "Yingping Wang", "Zhiqiang Xiao", "Hui Yang", "Sönke Zaehle", "Ning Zeng"], "year": 2016, "venue": "Nature Climate Change", "url": "https://doi.org/10.1038/nclimate3004", "doi": "10.1038/nclimate3004", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 13, "user_count": 8, "months": ["2019-02", "2019-07", "2020-12", "2021-11", "2022-01", "2023-04", "2023-05", "2023-07", "2023-08", "2023-10", "2025-07"], "last_mentioned": "2025-07-23T06:37:35Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.aaw2594", "slug": "8dbecad65e8c9d15", "title": "Different languages, similar encoding efficiency: Comparable information rates across the human communicative niche", "authors": ["Christophe Coupé", "Yoon Mi Oh", "Dan Dediu", "François Pellegrino"], "year": 2019, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.aaw2594", "doi": "10.1126/sciadv.aaw2594", "topics": ["Other research"], "description": "Human languages encode similar average information rates (~39 bits/s) despite their remarkable differences.", "description_method": "abstract-excerpt", "abstract": "Human languages encode similar average information rates (~39 bits/s) despite their remarkable differences.", "score": 0.0, "mention_count": 13, "user_count": 13, "months": ["2022-04", "2022-07", "2023-04", "2023-07", "2023-09", "2023-11", "2025-06", "2025-07", "2025-08", "2026-02", "2026-03"], "last_mentioned": "2026-03-14T18:10:46Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0706.1062", "slug": "7a7ae82ebeb7a54e", "title": "Power-law distributions in empirical data", "authors": ["Aaron Clauset", "Cosma Rohilla Shalizi", "M. E. J. Newman"], "year": 2007, "venue": "arXiv", "url": "https://arxiv.org/abs/0706.1062", "doi": "10.1137/070710111", "topics": ["Statistics & probability", "Psychology", "Science & research"], "description": "Power-law distributions occur in many situations of scientific interest and have significant consequences for our understanding of natural and man-made phenomena.", "description_method": "abstract-excerpt", "abstract": "Power-law distributions occur in many situations of scientific interest and have significant consequences for our understanding of natural and man-made phenomena. Unfortunately, the detection and characterization of power laws is complicated by the large fluctuations that occur in the tail of the distribution -- the part of the distribution representing large but rare events -- and by the difficulty of identifying the range over which power-law behavior holds. Commonly used methods for analyzing power-law data, such as least-squares fitting, can produce substantially inaccurate estimates of parameters for power-law distributions, and even in cases where such methods return accurate answers they are still unsatisfactory because they give no indication of whether the data obey a power law at all. Here we present a principled statistical framework for discerning and quantifying power-law behavior in empirical data. Our approach combines maximum-likelihood fitting methods with goodness-of-fit tests based on the Kolmogorov-Smirnov statistic and likelihood ratios. We evaluate the effectiveness of the approach with tests on synthetic data and give critical comparisons to previous approaches. We also apply the proposed methods to twenty-four real-world data sets from a range of different disciplines, each of which has been conjectured to follow a power-law distribution. In some cases we find these conjectures to be consistent with the data while in others the power law is ruled out.", "score": 0.0, "mention_count": 12, "user_count": 11, "months": ["2009-11", "2011-03", "2013-04", "2014-04", "2016-07", "2016-09", "2017-02", "2018-06", "2023-08", "2025-02", "2026-02"], "last_mentioned": "2026-02-12T17:30:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1311.3903", "slug": "163ede5294ad8171", "title": "A Categorical Theory of Patches", "authors": ["Samuel Mimram", "Cinzia Di Giusto"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1311.3903", "doi": "10.1016/j.entcs.2013.09.018", "topics": ["Economics"], "description": "When working with distant collaborators on the same documents, one often uses a version control system, which is a program tracking the history of files and helping importing modifications brought by others as patches.", "description_method": "abstract-excerpt", "abstract": "When working with distant collaborators on the same documents, one often uses a version control system, which is a program tracking the history of files and helping importing modifications brought by others as patches. The implementation of such a system requires to handle lots of situations depending on the operations performed by users on files, and it is thus difficult to ensure that all the corner cases have been correctly addressed. Here, instead of verifying the implementation of such a system, we adopt a complementary approach: we introduce a theoretical model, which is defined abstractly by the universal property that it should satisfy, and work out a concrete description of it. We begin by defining a category of files and patches, where the operation of merging the effect of two coinitial patches is defined by pushout. Since two patches can be incompatible, such a pushout does not necessarily exist in the category, which raises the question of which is the correct category to represent and manipulate files in conflicting state. We provide an answer by investigating the free completion of the category of files under finite colimits, and give an explicit description of this category: its objects are finite sets labeled by lines equipped with a transitive relation and morphisms are partial functions respecting labeling and relations.", "score": 0.0, "mention_count": 12, "user_count": 12, "months": ["2017-02", "2017-12", "2018-10", "2019-02", "2019-03", "2019-04", "2019-05", "2020-01", "2020-07", "2020-09", "2020-11", "2022-11"], "last_mentioned": "2022-11-11T15:38:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.0613", "slug": "57c58ea048395cd8", "title": "The Habitable Epoch of the Early Universe", "authors": ["Abraham Loeb"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.0613", "doi": "10.1017/S1473550414000196", "topics": ["Physics"], "description": "In the redshift range 100<(1+z)<137, the cosmic microwave background (CMB) had a temperature of 273-373K (0-100 degrees Celsius), allowing early rocky planets (if any existed) to have liquid water chemistry on their surface and be habitable, irrespective of their distance from a star.", "description_method": "abstract-excerpt", "abstract": "In the redshift range 100<(1+z)<137, the cosmic microwave background (CMB) had a temperature of 273-373K (0-100 degrees Celsius), allowing early rocky planets (if any existed) to have liquid water chemistry on their surface and be habitable, irrespective of their distance from a star. In the standard LCDM cosmology, the first star-forming halos within our Hubble volume started collapsing at these redshifts, allowing the chemistry of life to possibly begin when the Universe was merely 10-17 million years old. The possibility of life starting when the average matter density was a million times bigger than it is today argues against the anthropic explanation for the low value of the cosmological constant.", "score": 0.0, "mention_count": 12, "user_count": 11, "months": ["2015-07", "2019-04", "2019-07", "2019-12", "2021-01", "2021-05", "2022-04", "2023-01", "2024-01", "2025-02", "2025-11"], "last_mentioned": "2025-11-27T02:48:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1402.5644", "slug": "adb8865e33a2a332", "title": "Containment Control for a Social Network with State-Dependent Connectivity", "authors": ["Zhen Kan", "Justin Klotz", "Eduardo L. Pasiliao", "Warren E. Dixon"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1402.5644", "doi": null, "topics": ["Psychology"], "description": "Social interactions influence our thoughts, opinions and actions.", "description_method": "abstract-excerpt", "abstract": "Social interactions influence our thoughts, opinions and actions. In this paper, social interactions are studied within a group of individuals composed of influential social leaders and followers. Each person is assumed to maintain a social state, which can be an emotional state or an opinion. Followers update their social states based on the states of local neighbors, while social leaders maintain a constant desired state. Social interactions are modeled as a general directed graph where each directed edge represents an influence from one person to another. Motivated by the non-local property of fractional-order systems, the social response of individuals in the network are modeled by fractional-order dynamics whose states depend on influences from local neighbors and past experiences. A decentralized influence method is then developed to maintain existing social influence between individuals (i.e., without isolating peers in the group) and to influence the social group to a common desired state (i.e., within a convex hull spanned by social leaders). Mittag-Leffler stability methods are used to prove asymptotic stability of the networked fractional-order system.", "score": 0.0, "mention_count": 12, "user_count": 6, "months": ["2018-10", "2021-12", "2022-02", "2022-10", "2022-11", "2023-05", "2023-12", "2024-03", "2024-08", "2025-04", "2025-11"], "last_mentioned": "2025-11-24T01:25:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1412.1897", "slug": "e458ca9d02da9193", "title": "Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Images", "authors": ["Anh Nguyen", "Jason Yosinski", "Jeff Clune"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1412.1897", "doi": null, "topics": ["Machine learning", "Computer vision", "Psychology"], "description": "Deep neural networks (DNNs) have recently been achieving state-of-the-art performance on a variety of pattern-recognition tasks, most notably visual classification problems.", "description_method": "abstract-excerpt", "abstract": "Deep neural networks (DNNs) have recently been achieving state-of-the-art performance on a variety of pattern-recognition tasks, most notably visual classification problems. Given that DNNs are now able to classify objects in images with near-human-level performance, questions naturally arise as to what differences remain between computer and human vision. A recent study revealed that changing an image (e.g. of a lion) in a way imperceptible to humans can cause a DNN to label the image as something else entirely (e.g. mislabeling a lion a library). Here we show a related result: it is easy to produce images that are completely unrecognizable to humans, but that state-of-the-art DNNs believe to be recognizable objects with 99.99% confidence (e.g. labeling with certainty that white noise static is a lion). Specifically, we take convolutional neural networks trained to perform well on either the ImageNet or MNIST datasets and then find images with evolutionary algorithms or gradient ascent that DNNs label with high confidence as belonging to each dataset class. It is possible to produce images totally unrecognizable to human eyes that DNNs believe with near certainty are familiar objects, which we call \"fooling images\" (more generally, fooling examples). Our results shed light on interesting differences between human vision and current DNNs, and raise questions about the generality of DNN computer vision.", "score": 0.0, "mention_count": 12, "user_count": 12, "months": ["2014-12", "2015-03", "2015-06", "2016-02", "2016-05", "2016-06", "2016-10", "2017-04", "2017-07"], "last_mentioned": "2017-07-15T02:05:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1512.01895", "slug": "1e4ce4eba61d6308", "title": "Modular implicits", "authors": ["Leo White", "Frédéric Bour", "Jeremy Yallop"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1512.01895", "doi": "10.4204/EPTCS.198.2", "topics": ["Economics"], "description": "We present modular implicits, an extension to the OCaml language for ad-hoc polymorphism inspired by Scala implicits and modular type classes.", "description_method": "abstract-excerpt", "abstract": "We present modular implicits, an extension to the OCaml language for ad-hoc polymorphism inspired by Scala implicits and modular type classes. Modular implicits are based on type-directed implicit module parameters, and elaborate straightforwardly into OCaml's first-class functors. Basing the design on OCaml's modules leads to a system that naturally supports many features from other languages with systematic ad-hoc overloading, including inheritance, instance constraints, constructor classes and associated types.", "score": 0.0, "mention_count": 12, "user_count": 12, "months": ["2016-03", "2016-04", "2018-04", "2018-06", "2018-08", "2019-09", "2021-01", "2021-08", "2021-12", "2024-07", "2025-08"], "last_mentioned": "2025-08-13T23:14:31Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1602.04938", "slug": "cf506c9bf4fc6a0f", "title": "\"Why Should I Trust You?\": Explaining the Predictions of Any Classifier", "authors": ["Marco Tulio Ribeiro", "Sameer Singh", "Carlos Guestrin"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1602.04938", "doi": null, "topics": ["Machine learning", "Computer vision", "Algorithms"], "description": "Despite widespread adoption, machine learning models remain mostly black boxes.", "description_method": "abstract-excerpt", "abstract": "Despite widespread adoption, machine learning models remain mostly black boxes. Understanding the reasons behind predictions is, however, quite important in assessing trust, which is fundamental if one plans to take action based on a prediction, or when choosing whether to deploy a new model. Such understanding also provides insights into the model, which can be used to transform an untrustworthy model or prediction into a trustworthy one. In this work, we propose LIME, a novel explanation technique that explains the predictions of any classifier in an interpretable and faithful manner, by learning an interpretable model locally around the prediction. We also propose a method to explain models by presenting representative individual predictions and their explanations in a non-redundant way, framing the task as a submodular optimization problem. We demonstrate the flexibility of these methods by explaining different models for text (e.g. random forests) and image classification (e.g. neural networks). We show the utility of explanations via novel experiments, both simulated and with human subjects, on various scenarios that require trust: deciding if one should trust a prediction, choosing between models, improving an untrustworthy classifier, and identifying why a classifier should not be trusted.", "score": 0.0, "mention_count": 12, "user_count": 11, "months": ["2016-09", "2017-03", "2017-07", "2017-08", "2017-12", "2018-03", "2019-07", "2020-04", "2020-11", "2024-06"], "last_mentioned": "2024-06-05T12:58:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1605.05274", "slug": "8739f2fce0b22295", "title": "Java Generics are Turing Complete", "authors": ["Radu Grigore"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1605.05274", "doi": null, "topics": ["Other research"], "description": "This paper describes a reduction from the halting problem of Turing machines to subtype checking in Java.", "description_method": "abstract-excerpt", "abstract": "This paper describes a reduction from the halting problem of Turing machines to subtype checking in Java. It follows that subtype checking in Java is undecidable, which answers a question posed by Kennedy and Pierce in 2007. It also follows that Java's type checker can recognize any recursive language, which improves a result of Gil and Levy from 2016. The latter point is illustrated by a parser generator for fluent interfaces.", "score": 0.0, "mention_count": 12, "user_count": 11, "months": ["2017-08", "2017-10", "2018-05", "2019-04", "2019-06", "2020-05", "2021-01", "2021-07", "2021-12", "2022-10", "2023-09"], "last_mentioned": "2023-09-18T16:10:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1905.09749", "slug": "a5301de3b4b11f89", "title": "How to factor 2048 bit RSA integers in 8 hours using 20 million noisy qubits", "authors": ["Craig Gidney", "Martin Ekerå"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1905.09749", "doi": "10.22331/q-2021-04-15-433", "topics": ["Physics", "Security & cryptography", "Networks"], "description": "We significantly reduce the cost of factoring integers and computing discrete logarithms in finite fields on a quantum computer by combining techniques from Shor 1994, Griffiths-Niu 1996, Zalka 2006, Fowler 2012, Ekerå-Håstad 2017, Ekerå 2017, Ekerå 2018, Gidney-Fowler 2019, Gidney 2019.", "description_method": "abstract-excerpt", "abstract": "We significantly reduce the cost of factoring integers and computing discrete logarithms in finite fields on a quantum computer by combining techniques from Shor 1994, Griffiths-Niu 1996, Zalka 2006, Fowler 2012, Ekerå-Håstad 2017, Ekerå 2017, Ekerå 2018, Gidney-Fowler 2019, Gidney 2019. We estimate the approximate cost of our construction using plausible physical assumptions for large-scale superconducting qubit platforms: a planar grid of qubits with nearest-neighbor connectivity, a characteristic physical gate error rate of $10^{-3}$, a surface code cycle time of 1 microsecond, and a reaction time of 10 microseconds. We account for factors that are normally ignored such as noise, the need to make repeated attempts, and the spacetime layout of the computation. When factoring 2048 bit RSA integers, our construction's spacetime volume is a hundredfold less than comparable estimates from earlier works (Van Meter et al. 2009, Jones et al. 2010, Fowler et al. 2012, Gheorghiu et al. 2019). In the abstract circuit model (which ignores overheads from distillation, routing, and error correction) our construction uses $3 n + 0.002 n \\lg n$ logical qubits, $0.3 n^3 + 0.0005 n^3 \\lg n$ Toffolis, and $500 n^2 + n^2 \\lg n$ measurement depth to factor $n$-bit RSA integers. We quantify the cryptographic implications of our work, both for RSA and for schemes based on the DLP in finite fields.", "score": 0.0, "mention_count": 12, "user_count": 10, "months": ["2019-10", "2020-01", "2020-05", "2021-11", "2023-01", "2023-09", "2024-09", "2024-12", "2025-01", "2025-04", "2025-11", "2026-03"], "last_mentioned": "2026-03-31T21:14:39Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2005.11401", "slug": "8fc39e0ec1b6db48", "title": "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks", "authors": ["Patrick Lewis", "Ethan Perez", "Aleksandra Piktus", "Fabio Petroni", "Vladimir Karpukhin", "Naman Goyal", "Heinrich Küttler", "Mike Lewis", "Wen-tau Yih", "Tim Rocktäschel", "Sebastian Riedel", "Douwe Kiela"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2005.11401", "doi": null, "topics": ["Language models"], "description": "Large pre-trained language models have been shown to store factual knowledge in their parameters, and achieve state-of-the-art results when fine-tuned on downstream NLP tasks.", "description_method": "abstract-excerpt", "abstract": "Large pre-trained language models have been shown to store factual knowledge in their parameters, and achieve state-of-the-art results when fine-tuned on downstream NLP tasks. However, their ability to access and precisely manipulate knowledge is still limited, and hence on knowledge-intensive tasks, their performance lags behind task-specific architectures. Additionally, providing provenance for their decisions and updating their world knowledge remain open research problems. Pre-trained models with a differentiable access mechanism to explicit non-parametric memory can overcome this issue, but have so far been only investigated for extractive downstream tasks. We explore a general-purpose fine-tuning recipe for retrieval-augmented generation (RAG) -- models which combine pre-trained parametric and non-parametric memory for language generation. We introduce RAG models where the parametric memory is a pre-trained seq2seq model and the non-parametric memory is a dense vector index of Wikipedia, accessed with a pre-trained neural retriever. We compare two RAG formulations, one which conditions on the same retrieved passages across the whole generated sequence, the other can use different passages per token. We fine-tune and evaluate our models on a wide range of knowledge-intensive NLP tasks and set the state-of-the-art on three open domain QA tasks, outperforming parametric seq2seq models and task-specific retrieve-and-extract architectures. For language generation tasks, we find that RAG models generate more specific, diverse and factual language than a state-of-the-art parametric-only seq2seq baseline.", "score": 0.0, "mention_count": 12, "user_count": 12, "months": ["2023-01", "2023-05", "2023-06", "2023-07", "2023-08", "2023-10", "2023-11", "2023-12", "2024-03", "2024-10"], "last_mentioned": "2024-10-16T02:04:19Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2006.16236", "slug": "d119844a2dce8a3b", "title": "Transformers are RNNs: Fast Autoregressive Transformers with Linear Attention", "authors": ["Angelos Katharopoulos", "Apoorv Vyas", "Nikolaos Pappas", "François Fleuret"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2006.16236", "doi": null, "topics": ["Language models", "Algorithms", "Machine learning"], "description": "Transformers achieve remarkable performance in several tasks but due to their quadratic complexity, with respect to the input's length, they are prohibitively slow for very long sequences.", "description_method": "abstract-excerpt", "abstract": "Transformers achieve remarkable performance in several tasks but due to their quadratic complexity, with respect to the input's length, they are prohibitively slow for very long sequences. To address this limitation, we express the self-attention as a linear dot-product of kernel feature maps and make use of the associativity property of matrix products to reduce the complexity from $\\mathcal{O}\\left(N^2\\right)$ to $\\mathcal{O}\\left(N\\right)$, where $N$ is the sequence length. We show that this formulation permits an iterative implementation that dramatically accelerates autoregressive transformers and reveals their relationship to recurrent neural networks. Our linear transformers achieve similar performance to vanilla transformers and they are up to 4000x faster on autoregressive prediction of very long sequences.", "score": 0.0, "mention_count": 12, "user_count": 8, "months": ["2020-07", "2023-01", "2023-04", "2024-06", "2024-07", "2025-02", "2026-02"], "last_mentioned": "2026-02-04T17:41:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2405.08007", "slug": "ec6cede2a895468a", "title": "People cannot distinguish GPT-4 from a human in a Turing test", "authors": ["Cameron R. Jones", "Benjamin K. Bergen"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2405.08007", "doi": null, "topics": ["Language models", "Psychology"], "description": "We evaluated 3 systems (ELIZA, GPT-3.5 and GPT-4) in a randomized, controlled, and preregistered Turing test.", "description_method": "abstract-excerpt", "abstract": "We evaluated 3 systems (ELIZA, GPT-3.5 and GPT-4) in a randomized, controlled, and preregistered Turing test. Human participants had a 5 minute conversation with either a human or an AI, and judged whether or not they thought their interlocutor was human. GPT-4 was judged to be a human 54% of the time, outperforming ELIZA (22%) but lagging behind actual humans (67%). The results provide the first robust empirical demonstration that any artificial system passes an interactive 2-player Turing test. The results have implications for debates around machine intelligence and, more urgently, suggest that deception by current AI systems may go undetected. Analysis of participants' strategies and reasoning suggests that stylistic and socio-emotional factors play a larger role in passing the Turing test than traditional notions of intelligence.", "score": 0.0, "mention_count": 12, "user_count": 10, "months": ["2024-08", "2024-09", "2024-12", "2025-01", "2025-02", "2025-04", "2025-07", "2025-10", "2026-02", "2026-03", "2026-05"], "last_mentioned": "2026-05-04T02:40:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature25476", "slug": "f79df9bbd8224f04", "title": "Processing bulk natural wood into a high-performance structural material", "authors": ["Jianwei Song", "Chaoji Chen", "Shuze Zhu", "Mingwei Zhu", "Jiaqi Dai", "Upamanyu Ray", "Yiju Li", "Yudi Kuang", "Yongfeng Li", "Nelson Quispe", "Yonggang Yao", "Amy Gong", "Ulrich H. Leiste", "Hugh A. Bruck", "J. Y. Zhu", "Azhar Vellore", "Heng Li", "Marilyn L. Minus", "Zheng Jia", "Ashlie Martini", "Teng Li", "Liangbing Hu"], "year": 2018, "venue": "Nature", "url": "https://doi.org/10.1038/nature25476", "doi": "10.1038/nature25476", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 12, "user_count": 12, "months": ["2018-02", "2020-06", "2023-11", "2025-05", "2026-01"], "last_mentioned": "2026-01-29T09:50:55Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41579-022-00846-2", "slug": "4ceb22fcee750db0", "title": "Long COVID: major findings, mechanisms and recommendations", "authors": ["Hannah E. Davis", "Lisa McCorkell", "Julia Moore Vogel", "Eric J. Topol"], "year": 2023, "venue": "Nature Reviews Microbiology", "url": "https://doi.org/10.1038/s41579-022-00846-2", "doi": "10.1038/s41579-022-00846-2", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 12, "user_count": 9, "months": ["2023-03", "2023-04", "2023-05", "2023-06", "2023-07", "2023-08", "2023-09", "2023-12", "2024-07", "2025-07"], "last_mentioned": "2025-07-31T16:48:50Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-024-07566-y", "slug": "1bd0f3e40333ec40", "title": "AI models collapse when trained on recursively generated data", "authors": ["Ilia Shumailov", "Zakhar Shumaylov", "Yiren Zhao", "Nicolas Papernot", "Ross Anderson", "Yarin Gal"], "year": 2024, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-024-07566-y", "doi": "10.1038/s41586-024-07566-y", "topics": ["Language models", "Psychology", "Physics"], "description": "Abstract Stable diffusion revolutionized image creation from descriptive text.", "description_method": "abstract-excerpt", "abstract": "Abstract Stable diffusion revolutionized image creation from descriptive text. GPT-2 (ref. 1 ), GPT-3(.5) (ref. 2 ) and GPT-4 (ref. 3 ) demonstrated high performance across a variety of language tasks. ChatGPT introduced such language models to the public. It is now clear that generative artificial intelligence (AI) such as large language models (LLMs) is here to stay and will substantially change the ecosystem of online text and images. Here we consider what may happen to GPT-{ n } once LLMs contribute much of the text found online. We find that indiscriminate use of model-generated content in training causes irreversible defects in the resulting models, in which tails of the original content distribution disappear. We refer to this effect as ‘model collapse’ and show that it can occur in LLMs as well as in variational autoencoders (VAEs) and Gaussian mixture models (GMMs). We build theoretical intuition behind the phenomenon and portray its ubiquity among all learned generative models. We demonstrate that it must be taken seriously if we are to sustain the benefits of training from large-scale data scraped from the web. Indeed, the value of data collected about genuine human interactions with systems will be increasingly valuable in the presence of LLM-generated content in data crawled from the Internet.", "score": 0.0, "mention_count": 12, "user_count": 12, "months": ["2024-07", "2024-10", "2025-02", "2025-03", "2025-06", "2025-07", "2025-09", "2025-11", "2026-04", "2026-05", "2026-07"], "last_mentioned": "2026-07-07T08:13:27Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "url:https://proceedings.mlr.press/v81/buolamwini18a.html", "slug": "4c85dc2563e43eaf", "title": "Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification", "authors": ["Joy Buolamwini", "Timnit Gebru"], "year": 2018, "venue": "Conference on Fairness, Accountability and Transparency", "url": "https://proceedings.mlr.press/v81/buolamwini18a.html", "doi": null, "topics": ["Psychology", "Machine learning", "Algorithms"], "description": "Recent studies demonstrate that machine learning algorithms can discriminate based on classes like race and gender.", "description_method": "abstract-excerpt", "abstract": "Recent studies demonstrate that machine learning algorithms can discriminate based on classes like race and gender. In this work, we present an approach to evaluate bias present in automated facial...", "score": 0.0, "mention_count": 12, "user_count": 11, "months": ["2018-03", "2018-06", "2019-04", "2020-06", "2020-07", "2023-01", "2023-05", "2023-08"], "last_mentioned": "2023-08-06T22:24:07Z", "metadata_source": "publisher-citation-tags", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1311.0243", "slug": "759d83bdf8f75c15", "title": "Majority is not Enough: Bitcoin Mining is Vulnerable", "authors": ["Ittay Eyal", "Emin Gun Sirer"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1311.0243", "doi": null, "topics": ["Security & cryptography", "Economics", "Distributed systems"], "description": "The Bitcoin cryptocurrency records its transactions in a public log called the blockchain.", "description_method": "abstract-excerpt", "abstract": "The Bitcoin cryptocurrency records its transactions in a public log called the blockchain. Its security rests critically on the distributed protocol that maintains the blockchain, run by participants called miners. Conventional wisdom asserts that the protocol is incentive-compatible and secure against colluding minority groups, i.e., it incentivizes miners to follow the protocol as prescribed. We show that the Bitcoin protocol is not incentive-compatible. We present an attack with which colluding miners obtain a revenue larger than their fair share. This attack can have significant consequences for Bitcoin: Rational miners will prefer to join the selfish miners, and the colluding group will increase in size until it becomes a majority. At this point, the Bitcoin system ceases to be a decentralized currency. Selfish mining is feasible for any group size of colluding miners. We propose a practical modification to the Bitcoin protocol that protects against selfish mining pools that command less than 1/4 of the resources. This threshold is lower than the wrongly assumed 1/2 bound, but better than the current reality where a group of any size can compromise the system.", "score": 0.0, "mention_count": 11, "user_count": 10, "months": ["2013-12", "2014-01", "2014-06", "2014-07", "2015-07", "2015-10", "2017-09", "2018-04"], "last_mentioned": "2018-04-22T20:39:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1408.3421", "slug": "c46def498f04aed9", "title": "On the genetic architecture of intelligence and other quantitative traits", "authors": ["Stephen D. H. Hsu"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1408.3421", "doi": null, "topics": ["Psychology", "Statistics & probability", "Neuroscience"], "description": "How do genes affect cognitive ability or other human quantitative traits such as height or disease risk?", "description_method": "abstract-excerpt", "abstract": "How do genes affect cognitive ability or other human quantitative traits such as height or disease risk? Progress on this challenging question is likely to be significant in the near future. I begin with a brief review of psychometric measurements of intelligence, introducing the idea of a \"general factor\" or g score. The main results concern the stability, validity (predictive power), and heritability of adult g. The largest component of genetic variance for both height and intelligence is additive (linear), leading to important simplifications in predictive modeling and statistical estimation. Due mainly to the rapidly decreasing cost of genotyping, it is possible that within the coming decade researchers will identify loci which account for a significant fraction of total g variation. In the case of height analogous efforts are well under way. I describe some unpublished results concerning the genetic architecture of height and cognitive ability, which suggest that roughly 10k moderately rare causal variants of mostly negative effect are responsible for normal population variation. Using results from Compressed Sensing (L1-penalized regression), I estimate the statistical power required to characterize both linear and nonlinear models for quantitative traits. The main unknown parameter s (sparsity) is the number of loci which account for the bulk of the genetic variation. The required sample size is of order 100s, or roughly a million in the case of cognitive ability.", "score": 0.0, "mention_count": 11, "user_count": 9, "months": ["2014-10", "2015-12", "2016-03", "2016-04", "2017-03", "2017-05", "2017-07", "2018-11", "2020-07"], "last_mentioned": "2020-07-30T00:50:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1607.02318", "slug": "31a36af55e7ac83d", "title": "The Renewed Case for the Reduced Instruction Set Computer: Avoiding ISA Bloat with Macro-Op Fusion for RISC-V", "authors": ["Christopher Celio", "Palmer Dabbelt", "David A. Patterson", "Krste Asanović"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1607.02318", "doi": null, "topics": ["Physics", "Programming languages"], "description": "This report makes the case that a well-designed Reduced Instruction Set Computer (RISC) can match, and even exceed, the performance and code density of existing commercial Complex Instruction Set Computers (CISC) while maintaining the simplicity and cost-effectiveness that underpins the original RISC goals.", "description_method": "abstract-excerpt", "abstract": "This report makes the case that a well-designed Reduced Instruction Set Computer (RISC) can match, and even exceed, the performance and code density of existing commercial Complex Instruction Set Computers (CISC) while maintaining the simplicity and cost-effectiveness that underpins the original RISC goals. We begin by comparing the dynamic instruction counts and dynamic instruction bytes fetched for the popular proprietary ARMv7, ARMv8, IA-32, and x86-64 Instruction Set Architectures (ISAs) against the free and open RISC-V RV64G and RV64GC ISAs when running the SPEC CINT2006 benchmark suite. RISC-V was designed as a very small ISA to support a wide range of implementations, and has a less mature compiler toolchain. However, we observe that on SPEC CINT2006 RV64G executes on average 16% more instructions than x86-64, 3% more instructions than IA-32, 9% more instructions than ARMv8, but 4% fewer instructions than ARMv7. CISC x86 implementations break up complex instructions into smaller internal RISC-like micro-ops, and the RV64G instruction count is within 2% of the x86-64 retired micro-op count. RV64GC, the compressed variant of RV64G, is the densest ISA studied, fetching 8% fewer dynamic instruction bytes than x86-64. We observed that much of the increased RISC-V instruction count is due to a small set of common multi-instruction idioms. Exploiting this fact, the RV64G and RV64GC effective instruction count can be reduced by 5.4% on average by leveraging macro-op fusion. Combining the compressed RISC-V ISA extension with macro-op fusion provides both the densest ISA and the fewest dynamic operations retired per program, reducing the motivation to add more instructions to the ISA. This approach retains a single simple ISA suitable for both low-end and high-end implementations, where high-end implementations can boost performance through microarchitectural techniques.", "score": 0.0, "mention_count": 11, "user_count": 8, "months": ["2016-07", "2016-12", "2017-09", "2017-12", "2022-03", "2022-05", "2022-11", "2025-01"], "last_mentioned": "2025-01-03T18:50:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1712.01815", "slug": "993179c47eaf4928", "title": "Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm", "authors": ["David Silver", "Thomas Hubert", "Julian Schrittwieser", "Ioannis Antonoglou", "Matthew Lai", "Arthur Guez", "Marc Lanctot", "Laurent Sifre", "Dharshan Kumaran", "Thore Graepel", "Timothy Lillicrap", "Karen Simonyan", "Demis Hassabis"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1712.01815", "doi": null, "topics": ["Machine learning", "Algorithms", "Psychology"], "description": "The game of chess is the most widely-studied domain in the history of artificial intelligence.", "description_method": "abstract-excerpt", "abstract": "The game of chess is the most widely-studied domain in the history of artificial intelligence. The strongest programs are based on a combination of sophisticated search techniques, domain-specific adaptations, and handcrafted evaluation functions that have been refined by human experts over several decades. In contrast, the AlphaGo Zero program recently achieved superhuman performance in the game of Go, by tabula rasa reinforcement learning from games of self-play. In this paper, we generalise this approach into a single AlphaZero algorithm that can achieve, tabula rasa, superhuman performance in many challenging domains. Starting from random play, and given no domain knowledge except the game rules, AlphaZero achieved within 24 hours a superhuman level of play in the games of chess and shogi (Japanese chess) as well as Go, and convincingly defeated a world-champion program in each case.", "score": 0.0, "mention_count": 11, "user_count": 11, "months": ["2017-12", "2018-01", "2018-02", "2018-05", "2019-11", "2023-02", "2023-04", "2023-07", "2025-05"], "last_mentioned": "2025-05-03T07:19:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2106.06981", "slug": "4e3f633988378ba2", "title": "Thinking Like Transformers", "authors": ["Gail Weiss", "Yoav Goldberg", "Eran Yahav"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2106.06981", "doi": null, "topics": ["Language models", "Programming languages", "Machine learning"], "description": "What is the computational model behind a Transformer?", "description_method": "abstract-excerpt", "abstract": "What is the computational model behind a Transformer? Where recurrent neural networks have direct parallels in finite state machines, allowing clear discussion and thought around architecture variants or trained models, Transformers have no such familiar parallel. In this paper we aim to change that, proposing a computational model for the transformer-encoder in the form of a programming language. We map the basic components of a transformer-encoder -- attention and feed-forward computation -- into simple primitives, around which we form a programming language: the Restricted Access Sequence Processing Language (RASP). We show how RASP can be used to program solutions to tasks that could conceivably be learned by a Transformer, and how a Transformer can be trained to mimic a RASP solution. In particular, we provide RASP programs for histograms, sorting, and Dyck-languages. We further use our model to relate their difficulty in terms of the number of required layers and attention heads: analyzing a RASP program implies a maximum number of heads and layers necessary to encode a task in a transformer. Finally, we see how insights gained from our abstraction might be used to explain phenomena seen in recent works.", "score": 0.0, "mention_count": 11, "user_count": 9, "months": ["2021-12", "2022-10", "2023-05", "2023-06", "2023-08", "2023-09", "2024-04", "2025-11", "2026-02"], "last_mentioned": "2026-02-16T10:35:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.03860", "slug": "0d1475b528e9a024", "title": "Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Models", "authors": ["Gowthami Somepalli", "Vasu Singla", "Micah Goldblum", "Jonas Geiping", "Tom Goldstein"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.03860", "doi": null, "topics": ["Physics", "Machine learning", "Distributed systems"], "description": "Cutting-edge diffusion models produce images with high quality and customizability, enabling them to be used for commercial art and graphic design purposes.", "description_method": "abstract-excerpt", "abstract": "Cutting-edge diffusion models produce images with high quality and customizability, enabling them to be used for commercial art and graphic design purposes. But do diffusion models create unique works of art, or are they replicating content directly from their training sets? In this work, we study image retrieval frameworks that enable us to compare generated images with training samples and detect when content has been replicated. Applying our frameworks to diffusion models trained on multiple datasets including Oxford flowers, Celeb-A, ImageNet, and LAION, we discuss how factors such as training set size impact rates of content replication. We also identify cases where diffusion models, including the popular Stable Diffusion model, blatantly copy from their training data.", "score": 0.0, "mention_count": 11, "user_count": 10, "months": ["2022-12", "2023-01", "2023-02", "2024-07"], "last_mentioned": "2024-07-03T01:20:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2304.14717", "slug": "067ac44c8caf919d", "title": "faulTPM: Exposing AMD fTPMs' Deepest Secrets", "authors": ["Hans Niklas Jacob", "Christian Werling", "Robert Buhren", "Jean-Pierre Seifert"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2304.14717", "doi": null, "topics": ["Security & cryptography", "Computer systems", "Algorithms"], "description": "Trusted Platform Modules constitute an integral building block of modern security features.", "description_method": "abstract-excerpt", "abstract": "Trusted Platform Modules constitute an integral building block of modern security features. Moreover, as Windows 11 made a TPM 2.0 mandatory, they are subject to an ever-increasing academic challenge. While discrete TPMs - as found in higher-end systems - have been susceptible to attacks on their exposed communication interface, more common firmware TPMs (fTPMs) are immune to this attack vector as they do not communicate with the CPU via an exposed bus. In this paper, we analyze a new class of attacks against fTPMs: Attacking their Trusted Execution Environment can lead to a full TPM state compromise. We experimentally verify this attack by compromising the AMD Secure Processor, which constitutes the TEE for AMD's fTPMs. In contrast to previous dTPM sniffing attacks, this vulnerability exposes the complete internal TPM state of the fTPM. It allows us to extract any cryptographic material stored or sealed by the fTPM regardless of authentication mechanisms such as Platform Configuration Register validation or passphrases with anti-hammering protection. First, we demonstrate the impact of our findings by - to the best of our knowledge - enabling the first attack against Full Disk Encryption solutions backed by an fTPM. Furthermore, we lay out how any application relying solely on the security properties of the TPM - like Bitlocker's TPM- only protector - can be defeated by an attacker with 2-3 hours of physical access to the target device. Lastly, we analyze the impact of our attack on FDE solutions protected by a TPM and PIN strategy. While a naive implementation also leaves the disk completely unprotected, we find that BitLocker's FDE implementation withholds some protection depending on the complexity of the used PIN. Our results show that when an fTPM's internal state is compromised, a TPM and PIN strategy for FDE is less secure than TPM-less protection with a reasonable passphrase.", "score": 0.0, "mention_count": 11, "user_count": 4, "months": ["2023-07", "2023-08", "2023-11", "2026-01", "2026-05"], "last_mentioned": "2026-05-17T18:51:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.11453", "slug": "5184589093dcf78a", "title": "BitNet: Scaling 1-bit Transformers for Large Language Models", "authors": ["Hongyu Wang", "Shuming Ma", "Li Dong", "Shaohan Huang", "Huaijie Wang", "Lingxiao Ma", "Fan Yang", "Ruiping Wang", "Yi Wu", "Furu Wei"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.11453", "doi": null, "topics": ["Language models"], "description": "The increasing size of large language models has posed challenges for deployment and raised concerns about environmental impact due to high energy consumption.", "description_method": "abstract-excerpt", "abstract": "The increasing size of large language models has posed challenges for deployment and raised concerns about environmental impact due to high energy consumption. In this work, we introduce BitNet, a scalable and stable 1-bit Transformer architecture designed for large language models. Specifically, we introduce BitLinear as a drop-in replacement of the nn.Linear layer in order to train 1-bit weights from scratch. Experimental results on language modeling show that BitNet achieves competitive performance while substantially reducing memory footprint and energy consumption, compared to state-of-the-art 8-bit quantization methods and FP16 Transformer baselines. Furthermore, BitNet exhibits a scaling law akin to full-precision Transformers, suggesting its potential for effective scaling to even larger language models while maintaining efficiency and performance benefits.", "score": 0.0, "mention_count": 11, "user_count": 10, "months": ["2024-02", "2024-03", "2024-06", "2024-11", "2025-04", "2025-05", "2025-06", "2026-03", "2026-07"], "last_mentioned": "2026-07-28T05:53:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2602.11988", "slug": "3a3cb6b676d84987", "title": "Evaluating AGENTS.md: Are Repository-Level Context Files Helpful for Coding Agents?", "authors": ["Thibaud Gloaguen", "Niels Mündler-Sasahara", "Mark Niklas Müller", "Veselin Raychev", "Martin Vechev"], "year": 2026, "venue": "arXiv", "url": "https://arxiv.org/abs/2602.11988", "doi": null, "topics": ["Language models"], "description": "A widespread practice in software development is to tailor coding agents to repositories using context files, such as AGENTS.md.", "description_method": "abstract-excerpt", "abstract": "A widespread practice in software development is to tailor coding agents to repositories using context files, such as AGENTS.md. Although this practice is strongly encouraged by agent developers, there is currently no rigorous investigation into whether such context files are actually effective for real-world tasks. In this work, we study this question and evaluate coding agents' task completion performance in two complementary settings: established SWE-bench tasks from popular repositories, with LLM-generated context files, and a novel collection of issues from repositories containing developer-committed context files. Surprisingly, we find that providing context files does not generally improve task success rates, while increasing inference cost by over 20% on average. This observation holds across different LLMs, coding agents, and for both LLM-generated and developer-committed context files. Specifically, we find that while instructions in the context files are well followed by coding agents, repository overviews, although popular and recommended by model providers, are not helpful. We conclude that while context files are useful for specifying non-standard coding practices, any attempts to improve performance should be rigorously evaluated before deployment.", "score": 0.0, "mention_count": 11, "user_count": 11, "months": ["2026-02", "2026-03", "2026-04", "2026-06", "2026-08"], "last_mentioned": "2026-08-20T04:13:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/483531a", "slug": "6be3c90ca137af17", "title": "Raise standards for preclinical cancer research", "authors": ["C. Glenn Begley", "Lee M. Ellis"], "year": 2012, "venue": "Nature", "url": "https://doi.org/10.1038/483531a", "doi": "10.1038/483531a", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 11, "user_count": 8, "months": ["2015-11", "2018-08", "2018-12", "2019-06", "2021-09", "2021-12", "2022-06", "2023-09", "2024-05", "2025-03"], "last_mentioned": "2025-03-23T16:06:57Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41422-020-0282-0", "slug": "a57d7057ecb9d9e2", "title": "Remdesivir and chloroquine effectively inhibit the recently emerged novel coronavirus (2019-nCoV) in vitro", "authors": ["Manli Wang", "Ruiyuan Cao", "Leike Zhang", "Xinglou Yang", "Jia Liu", "Mingyue Xu", "Zhengli Shi", "Zhihong Hu", "Wu Zhong", "Gengfu Xiao"], "year": 2020, "venue": "Cell Research", "url": "https://doi.org/10.1038/s41422-020-0282-0", "doi": "10.1038/s41422-020-0282-0", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 11, "user_count": 8, "months": ["2020-02", "2020-03", "2020-04", "2022-12"], "last_mentioned": "2022-12-29T22:11:41Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abm0620", "slug": "1505023969d69b61", "title": "SARS-CoV-2 vaccine protection and deaths among US veterans during 2021", "authors": ["Barbara A. Cohn", "Piera M. Cirillo", "Caitlin C. Murphy", "Nickilou Y. Krigbaum", "Arthur W. Wallace"], "year": 2022, "venue": "Science", "url": "https://doi.org/10.1126/science.abm0620", "doi": "10.1126/science.abm0620", "topics": ["Biology & medicine"], "description": "We report severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccine effectiveness against infection (VE-I) and death (VE-D) by vaccine type in 780,225 veterans in the Veterans Health Administration, covering 2.7% of the US population.", "description_method": "abstract-excerpt", "abstract": "We report severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccine effectiveness against infection (VE-I) and death (VE-D) by vaccine type in 780,225 veterans in the Veterans Health Administration, covering 2.7% of the US population. From February to October 2021, VE-I declined for all vaccine types, and the decline was greatest for the Janssen vaccine, resulting in a VE-I of 13.1%. Although breakthrough infection increased risk of death, vaccination remained protective against death in persons who became infected during the Delta variant surge. From July to October 2021, VE-D for age <65 years was 73.0% for Janssen, 81.5% for Moderna, and 84.3% for Pfizer-BioNTech; VE-D for age ≥65 years was 52.2% for Janssen, 75.5% for Moderna, and 70.1% for Pfizer-BioNTech. Findings support continued efforts to increase vaccination, booster campaigns, and multiple additional layers of protection against infection.", "score": 0.0, "mention_count": 11, "user_count": 8, "months": ["2021-11", "2021-12", "2022-02"], "last_mentioned": "2022-02-11T00:08:51Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abq7487", "slug": "9f00870921934097", "title": "Genomic inference of a severe human bottleneck during the Early to Middle Pleistocene transition", "authors": ["Wangjie Hu", "Ziqian Hao", "Pengyuan Du", "Fabio Di Vincenzo", "Giorgio Manzi", "Jialong Cui", "Yun-Xin Fu", "Yi-Hsuan Pan", "Haipeng Li"], "year": 2023, "venue": "Science", "url": "https://doi.org/10.1126/science.abq7487", "doi": "10.1126/science.abq7487", "topics": ["Biology & medicine"], "description": "Population size history is essential for studying human evolution.", "description_method": "abstract-excerpt", "abstract": "Population size history is essential for studying human evolution. However, ancient population size history during the Pleistocene is notoriously difficult to unravel. In this study, we developed a fast infinitesimal time coalescent process (FitCoal) to circumvent this difficulty and calculated the composite likelihood for present-day human genomic sequences of 3154 individuals. Results showed that human ancestors went through a severe population bottleneck with about 1280 breeding individuals between around 930,000 and 813,000 years ago. The bottleneck lasted for about 117,000 years and brought human ancestors close to extinction. This bottleneck is congruent with a substantial chronological gap in the available African and Eurasian fossil record. Our results provide new insights into our ancestry and suggest a coincident speciation event.", "score": 0.0, "mention_count": 11, "user_count": 11, "months": ["2023-08", "2023-09", "2023-12", "2024-04", "2024-10", "2025-05", "2025-07", "2025-09", "2026-05"], "last_mentioned": "2026-05-03T09:44:52Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1404.7828", "slug": "aa32da249c371bb5", "title": "Deep Learning in Neural Networks: An Overview", "authors": ["Juergen Schmidhuber"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1404.7828", "doi": "10.1016/j.neunet.2014.09.003", "topics": ["Machine learning", "Psychology"], "description": "In recent years, deep artificial neural networks (including recurrent ones) have won numerous contests in pattern recognition and machine learning.", "description_method": "abstract-excerpt", "abstract": "In recent years, deep artificial neural networks (including recurrent ones) have won numerous contests in pattern recognition and machine learning. This historical survey compactly summarises relevant work, much of it from the previous millennium. Shallow and deep learners are distinguished by the depth of their credit assignment paths, which are chains of possibly learnable, causal links between actions and effects. I review deep supervised learning (also recapitulating the history of backpropagation), unsupervised learning, reinforcement learning & evolutionary computation, and indirect search for short programs encoding deep and large networks.", "score": 0.0, "mention_count": 10, "user_count": 10, "months": ["2014-12", "2015-06", "2016-09", "2017-01", "2017-09", "2018-07", "2019-10", "2023-11", "2024-09"], "last_mentioned": "2024-09-26T18:33:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1602.00602", "slug": "0a4695978dd0433c", "title": "Virtual Machine Warmup Blows Hot and Cold", "authors": ["Edd Barrett", "Carl Friedrich Bolz-Tereick", "Rebecca Killick", "Sarah Mount", "Laurence Tratt"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1602.00602", "doi": "10.1145/3133876", "topics": ["Computer systems", "Statistics & probability", "Programming languages"], "description": "Virtual Machines (VMs) with Just-In-Time (JIT) compilers are traditionally thought to execute programs in two phases: the initial warmup phase determines which parts of a program would most benefit from dynamic compilation, before JIT compiling those parts into machine code; subsequently the program is said to be at a steady state of peak", "description_method": "abstract-excerpt", "abstract": "Virtual Machines (VMs) with Just-In-Time (JIT) compilers are traditionally thought to execute programs in two phases: the initial warmup phase determines which parts of a program would most benefit from dynamic compilation, before JIT compiling those parts into machine code; subsequently the program is said to be at a steady state of peak performance. Measurement methodologies almost always discard data collected during the warmup phase such that reported measurements focus entirely on peak performance. We introduce a fully automated statistical approach, based on changepoint analysis, which allows us to determine if a program has reached a steady state and, if so, whether that represents peak performance or not. Using this, we show that even when run in the most controlled of circumstances, small, deterministic, widely studied microbenchmarks often fail to reach a steady state of peak performance on a variety of common VMs. Repeating our experiment on 3 different machines, we found that at most 43.5% of pairs consistently reach a steady state of peak performance.", "score": 0.0, "mention_count": 10, "user_count": 7, "months": ["2016-03", "2016-10", "2016-12", "2017-09", "2021-10", "2022-08", "2022-11", "2024-01", "2024-12", "2025-12"], "last_mentioned": "2025-12-29T23:19:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1606.04474", "slug": "3baea075d7fdab3c", "title": "Learning to learn by gradient descent by gradient descent", "authors": ["Marcin Andrychowicz", "Misha Denil", "Sergio Gomez", "Matthew W. Hoffman", "David Pfau", "Tom Schaul", "Brendan Shillingford", "Nando de Freitas"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1606.04474", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "The move from hand-designed features to learned features in machine learning has been wildly successful.", "description_method": "abstract-excerpt", "abstract": "The move from hand-designed features to learned features in machine learning has been wildly successful. In spite of this, optimization algorithms are still designed by hand. In this paper we show how the design of an optimization algorithm can be cast as a learning problem, allowing the algorithm to learn to exploit structure in the problems of interest in an automatic way. Our learned algorithms, implemented by LSTMs, outperform generic, hand-designed competitors on the tasks for which they are trained, and also generalize well to new tasks with similar structure. We demonstrate this on a number of tasks, including simple convex problems, training neural networks, and styling images with neural art.", "score": 0.0, "mention_count": 10, "user_count": 10, "months": ["2016-08", "2016-11", "2017-01", "2017-02", "2017-04", "2017-07", "2019-03", "2022-08"], "last_mentioned": "2022-08-01T05:26:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1611.04135", "slug": "6615c2b3703ca6e2", "title": "Responses to Critiques on Machine Learning of Criminality Perceptions (Addendum of arXiv:1611.04135)", "authors": ["Xiaolin Wu", "Xi Zhang"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1611.04135", "doi": null, "topics": ["Psychology", "Machine learning", "Networks"], "description": "In November 2016 we submitted to arXiv our paper \"Automated Inference on Criminality Using Face Images\".", "description_method": "abstract-excerpt", "abstract": "In November 2016 we submitted to arXiv our paper \"Automated Inference on Criminality Using Face Images\". It generated a great deal of discussions in the Internet and some media outlets. Our work is only intended for pure academic discussions; how it has become a media consumption is a total surprise to us. Although in agreement with our critics on the need and importance of policing AI research for the general good of the society, we are deeply baffled by the ways some of them mispresented our work, in particular the motive and objective of our research.", "score": 0.0, "mention_count": 10, "user_count": 10, "months": ["2016-12", "2017-04", "2017-08", "2017-09", "2018-02", "2019-06", "2019-12", "2021-04", "2022-08"], "last_mentioned": "2022-08-21T19:38:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1704.04760", "slug": "d34b1ca72d39e8c6", "title": "In-Datacenter Performance Analysis of a Tensor Processing Unit", "authors": ["Norman P. Jouppi", "Cliff Young", "Nishant Patil", "David Patterson", "Gaurav Agrawal", "Raminder Bajwa", "Sarah Bates", "Suresh Bhatia", "Nan Boden", "Al Borchers", "Rick Boyle", "Pierre-luc Cantin", "Clifford Chao", "Chris Clark", "Jeremy Coriell", "Mike Daley", "Matt Dau", "Jeffrey Dean", "Ben Gelb", "Tara Vazir Ghaemmaghami", "Rajendra Gottipati", "William Gulland", "Robert Hagmann", "C. Richard Ho", "Doug Hogberg", "John Hu", "Robert Hundt", "Dan Hurt", "Julian Ibarz", "Aaron Jaffey", "Alek Jaworski", "Alexander Kaplan", "Harshit Khaitan", "Andy Koch", "Naveen Kumar", "Steve Lacy", "James Laudon", "James Law", "Diemthu Le", "Chris Leary", "Zhuyuan Liu", "Kyle Lucke", "Alan Lundin", "Gordon MacKean", "Adriana Maggiore", "Maire Mahony", "Kieran Miller", "Rahul Nagarajan", "Ravi Narayanaswami", "Ray Ni", "Kathy Nix", "Thomas Norrie", "Mark Omernick", "Narayana Penukonda", "Andy Phelps", "Jonathan Ross", "Matt Ross", "Amir Salek", "Emad Samadiani", "Chris Severn", "Gregory Sizikov", "Matthew Snelham", "Jed Souter", "Dan Steinberg", "Andy Swing", "Mercedes Tan", "Gregory Thorson", "Bo Tian", "Horia Toma", "Erick Tuttle", "Vijay Vasudevan", "Richard Walter", "Walter Wang", "Eric Wilcox", "Doe Hyun Yoon"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1704.04760", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "Many architects believe that major improvements in cost-energy-performance must now come from domain-specific hardware.", "description_method": "abstract-excerpt", "abstract": "Many architects believe that major improvements in cost-energy-performance must now come from domain-specific hardware. This paper evaluates a custom ASIC---called a Tensor Processing Unit (TPU)---deployed in datacenters since 2015 that accelerates the inference phase of neural networks (NN). The heart of the TPU is a 65,536 8-bit MAC matrix multiply unit that offers a peak throughput of 92 TeraOps/second (TOPS) and a large (28 MiB) software-managed on-chip memory. The TPU's deterministic execution model is a better match to the 99th-percentile response-time requirement of our NN applications than are the time-varying optimizations of CPUs and GPUs (caches, out-of-order execution, multithreading, multiprocessing, prefetching, ...) that help average throughput more than guaranteed latency. The lack of such features helps explain why, despite having myriad MACs and a big memory, the TPU is relatively small and low power. We compare the TPU to a server-class Intel Haswell CPU and an Nvidia K80 GPU, which are contemporaries deployed in the same datacenters. Our workload, written in the high-level TensorFlow framework, uses production NN applications (MLPs, CNNs, and LSTMs) that represent 95% of our datacenters' NN inference demand. Despite low utilization for some applications, the TPU is on average about 15X - 30X faster than its contemporary GPU or CPU, with TOPS/Watt about 30X - 80X higher. Moreover, using the GPU's GDDR5 memory in the TPU would triple achieved TOPS and raise TOPS/Watt to nearly 70X the GPU and 200X the CPU.", "score": 0.0, "mention_count": 10, "user_count": 9, "months": ["2017-06", "2017-07", "2017-08", "2018-02", "2018-04", "2019-04", "2023-02", "2024-02", "2024-10", "2025-04"], "last_mentioned": "2025-04-12T15:25:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1705.05937", "slug": "a6dd3e0fa4171264", "title": "Engineering Record And Replay For Deployability: Extended Technical Report", "authors": ["Robert O'Callahan", "Chris Jones", "Nathan Froyd", "Kyle Huey", "Albert Noll", "Nimrod Partush"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1705.05937", "doi": null, "topics": ["Computer systems", "Programming languages", "Algorithms"], "description": "The ability to record and replay program executions with low overhead enables many applications, such as reverse-execution debugging, debugging of hard-to-reproduce test failures, and \"black box\" forensic analysis of failures in deployed systems.", "description_method": "abstract-excerpt", "abstract": "The ability to record and replay program executions with low overhead enables many applications, such as reverse-execution debugging, debugging of hard-to-reproduce test failures, and \"black box\" forensic analysis of failures in deployed systems. Existing record-and-replay approaches limit deployability by recording an entire virtual machine (heavyweight), modifying the OS kernel (adding deployment and maintenance costs), requiring pervasive code instrumentation (imposing significant performance and complexity overhead), or modifying compilers and runtime systems (limiting generality). We investigated whether it is possible to build a practical record-and-replay system avoiding all these issues. The answer turns out to be yes - if the CPU and operating system meet certain non-obvious constraints. Fortunately modern Intel CPUs, Linux kernels and user-space frameworks do meet these constraints, although this has only become true recently. With some novel optimizations, our system 'rr' records and replays real-world low-parallelism workloads with low overhead, with an entirely user-space implementation, using stock hardware, compilers, runtimes and operating systems. \"rr\" forms the basis of an open-source reverse-execution debugger seeing significant use in practice. We present the design and implementation of 'rr', describe its performance on a variety of workloads, and identify constraints on hardware and operating system design required to support our approach.", "score": 0.0, "mention_count": 10, "user_count": 7, "months": ["2018-11", "2019-11", "2020-05", "2022-06", "2024-07", "2025-03"], "last_mentioned": "2025-03-06T17:12:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1806.07366", "slug": "327916ebd5c1b401", "title": "Neural Ordinary Differential Equations", "authors": ["Ricky T. Q. Chen", "Yulia Rubanova", "Jesse Bettencourt", "David Duvenaud"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1806.07366", "doi": null, "topics": ["Machine learning"], "description": "We introduce a new family of deep neural network models.", "description_method": "abstract-excerpt", "abstract": "We introduce a new family of deep neural network models. Instead of specifying a discrete sequence of hidden layers, we parameterize the derivative of the hidden state using a neural network. The output of the network is computed using a black-box differential equation solver. These continuous-depth models have constant memory cost, adapt their evaluation strategy to each input, and can explicitly trade numerical precision for speed. We demonstrate these properties in continuous-depth residual networks and continuous-time latent variable models. We also construct continuous normalizing flows, a generative model that can train by maximum likelihood, without partitioning or ordering the data dimensions. For training, we show how to scalably backpropagate through any ODE solver, without access to its internal operations. This allows end-to-end training of ODEs within larger models.", "score": 0.0, "mention_count": 10, "user_count": 10, "months": ["2018-06", "2018-12", "2019-01", "2019-03", "2019-09", "2021-03", "2024-03"], "last_mentioned": "2024-03-07T12:49:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1810.03993", "slug": "54fe162d31133deb", "title": "Model Cards for Model Reporting", "authors": ["Margaret Mitchell", "Simone Wu", "Andrew Zaldivar", "Parker Barnes", "Lucy Vasserman", "Ben Hutchinson", "Elena Spitzer", "Inioluwa Deborah Raji", "Timnit Gebru"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1810.03993", "doi": "10.1145/3287560.3287596", "topics": ["Machine learning", "Computer vision"], "description": "Trained machine learning models are increasingly used to perform high-impact tasks in areas such as law enforcement, medicine, education, and employment.", "description_method": "abstract-excerpt", "abstract": "Trained machine learning models are increasingly used to perform high-impact tasks in areas such as law enforcement, medicine, education, and employment. In order to clarify the intended use cases of machine learning models and minimize their usage in contexts for which they are not well suited, we recommend that released models be accompanied by documentation detailing their performance characteristics. In this paper, we propose a framework that we call model cards, to encourage such transparent model reporting. Model cards are short documents accompanying trained machine learning models that provide benchmarked evaluation in a variety of conditions, such as across different cultural, demographic, or phenotypic groups (e.g., race, geographic location, sex, Fitzpatrick skin type) and intersectional groups (e.g., age and race, or sex and Fitzpatrick skin type) that are relevant to the intended application domains. Model cards also disclose the context in which models are intended to be used, details of the performance evaluation procedures, and other relevant information. While we focus primarily on human-centered machine learning models in the application fields of computer vision and natural language processing, this framework can be used to document any trained machine learning model. To solidify the concept, we provide cards for two supervised models: One trained to detect smiling faces in images, and one trained to detect toxic comments in text. We propose model cards as a step towards the responsible democratization of machine learning and related AI technology, increasing transparency into how well AI technology works. We hope this work encourages those releasing trained machine learning models to accompany model releases with similar detailed evaluation numbers and other relevant documentation.", "score": 0.0, "mention_count": 10, "user_count": 9, "months": ["2020-07", "2021-06", "2023-08", "2023-12", "2024-08", "2024-12", "2025-10", "2026-05", "2026-06"], "last_mentioned": "2026-06-09T18:17:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1901.10220", "slug": "d0467299cd64373d", "title": "On the Impact of Programming Languages on Code Quality", "authors": ["Emery D. Berger", "Celeste Hollenbeck", "Petr Maj", "Olga Vitek", "Jan Vitek"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1901.10220", "doi": "10.1145/3340571", "topics": ["Programming languages", "Statistics & probability", "Psychology"], "description": "This paper is a reproduction of work by Ray et al.", "description_method": "abstract-excerpt", "abstract": "This paper is a reproduction of work by Ray et al. which claimed to have uncovered a statistically significant association between eleven programming languages and software defects in projects hosted on GitHub. First we conduct an experimental repetition, repetition is only partially successful, but it does validate one of the key claims of the original work about the association of ten programming languages with defects. Next, we conduct a complete, independent reanalysis of the data and statistical modeling steps of the original study. We uncover a number of flaws that undermine the conclusions of the original study as only four languages are found to have a statistically significant association with defects, and even for those the effect size is exceedingly small. We conclude with some additional sources of bias that should be investigated in follow up work and a few best practice recommendations for similar efforts.", "score": 0.0, "mention_count": 10, "user_count": 9, "months": ["2019-04", "2019-06", "2019-07", "2021-05", "2021-06", "2022-05", "2022-06", "2023-05", "2025-12"], "last_mentioned": "2025-12-08T11:02:44Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1906.05433", "slug": "de4ce75eeab2edf0", "title": "Tackling Climate Change with Machine Learning", "authors": ["David Rolnick", "Priya L. Donti", "Lynn H. Kaack", "Kelly Kochanski", "Alexandre Lacoste", "Kris Sankaran", "Andrew Slavin Ross", "Nikola Milojevic-Dupont", "Natasha Jaques", "Anna Waldman-Brown", "Alexandra Luccioni", "Tegan Maharaj", "Evan D. Sherwin", "S. Karthik Mukkavilli", "Konrad P. Kording", "Carla Gomes", "Andrew Y. Ng", "Demis Hassabis", "John C. Platt", "Felix Creutzig", "Jennifer Chayes", "Yoshua Bengio"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1906.05433", "doi": null, "topics": ["Machine learning", "Economics"], "description": "Climate change is one of the greatest challenges facing humanity, and we, as machine learning experts, may wonder how we can help.", "description_method": "abstract-excerpt", "abstract": "Climate change is one of the greatest challenges facing humanity, and we, as machine learning experts, may wonder how we can help. Here we describe how machine learning can be a powerful tool in reducing greenhouse gas emissions and helping society adapt to a changing climate. From smart grids to disaster management, we identify high impact problems where existing gaps can be filled by machine learning, in collaboration with other fields. Our recommendations encompass exciting research questions as well as promising business opportunities. We call on the machine learning community to join the global effort against climate change.", "score": 0.0, "mention_count": 10, "user_count": 6, "months": ["2019-06", "2019-10", "2019-12", "2022-08", "2022-10"], "last_mentioned": "2022-10-05T12:45:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2102.06824", "slug": "7ea48cef50731862", "title": "Introducing Physical Warp Drives", "authors": ["Alexey Bobrick", "Gianni Martire"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2102.06824", "doi": "10.1088/1361-6382/abdf6e", "topics": ["Physics", "Algorithms"], "description": "The Alcubierre warp drive is an exotic solution in general relativity.", "description_method": "abstract-excerpt", "abstract": "The Alcubierre warp drive is an exotic solution in general relativity. It allows for superluminal travel at the cost of enormous amounts of matter with negative mass density. For this reason, the Alcubierre warp drive has been widely considered unphysical. In this study, we develop a model of a general warp drive spacetime in classical relativity that encloses all existing warp drive definitions and allows for new metrics without the most serious issues present in the Alcubierre solution. We present the first general model for subluminal positive-energy, spherically symmetric warp drives; construct superluminal warp-drive solutions which satisfy quantum inequalities; provide optimizations for the Alcubierre metric that decrease the negative energy requirements by two orders of magnitude; and introduce a warp drive spacetime in which space capacity and the rate of time can be chosen in a controlled manner. Conceptually, we demonstrate that any warp drive, including the Alcubierre drive, is a shell of regular or exotic material moving inertially with a certain velocity. Therefore, any warp drive requires propulsion. We show that a class of subluminal, spherically symmetric warp drive spacetimes, at least in principle, can be constructed based on the physical principles known to humanity today.", "score": 0.0, "mention_count": 10, "user_count": 9, "months": ["2021-03", "2021-05", "2021-12", "2022-11", "2024-07"], "last_mentioned": "2024-07-04T04:16:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2103.00020", "slug": "2f4691de4e006980", "title": "Learning Transferable Visual Models From Natural Language Supervision", "authors": ["Alec Radford", "Jong Wook Kim", "Chris Hallacy", "Aditya Ramesh", "Gabriel Goh", "Sandhini Agarwal", "Girish Sastry", "Amanda Askell", "Pamela Mishkin", "Jack Clark", "Gretchen Krueger", "Ilya Sutskever"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2103.00020", "doi": null, "topics": ["Computer vision", "Psychology", "Networks"], "description": "State-of-the-art computer vision systems are trained to predict a fixed set of predetermined object categories.", "description_method": "abstract-excerpt", "abstract": "State-of-the-art computer vision systems are trained to predict a fixed set of predetermined object categories. This restricted form of supervision limits their generality and usability since additional labeled data is needed to specify any other visual concept. Learning directly from raw text about images is a promising alternative which leverages a much broader source of supervision. We demonstrate that the simple pre-training task of predicting which caption goes with which image is an efficient and scalable way to learn SOTA image representations from scratch on a dataset of 400 million (image, text) pairs collected from the internet. After pre-training, natural language is used to reference learned visual concepts (or describe new ones) enabling zero-shot transfer of the model to downstream tasks. We study the performance of this approach by benchmarking on over 30 different existing computer vision datasets, spanning tasks such as OCR, action recognition in videos, geo-localization, and many types of fine-grained object classification. The model transfers non-trivially to most tasks and is often competitive with a fully supervised baseline without the need for any dataset specific training. For instance, we match the accuracy of the original ResNet-50 on ImageNet zero-shot without needing to use any of the 1.28 million training examples it was trained on. We release our code and pre-trained model weights at https://github.com/OpenAI/CLIP.", "score": 0.0, "mention_count": 10, "user_count": 7, "months": ["2021-03", "2021-08", "2022-04", "2022-10", "2024-03", "2024-08", "2025-01"], "last_mentioned": "2025-01-15T17:54:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2201.08239", "slug": "e5238d03940898cc", "title": "LaMDA: Language Models for Dialog Applications", "authors": ["Romal Thoppilan", "Daniel De Freitas", "Jamie Hall", "Noam Shazeer", "Apoorv Kulshreshtha", "Heng-Tze Cheng", "Alicia Jin", "Taylor Bos", "Leslie Baker", "Yu Du", "YaGuang Li", "Hongrae Lee", "Huaixiu Steven Zheng", "Amin Ghafouri", "Marcelo Menegali", "Yanping Huang", "Maxim Krikun", "Dmitry Lepikhin", "James Qin", "Dehao Chen", "Yuanzhong Xu", "Zhifeng Chen", "Adam Roberts", "Maarten Bosma", "Vincent Zhao", "Yanqi Zhou", "Chung-Ching Chang", "Igor Krivokon", "Will Rusch", "Marc Pickett", "Pranesh Srinivasan", "Laichee Man", "Kathleen Meier-Hellstern", "Meredith Ringel Morris", "Tulsee Doshi", "Renelito Delos Santos", "Toju Duke", "Johnny Soraker", "Ben Zevenbergen", "Vinodkumar Prabhakaran", "Mark Diaz", "Ben Hutchinson", "Kristen Olson", "Alejandra Molina", "Erin Hoffman-John", "Josh Lee", "Lora Aroyo", "Ravi Rajakumar", "Alena Butryna", "Matthew Lamm", "Viktoriya Kuzmina", "Joe Fenton", "Aaron Cohen", "Rachel Bernstein", "Ray Kurzweil", "Blaise Aguera-Arcas", "Claire Cui", "Marian Croak", "Ed Chi", "Quoc Le"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2201.08239", "doi": null, "topics": ["Language models", "Psychology", "Machine learning"], "description": "We present LaMDA: Language Models for Dialog Applications.", "description_method": "abstract-excerpt", "abstract": "We present LaMDA: Language Models for Dialog Applications. LaMDA is a family of Transformer-based neural language models specialized for dialog, which have up to 137B parameters and are pre-trained on 1.56T words of public dialog data and web text. While model scaling alone can improve quality, it shows less improvements on safety and factual grounding. We demonstrate that fine-tuning with annotated data and enabling the model to consult external knowledge sources can lead to significant improvements towards the two key challenges of safety and factual grounding. The first challenge, safety, involves ensuring that the model's responses are consistent with a set of human values, such as preventing harmful suggestions and unfair bias. We quantify safety using a metric based on an illustrative set of human values, and we find that filtering candidate responses using a LaMDA classifier fine-tuned with a small amount of crowdworker-annotated data offers a promising approach to improving model safety. The second challenge, factual grounding, involves enabling the model to consult external knowledge sources, such as an information retrieval system, a language translator, and a calculator. We quantify factuality using a groundedness metric, and we find that our approach enables the model to generate responses grounded in known sources, rather than responses that merely sound plausible. Finally, we explore the use of LaMDA in the domains of education and content recommendations, and analyze their helpfulness and role consistency.", "score": 0.0, "mention_count": 10, "user_count": 8, "months": ["2022-02", "2022-05", "2022-06", "2022-09", "2022-12", "2023-02", "2023-03"], "last_mentioned": "2023-03-09T01:39:19Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.05217", "slug": "f71e3b65a8b3f35b", "title": "Progress measures for grokking via mechanistic interpretability", "authors": ["Neel Nanda", "Lawrence Chan", "Tom Lieberum", "Jess Smith", "Jacob Steinhardt"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.05217", "doi": null, "topics": ["Psychology", "Algorithms", "Machine learning"], "description": "Neural networks often exhibit emergent behavior, where qualitatively new capabilities arise from scaling up the amount of parameters, training data, or training steps.", "description_method": "abstract-excerpt", "abstract": "Neural networks often exhibit emergent behavior, where qualitatively new capabilities arise from scaling up the amount of parameters, training data, or training steps. One approach to understanding emergence is to find continuous \\textit{progress measures} that underlie the seemingly discontinuous qualitative changes. We argue that progress measures can be found via mechanistic interpretability: reverse-engineering learned behaviors into their individual components. As a case study, we investigate the recently-discovered phenomenon of ``grokking'' exhibited by small transformers trained on modular addition tasks. We fully reverse engineer the algorithm learned by these networks, which uses discrete Fourier transforms and trigonometric identities to convert addition to rotation about a circle. We confirm the algorithm by analyzing the activations and weights and by performing ablations in Fourier space. Based on this understanding, we define progress measures that allow us to study the dynamics of training and split training into three continuous phases: memorization, circuit formation, and cleanup. Our results show that grokking, rather than being a sudden shift, arises from the gradual amplification of structured mechanisms encoded in the weights, followed by the later removal of memorizing components.", "score": 0.0, "mention_count": 10, "user_count": 7, "months": ["2023-01", "2023-04", "2023-05", "2023-08", "2024-02", "2024-10", "2026-01"], "last_mentioned": "2026-01-15T22:55:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2302.02083", "slug": "f716bd82d12aed22", "title": "Evaluating Large Language Models in Theory of Mind Tasks", "authors": ["Michal Kosinski"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2302.02083", "doi": "10.1073/pnas.2405460121", "topics": ["Language models"], "description": "Eleven Large Language Models (LLMs) were assessed using a custom-made battery of false-belief tasks, considered a gold standard in testing Theory of Mind (ToM) in humans.", "description_method": "abstract-excerpt", "abstract": "Eleven Large Language Models (LLMs) were assessed using a custom-made battery of false-belief tasks, considered a gold standard in testing Theory of Mind (ToM) in humans. The battery included 640 prompts spread across 40 diverse tasks, each one including a false-belief scenario, three closely matched true-belief control scenarios, and the reversed versions of all four. To solve a single task, a model needed to correctly answer 16 prompts across all eight scenarios. Smaller and older models solved no tasks; GPT-3-davinci-003 (from November 2022) and ChatGPT-3.5-turbo (from March 2023) solved 20% of the tasks; ChatGPT-4 (from June 2023) solved 75% of the tasks, matching the performance of six-year-old children observed in past studies. We explore the potential interpretation of these findings, including the intriguing possibility that ToM, previously considered exclusive to humans, may have spontaneously emerged as a byproduct of LLMs' improving language skills.", "score": 0.0, "mention_count": 10, "user_count": 9, "months": ["2023-02", "2023-03", "2023-04", "2023-05"], "last_mentioned": "2023-05-14T06:43:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2304.01433", "slug": "eb01c6279da01afe", "title": "TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings", "authors": ["Norman P. Jouppi", "George Kurian", "Sheng Li", "Peter Ma", "Rahul Nagarajan", "Lifeng Nai", "Nishant Patil", "Suvinay Subramanian", "Andy Swing", "Brian Towles", "Cliff Young", "Xiang Zhou", "Zongwei Zhou", "David Patterson"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2304.01433", "doi": null, "topics": ["Machine learning", "Mathematics", "Security & cryptography"], "description": "In response to innovations in machine learning (ML) models, production workloads changed radically and rapidly.", "description_method": "abstract-excerpt", "abstract": "In response to innovations in machine learning (ML) models, production workloads changed radically and rapidly. TPU v4 is the fifth Google domain specific architecture (DSA) and its third supercomputer for such ML models. Optical circuit switches (OCSes) dynamically reconfigure its interconnect topology to improve scale, availability, utilization, modularity, deployment, security, power, and performance; users can pick a twisted 3D torus topology if desired. Much cheaper, lower power, and faster than Infiniband, OCSes and underlying optical components are <5% of system cost and <3% of system power. Each TPU v4 includes SparseCores, dataflow processors that accelerate models that rely on embeddings by 5x-7x yet use only 5% of die area and power. Deployed since 2020, TPU v4 outperforms TPU v3 by 2.1x and improves performance/Watt by 2.7x. The TPU v4 supercomputer is 4x larger at 4096 chips and thus ~10x faster overall, which along with OCS flexibility helps large language models. For similar sized systems, it is ~4.3x-4.5x faster than the Graphcore IPU Bow and is 1.2x-1.7x faster and uses 1.3x-1.9x less power than the Nvidia A100. TPU v4s inside the energy-optimized warehouse scale computers of Google Cloud use ~3x less energy and produce ~20x less CO2e than contemporary DSAs in a typical on-premise data center.", "score": 0.0, "mention_count": 10, "user_count": 10, "months": ["2023-04", "2023-05", "2024-06", "2025-06", "2025-10"], "last_mentioned": "2025-10-01T23:13:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.14314", "slug": "5324ffa2c26905b3", "title": "QLoRA: Efficient Finetuning of Quantized LLMs", "authors": ["Tim Dettmers", "Artidoro Pagnoni", "Ari Holtzman", "Luke Zettlemoyer"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.14314", "doi": null, "topics": ["Language models", "Machine learning", "Distributed systems"], "description": "We present QLoRA, an efficient finetuning approach that reduces memory usage enough to finetune a 65B parameter model on a single 48GB GPU while preserving full 16-bit finetuning task performance.", "description_method": "abstract-excerpt", "abstract": "We present QLoRA, an efficient finetuning approach that reduces memory usage enough to finetune a 65B parameter model on a single 48GB GPU while preserving full 16-bit finetuning task performance. QLoRA backpropagates gradients through a frozen, 4-bit quantized pretrained language model into Low Rank Adapters~(LoRA). Our best model family, which we name Guanaco, outperforms all previous openly released models on the Vicuna benchmark, reaching 99.3% of the performance level of ChatGPT while only requiring 24 hours of finetuning on a single GPU. QLoRA introduces a number of innovations to save memory without sacrificing performance: (a) 4-bit NormalFloat (NF4), a new data type that is information theoretically optimal for normally distributed weights (b) double quantization to reduce the average memory footprint by quantizing the quantization constants, and (c) paged optimziers to manage memory spikes. We use QLoRA to finetune more than 1,000 models, providing a detailed analysis of instruction following and chatbot performance across 8 instruction datasets, multiple model types (LLaMA, T5), and model scales that would be infeasible to run with regular finetuning (e.g. 33B and 65B parameter models). Our results show that QLoRA finetuning on a small high-quality dataset leads to state-of-the-art results, even when using smaller models than the previous SoTA. We provide a detailed analysis of chatbot performance based on both human and GPT-4 evaluations showing that GPT-4 evaluations are a cheap and reasonable alternative to human evaluation. Furthermore, we find that current chatbot benchmarks are not trustworthy to accurately evaluate the performance levels of chatbots. A lemon-picked analysis demonstrates where Guanaco fails compared to ChatGPT. We release all of our models and code, including CUDA kernels for 4-bit training.", "score": 0.0, "mention_count": 10, "user_count": 7, "months": ["2023-05", "2023-06", "2023-12", "2024-05", "2024-10", "2026-03"], "last_mentioned": "2026-03-04T04:19:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.18654", "slug": "e202f38244f8a892", "title": "Faith and Fate: Limits of Transformers on Compositionality", "authors": ["Nouha Dziri", "Ximing Lu", "Melanie Sclar", "Xiang Lorraine Li", "Liwei Jiang", "Bill Yuchen Lin", "Peter West", "Chandra Bhagavatula", "Ronan Le Bras", "Jena D. Hwang", "Soumya Sanyal", "Sean Welleck", "Xiang Ren", "Allyson Ettinger", "Zaid Harchaoui", "Yejin Choi"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.18654", "doi": null, "topics": ["Language models", "Algorithms", "Programming languages"], "description": "Transformer large language models (LLMs) have sparked admiration for their exceptional performance on tasks that demand intricate multi-step reasoning.", "description_method": "abstract-excerpt", "abstract": "Transformer large language models (LLMs) have sparked admiration for their exceptional performance on tasks that demand intricate multi-step reasoning. Yet, these models simultaneously show failures on surprisingly trivial problems. This begs the question: Are these errors incidental, or do they signal more substantial limitations? In an attempt to demystify transformer LLMs, we investigate the limits of these models across three representative compositional tasks -- multi-digit multiplication, logic grid puzzles, and a classic dynamic programming problem. These tasks require breaking problems down into sub-steps and synthesizing these steps into a precise answer. We formulate compositional tasks as computation graphs to systematically quantify the level of complexity, and break down reasoning steps into intermediate sub-procedures. Our empirical findings suggest that transformer LLMs solve compositional tasks by reducing multi-step compositional reasoning into linearized subgraph matching, without necessarily developing systematic problem-solving skills. To round off our empirical study, we provide theoretical arguments on abstract multi-step reasoning problems that highlight how autoregressive generations' performance can rapidly decay with\\,increased\\,task\\,complexity.", "score": 0.0, "mention_count": 10, "user_count": 7, "months": ["2023-06", "2023-07", "2023-12", "2024-03", "2024-08", "2025-01", "2025-06", "2025-10"], "last_mentioned": "2025-10-20T13:07:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2410.02724", "slug": "76f21dee5a11a7b8", "title": "Large Language Models as Markov Chains", "authors": ["Oussama Zekri", "Ambroise Odonnat", "Abdelhakim Benechehab", "Linus Bleistein", "Nicolas Boullé", "Ievgen Redko"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2410.02724", "doi": null, "topics": ["Language models", "Psychology", "Machine learning"], "description": "Large language models (LLMs) are remarkably efficient across a wide range of natural language processing tasks and well beyond them.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLMs) are remarkably efficient across a wide range of natural language processing tasks and well beyond them. However, a comprehensive theoretical analysis of the LLMs' generalization capabilities remains elusive. In our paper, we approach this task by drawing an equivalence between autoregressive transformer-based language models and Markov chains defined on a finite state space. This allows us to study the multi-step inference mechanism of LLMs from first principles. We relate the obtained results to the pathological behavior observed with LLMs such as repetitions and incoherent replies with high temperature. Finally, we leverage the proposed formalization to derive pre-training and in-context learning generalization bounds for LLMs under realistic data and model assumptions. Experiments with the most recent Llama and Gemma herds of models show that our theory correctly captures their behavior in practice.", "score": 0.0, "mention_count": 10, "user_count": 7, "months": ["2025-07", "2025-09", "2025-10", "2025-11", "2025-12", "2026-05", "2026-06"], "last_mentioned": "2026-06-08T03:50:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2501.19393", "slug": "2a88818f7605b71c", "title": "s1: Simple test-time scaling", "authors": ["Niklas Muennighoff", "Zitong Yang", "Weijia Shi", "Xiang Lisa Li", "Li Fei-Fei", "Hannaneh Hajishirzi", "Luke Zettlemoyer", "Percy Liang", "Emmanuel Candès", "Tatsunori Hashimoto"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2501.19393", "doi": null, "topics": ["Language models", "Distributed systems"], "description": "Test-time scaling is a promising new approach to language modeling that uses extra test-time compute to improve performance.", "description_method": "abstract-excerpt", "abstract": "Test-time scaling is a promising new approach to language modeling that uses extra test-time compute to improve performance. Recently, OpenAI's o1 model showed this capability but did not publicly share its methodology, leading to many replication efforts. We seek the simplest approach to achieve test-time scaling and strong reasoning performance. First, we curate a small dataset s1K of 1,000 questions paired with reasoning traces relying on three criteria we validate through ablations: difficulty, diversity, and quality. Second, we develop budget forcing to control test-time compute by forcefully terminating the model's thinking process or lengthening it by appending \"Wait\" multiple times to the model's generation when it tries to end. This can lead the model to double-check its answer, often fixing incorrect reasoning steps. After supervised finetuning the Qwen2.5-32B-Instruct language model on s1K and equipping it with budget forcing, our model s1-32B exceeds o1-preview on competition math questions by up to 27% (MATH and AIME24). Further, scaling s1-32B with budget forcing allows extrapolating beyond its performance without test-time intervention: from 50% to 57% on AIME24. Our model, data, and code are open-source at https://github.com/simplescaling/s1", "score": 0.0, "mention_count": 10, "user_count": 10, "months": ["2025-02", "2025-03", "2025-04", "2026-04", "2026-07"], "last_mentioned": "2026-07-20T21:21:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2503.18813", "slug": "13649d62aa5d7b1b", "title": "Defeating Prompt Injections by Design", "authors": ["Edoardo Debenedetti", "Ilia Shumailov", "Tianqi Fan", "Jamie Hayes", "Nicholas Carlini", "Daniel Fabian", "Christoph Kern", "Chongyang Shi", "Andreas Terzis", "Florian Tramèr"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2503.18813", "doi": null, "topics": ["Security & cryptography", "Language models", "Databases"], "description": "Large Language Models (LLMs) are increasingly deployed in agentic systems that interact with an untrusted environment.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) are increasingly deployed in agentic systems that interact with an untrusted environment. However, LLM agents are vulnerable to prompt injection attacks when handling untrusted data. In this paper we propose CaMeL, a robust defense that creates a protective system layer around the LLM, securing it even when underlying models are susceptible to attacks. To operate, CaMeL explicitly extracts the control and data flows from the (trusted) query; therefore, the untrusted data retrieved by the LLM can never impact the program flow. To further improve security, CaMeL uses a notion of a capability to prevent the exfiltration of private data over unauthorized data flows by enforcing security policies when tools are called. We demonstrate effectiveness of CaMeL by solving $77\\%$ of tasks with provable security (compared to $84\\%$ with an undefended system) in AgentDojo. We release CaMeL at https://github.com/google-research/camel-prompt-injection.", "score": 0.0, "mention_count": 10, "user_count": 7, "months": ["2025-04", "2025-07", "2025-08", "2025-09", "2026-01", "2026-02", "2026-03"], "last_mentioned": "2026-03-06T13:08:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:quant-ph/0101012", "slug": "4e04853e4e10bf46", "title": "Quantum Theory From Five Reasonable Axioms", "authors": ["Lucien Hardy"], "year": 2001, "venue": "arXiv", "url": "https://arxiv.org/abs/quant-ph/0101012", "doi": null, "topics": ["Physics", "Statistics & probability", "Language models"], "description": "The usual formulation of quantum theory is based on rather obscure axioms (employing complex Hilbert spaces, Hermitean operators, and the trace rule for calculating probabilities).", "description_method": "abstract-excerpt", "abstract": "The usual formulation of quantum theory is based on rather obscure axioms (employing complex Hilbert spaces, Hermitean operators, and the trace rule for calculating probabilities). In this paper it is shown that quantum theory can be derived from five very reasonable axioms. The first four of these are obviously consistent with both quantum theory and classical probability theory. Axiom 5 (which requires that there exists continuous reversible transformations between pure states) rules out classical probability theory. If Axiom 5 (or even just the word \"continuous\" from Axiom 5) is dropped then we obtain classical probability theory instead. This work provides some insight into the reasons quantum theory is the way it is. For example, it explains the need for complex numbers and where the trace formula comes from. We also gain insight into the relationship between quantum theory and classical probability theory.", "score": 0.0, "mention_count": 10, "user_count": 7, "months": ["2013-03", "2014-09", "2017-03", "2017-08", "2021-04", "2021-09", "2021-10", "2021-12", "2023-12", "2025-03"], "last_mentioned": "2025-03-03T11:27:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature.2015.17700", "slug": "9e5e911d700dabc2", "title": "Climate-change ‘hiatus’ disappears with new data", "authors": ["Jeff Tollefson"], "year": 2015, "venue": "Nature", "url": "https://doi.org/10.1038/nature.2015.17700", "doi": "10.1038/nature.2015.17700", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 10, "user_count": 4, "months": ["2022-03", "2022-04", "2022-05", "2022-07", "2022-10", "2022-11", "2023-08", "2023-12"], "last_mentioned": "2023-12-17T18:12:11Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41421-020-0156-0", "slug": "6c358828382f5506", "title": "Hydroxychloroquine, a less toxic derivative of chloroquine, is effective in inhibiting SARS-CoV-2 infection in vitro", "authors": ["Jia Liu", "Ruiyuan Cao", "Mingyue Xu", "Xi Wang", "Huanyu Zhang", "Hengrui Hu", "Yufeng Li", "Zhihong Hu", "Wu Zhong", "Manli Wang"], "year": 2020, "venue": "Cell Discovery", "url": "https://doi.org/10.1038/s41421-020-0156-0", "doi": "10.1038/s41421-020-0156-0", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 10, "user_count": 9, "months": ["2020-03", "2020-04", "2020-09"], "last_mentioned": "2020-09-15T23:51:04Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2918-0", "slug": "4dd050aab831d0bf", "title": "Age-specific mortality and immunity patterns of SARS-CoV-2", "authors": ["Megan O’Driscoll", "Gabriel Ribeiro Dos Santos", "Lin Wang", "Derek A. T. Cummings", "Andrew S. Azman", "Juliette Paireau", "Arnaud Fontanet", "Simon Cauchemez", "Henrik Salje"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2918-0", "doi": "10.1038/s41586-020-2918-0", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 10, "user_count": 8, "months": ["2020-12", "2021-03", "2021-04", "2021-06", "2021-07", "2021-08", "2021-09", "2022-10"], "last_mentioned": "2022-10-24T13:29:05Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-024-08025-4", "slug": "d625b0dd7b5e9293", "title": "Scalable watermarking for identifying large language model outputs", "authors": ["Sumanth Dathathri", "Abigail See", "Sumedh Ghaisas", "Po-Sen Huang", "Rob McAdam", "Johannes Welbl", "Vandana Bachani", "Alex Kaskasoli", "Robert Stanforth", "Tatiana Matejovicova", "Jamie Hayes", "Nidhi Vyas", "Majd Al Merey", "Jonah Brown-Cohen", "Rudy Bunel", "Borja Balle", "Taylan Cemgil", "Zahra Ahmed", "Kitty Stacpoole", "Ilia Shumailov", "Ciprian Baetu", "Sven Gowal", "Demis Hassabis", "Pushmeet Kohli"], "year": 2024, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-024-08025-4", "doi": "10.1038/s41586-024-08025-4", "topics": ["Language models", "Algorithms"], "description": "Abstract Large language models (LLMs) have enabled the generation of high-quality synthetic text, often indistinguishable from human-written content, at a scale that can markedly affect the nature of the information ecosystem 1–3 .", "description_method": "abstract-excerpt", "abstract": "Abstract Large language models (LLMs) have enabled the generation of high-quality synthetic text, often indistinguishable from human-written content, at a scale that can markedly affect the nature of the information ecosystem 1–3 . Watermarking can help identify synthetic text and limit accidental or deliberate misuse 4 , but has not been adopted in production systems owing to stringent quality, detectability and computational efficiency requirements. Here we describe SynthID-Text, a production-ready text watermarking scheme that preserves text quality and enables high detection accuracy, with minimal latency overhead. SynthID-Text does not affect LLM training and modifies only the sampling procedure; watermark detection is computationally efficient, without using the underlying LLM. To enable watermarking at scale, we develop an algorithm integrating watermarking with speculative sampling, an efficiency technique frequently used in production systems 5 . Evaluations across multiple LLMs empirically show that SynthID-Text provides improved detectability over comparable methods, and standard benchmarks and human side-by-side ratings indicate no change in LLM capabilities. To demonstrate the feasibility of watermarking in large-scale-production systems, we conducted a live experiment that assessed feedback from nearly 20 million Gemini 6 responses, again confirming the preservation of text quality. We hope that the availability of SynthID-Text 7 will facilitate further development of watermarking and responsible use of LLM systems.", "score": 0.0, "mention_count": 10, "user_count": 8, "months": ["2024-11", "2024-12", "2025-05", "2025-08", "2026-08"], "last_mentioned": "2026-08-21T01:31:58Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-021-01336-3", "slug": "0f1fbc8b127b74d9", "title": "MDMA-assisted therapy for severe PTSD: a randomized, double-blind, placebo-controlled phase 3 study", "authors": ["Jennifer M. Mitchell", "Michael Bogenschutz", "Alia Lilienstein", "Charlotte Harrison", "Sarah Kleiman", "Kelly Parker-Guilbert", "Marcela Ot’alora G.", "Wael Garas", "Casey Paleos", "Ingmar Gorman", "Christopher Nicholas", "Michael Mithoefer", "Shannon Carlin", "Bruce Poulter", "Ann Mithoefer", "Sylvestre Quevedo", "Gregory Wells", "Sukhpreet S. Klaire", "Bessel van der Kolk", "Keren Tzarfaty", "Revital Amiaz", "Ray Worthy", "Scott Shannon", "Joshua D. Woolley", "Cole Marta", "Yevgeniy Gelfand", "Emma Hapke", "Simon Amar", "Yair Wallach", "Randall Brown", "Scott Hamilton", "Julie B. Wang", "Allison Coker", "Rebecca Matthews", "Alberdina de Boer", "Berra Yazar-Klosinski", "Amy Emerson", "Rick Doblin"], "year": 2021, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-021-01336-3", "doi": "10.1038/s41591-021-01336-3", "topics": ["Biology & medicine"], "description": "Abstract Post-traumatic stress disorder (PTSD) presents a major public health problem for which currently available treatments are modestly effective.", "description_method": "abstract-excerpt", "abstract": "Abstract Post-traumatic stress disorder (PTSD) presents a major public health problem for which currently available treatments are modestly effective. We report the findings of a randomized, double-blind, placebo-controlled, multi-site phase 3 clinical trial (NCT03537014) to test the efficacy and safety of 3,4-methylenedioxymethamphetamine (MDMA)-assisted therapy for the treatment of patients with severe PTSD, including those with common comorbidities such as dissociation, depression, a history of alcohol and substance use disorders, and childhood trauma. After psychiatric medication washout, participants ( n = 90) were randomized 1:1 to receive manualized therapy with MDMA or with placebo, combined with three preparatory and nine integrative therapy sessions. PTSD symptoms, measured with the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5, the primary endpoint), and functional impairment, measured with the Sheehan Disability Scale (SDS, the secondary endpoint) were assessed at baseline and at 2 months after the last experimental session. Adverse events and suicidality were tracked throughout the study. MDMA was found to induce significant and robust attenuation in CAPS-5 score compared with placebo ( P < 0.0001, d = 0.91) and to significantly decrease the SDS total score ( P = 0.0116, d = 0.43). The mean change in CAPS-5 scores in participants completing treatment was −24.4 (s.d. 11.6) in the MDMA group and −13.9 (s.d. 11.5) in the placebo group. MDMA did not induce adverse events of abuse potential, suicidality or QT prolongation. These data indicate that, compared with manualized therapy with inactive placebo, MDMA-assisted therapy is highly efficacious in individuals with severe PTSD, and treatment is safe and well-tolerated, even in those with comorbidities. We conclude that MDMA-assisted therapy represents a potential breakthrough treatment that merits expedited clinical evaluation.", "score": 0.0, "mention_count": 10, "user_count": 9, "months": ["2022-01", "2022-04", "2023-03", "2023-04", "2023-05", "2023-06", "2023-08"], "last_mentioned": "2023-08-31T13:25:01Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-019-49660-6", "slug": "550f51d39b1b20b3", "title": "Transgenic Aedes aegypti Mosquitoes Transfer Genes into a Natural Population", "authors": ["Benjamin R. Evans", "Panayiota Kotsakiozi", "Andre Luis Costa-da-Silva", "Rafaella Sayuri Ioshino", "Luiza Garziera", "Michele C. Pedrosa", "Aldo Malavasi", "Jair F. Virginio", "Margareth L. Capurro", "Jeffrey R. Powell"], "year": 2019, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-019-49660-6", "doi": "10.1038/s41598-019-49660-6", "topics": ["Biology & medicine"], "description": "AbstractIn an attempt to control the mosquito-borne diseases yellow fever, dengue, chikungunya, and Zika fevers, a strain of transgenically modified Aedes aegypti mosquitoes containing a dominant lethal gene has been developed by a commercial company, Oxitec Ltd.", "description_method": "abstract-excerpt", "abstract": "AbstractIn an attempt to control the mosquito-borne diseases yellow fever, dengue, chikungunya, and Zika fevers, a strain of transgenically modified Aedes aegypti mosquitoes containing a dominant lethal gene has been developed by a commercial company, Oxitec Ltd. If lethality is complete, releasing this strain should only reduce population size and not affect the genetics of the target populations. Approximately 450 thousand males of this strain were released each week for 27 months in Jacobina, Bahia, Brazil. We genotyped the release strain and the target Jacobina population before releases began for >21,000 single nucleotide polymorphisms (SNPs). Genetic sampling from the target population six, 12, and 27–30 months after releases commenced provides clear evidence that portions of the transgenic strain genome have been incorporated into the target population. Evidently, rare viable hybrid offspring between the release strain and the Jacobina population are sufficiently robust to be able to reproduce in nature. The release strain was developed using a strain originally from Cuba, then outcrossed to a Mexican population. Thus, Jacobina Ae. aegypti are now a mix of three populations. It is unclear how this may affect disease transmission or affect other efforts to control these dangerous vectors. These results highlight the importance of having in place a genetic monitoring program during such releases to detect un-anticipated outcomes.", "score": 0.0, "mention_count": 10, "user_count": 9, "months": ["2019-09", "2020-01", "2020-05", "2020-11", "2021-05", "2022-03"], "last_mentioned": "2022-03-11T05:11:09Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3385412.3386037", "slug": "e18e980dc2a7002e", "title": "Towards an API for the real numbers", "authors": ["Hans-J. Boehm"], "year": 2020, "venue": "Proceedings of the 41st ACM SIGPLAN Conference on Programming Language Design and Implementation", "url": "https://doi.org/10.1145/3385412.3386037", "doi": "10.1145/3385412.3386037", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 10, "user_count": 9, "months": ["2022-01", "2022-03", "2024-08", "2025-02", "2025-05", "2026-03"], "last_mentioned": "2026-03-22T21:25:05Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1201.3798", "slug": "5844a9b5af908c85", "title": "No need for conspiracy: Self-organized cartel formation in a modified trust game", "authors": ["Tiago P. Peixoto", "Stefan Bornholdt"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1201.3798", "doi": "10.1103/PhysRevLett.108.218702", "topics": ["Other research"], "description": "We investigate the dynamics of a trust game on a mixed population where individuals with the role of buyers are forced to play against a predetermined number of sellers, whom they choose dynamically.", "description_method": "abstract-excerpt", "abstract": "We investigate the dynamics of a trust game on a mixed population where individuals with the role of buyers are forced to play against a predetermined number of sellers, whom they choose dynamically. Agents with the role of sellers are also allowed to adapt the level of value for money of their products, based on payoff. The dynamics undergoes a transition at a specific value of the strategy update rate, above which an emergent cartel organization is observed, where sellers have similar values of below optimal value for money. This cartel organization is not due to an explicit collusion among agents; instead it arises spontaneously from the maximization of the individual payoffs. This dynamics is marked by large fluctuations and a high degree of unpredictability for most of the parameter space, and serves as a plausible qualitative explanation for observed elevated levels and fluctuations of certain commodity prices.", "score": 0.0, "mention_count": 9, "user_count": 4, "months": ["2012-01", "2014-03", "2015-02", "2015-07", "2016-01", "2016-09", "2017-08", "2021-03", "2024-08"], "last_mentioned": "2024-08-13T15:50:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1307.1714", "slug": "178724ed8d053c4f", "title": "Can Bohmian mechanics be made relativistic?", "authors": ["Detlef Dürr", "Sheldon Goldstein", "Travis Norsen", "Ward Struyve", "Nino Zanghì"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1307.1714", "doi": "10.1098/rspa.2013.0699", "topics": ["Physics"], "description": "In relativistic space-time, Bohmian theories can be formulated by introducing a privileged foliation of space-time.", "description_method": "abstract-excerpt", "abstract": "In relativistic space-time, Bohmian theories can be formulated by introducing a privileged foliation of space-time. The introduction of such a foliation - as extra absolute space-time structure - would seem to imply a clear violation of Lorentz invariance, and thus a conflict with fundamental relativity. Here, we consider the possibility that, instead of positing it as extra structure, the required foliation could be covariantly determined by the wave function. We argue that this allows for the formulation of Bohmian theories that seem to qualify as fundamentally Lorentz invariant. We conclude with some discussion of whether or not they might also qualify as fundamentally relativistic.", "score": 0.0, "mention_count": 9, "user_count": 4, "months": ["2016-10", "2016-12", "2017-01", "2018-07", "2018-08", "2019-07", "2021-01", "2023-01"], "last_mentioned": "2023-01-24T16:53:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1311.2540", "slug": "41b81ea2c5bc776b", "title": "Asymmetric numeral systems: entropy coding combining speed of Huffman coding with compression rate of arithmetic coding", "authors": ["Jarek Duda"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1311.2540", "doi": null, "topics": ["Statistics & probability", "Security & cryptography", "Psychology"], "description": "The modern data compression is mainly based on two approaches to entropy coding: Huffman (HC) and arithmetic/range coding (AC).", "description_method": "abstract-excerpt", "abstract": "The modern data compression is mainly based on two approaches to entropy coding: Huffman (HC) and arithmetic/range coding (AC). The former is much faster, but approximates probabilities with powers of 2, usually leading to relatively low compression rates. The latter uses nearly exact probabilities - easily approaching theoretical compression rate limit (Shannon entropy), but at cost of much larger computational cost. Asymmetric numeral systems (ANS) is a new approach to accurate entropy coding, which allows to end this trade-off between speed and rate: the recent implementation [1] provides about $50\\%$ faster decoding than HC for 256 size alphabet, with compression rate similar to provided by AC. This advantage is due to being simpler than AC: using single natural number as the state, instead of two to represent a range. Beside simplifying renormalization, it allows to put the entire behavior for given probability distribution into a relatively small table: defining entropy coding automaton. The memory cost of such table for 256 size alphabet is a few kilobytes. There is a large freedom while choosing a specific table - using pseudorandom number generator initialized with cryptographic key for this purpose allows to simultaneously encrypt the data. This article also introduces and discusses many other variants of this new entropy coding approach, which can provide direct alternatives for standard AC, for large alphabet range coding, or for approximated quasi arithmetic coding.", "score": 0.0, "mention_count": 9, "user_count": 9, "months": ["2016-07", "2016-08", "2017-07", "2017-08", "2022-05", "2024-03"], "last_mentioned": "2024-03-15T23:53:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1311.2901", "slug": "5051fd885d810f9a", "title": "Visualizing and Understanding Convolutional Networks", "authors": ["Matthew D Zeiler", "Rob Fergus"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1311.2901", "doi": null, "topics": ["Computer vision", "Machine learning", "Human–computer interaction"], "description": "Large Convolutional Network models have recently demonstrated impressive classification performance on the ImageNet benchmark.", "description_method": "abstract-excerpt", "abstract": "Large Convolutional Network models have recently demonstrated impressive classification performance on the ImageNet benchmark. However there is no clear understanding of why they perform so well, or how they might be improved. In this paper we address both issues. We introduce a novel visualization technique that gives insight into the function of intermediate feature layers and the operation of the classifier. We also perform an ablation study to discover the performance contribution from different model layers. This enables us to find model architectures that outperform Krizhevsky \\etal on the ImageNet classification benchmark. We show our ImageNet model generalizes well to other datasets: when the softmax classifier is retrained, it convincingly beats the current state-of-the-art results on Caltech-101 and Caltech-256 datasets.", "score": 0.0, "mention_count": 9, "user_count": 9, "months": ["2015-01", "2015-07", "2016-08", "2017-02", "2018-02", "2023-09", "2023-10", "2024-03"], "last_mentioned": "2024-03-19T07:47:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1409.3215", "slug": "d01e8a1640b76acd", "title": "Sequence to Sequence Learning with Neural Networks", "authors": ["Ilya Sutskever", "Oriol Vinyals", "Quoc V. Le"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1409.3215", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "Deep Neural Networks (DNNs) are powerful models that have achieved excellent performance on difficult learning tasks.", "description_method": "abstract-excerpt", "abstract": "Deep Neural Networks (DNNs) are powerful models that have achieved excellent performance on difficult learning tasks. Although DNNs work well whenever large labeled training sets are available, they cannot be used to map sequences to sequences. In this paper, we present a general end-to-end approach to sequence learning that makes minimal assumptions on the sequence structure. Our method uses a multilayered Long Short-Term Memory (LSTM) to map the input sequence to a vector of a fixed dimensionality, and then another deep LSTM to decode the target sequence from the vector. Our main result is that on an English to French translation task from the WMT'14 dataset, the translations produced by the LSTM achieve a BLEU score of 34.8 on the entire test set, where the LSTM's BLEU score was penalized on out-of-vocabulary words. Additionally, the LSTM did not have difficulty on long sentences. For comparison, a phrase-based SMT system achieves a BLEU score of 33.3 on the same dataset. When we used the LSTM to rerank the 1000 hypotheses produced by the aforementioned SMT system, its BLEU score increases to 36.5, which is close to the previous best result on this task. The LSTM also learned sensible phrase and sentence representations that are sensitive to word order and are relatively invariant to the active and the passive voice. Finally, we found that reversing the order of the words in all source sentences (but not target sentences) improved the LSTM's performance markedly, because doing so introduced many short term dependencies between the source and the target sentence which made the optimization problem easier.", "score": 0.0, "mention_count": 9, "user_count": 8, "months": ["2015-01", "2015-08", "2016-05", "2017-07", "2018-03", "2020-12", "2023-06", "2023-10", "2025-04"], "last_mentioned": "2025-04-30T18:14:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1412.0233", "slug": "71fcb6fb0595c5ec", "title": "The Loss Surfaces of Multilayer Networks", "authors": ["Anna Choromanska", "Mikael Henaff", "Michael Mathieu", "Gérard Ben Arous", "Yann LeCun"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1412.0233", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Psychology"], "description": "We study the connection between the highly non-convex loss function of a simple model of the fully-connected feed-forward neural network and the Hamiltonian of the spherical spin-glass model under the assumptions of: i) variable independence, ii) redundancy in network parametrization, and iii) uniformity.", "description_method": "abstract-excerpt", "abstract": "We study the connection between the highly non-convex loss function of a simple model of the fully-connected feed-forward neural network and the Hamiltonian of the spherical spin-glass model under the assumptions of: i) variable independence, ii) redundancy in network parametrization, and iii) uniformity. These assumptions enable us to explain the complexity of the fully decoupled neural network through the prism of the results from random matrix theory. We show that for large-size decoupled networks the lowest critical values of the random loss function form a layered structure and they are located in a well-defined band lower-bounded by the global minimum. The number of local minima outside that band diminishes exponentially with the size of the network. We empirically verify that the mathematical model exhibits similar behavior as the computer simulations, despite the presence of high dependencies in real networks. We conjecture that both simulated annealing and SGD converge to the band of low critical points, and that all critical points found there are local minima of high quality measured by the test error. This emphasizes a major difference between large- and small-size networks where for the latter poor quality local minima have non-zero probability of being recovered. Finally, we prove that recovering the global minimum becomes harder as the network size increases and that it is in practice irrelevant as global minimum often leads to overfitting.", "score": 0.0, "mention_count": 9, "user_count": 6, "months": ["2015-03", "2015-08", "2016-02", "2016-08", "2017-05", "2017-12", "2020-08"], "last_mentioned": "2020-08-27T13:35:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1611.02779", "slug": "75154ba5686a380a", "title": "RL$^2$: Fast Reinforcement Learning via Slow Reinforcement Learning", "authors": ["Yan Duan", "John Schulman", "Xi Chen", "Peter L. Bartlett", "Ilya Sutskever", "Pieter Abbeel"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1611.02779", "doi": null, "topics": ["Machine learning", "Algorithms", "Psychology"], "description": "Deep reinforcement learning (deep RL) has been successful in learning sophisticated behaviors automatically; however, the learning process requires a huge number of trials.", "description_method": "abstract-excerpt", "abstract": "Deep reinforcement learning (deep RL) has been successful in learning sophisticated behaviors automatically; however, the learning process requires a huge number of trials. In contrast, animals can learn new tasks in just a few trials, benefiting from their prior knowledge about the world. This paper seeks to bridge this gap. Rather than designing a \"fast\" reinforcement learning algorithm, we propose to represent it as a recurrent neural network (RNN) and learn it from data. In our proposed method, RL$^2$, the algorithm is encoded in the weights of the RNN, which are learned slowly through a general-purpose (\"slow\") RL algorithm. The RNN receives all information a typical RL algorithm would receive, including observations, actions, rewards, and termination flags; and it retains its state across episodes in a given Markov Decision Process (MDP). The activations of the RNN store the state of the \"fast\" RL algorithm on the current (previously unseen) MDP. We evaluate RL$^2$ experimentally on both small-scale and large-scale problems. On the small-scale side, we train it to solve randomly generated multi-arm bandit problems and finite MDPs. After RL$^2$ is trained, its performance on new MDPs is close to human-designed algorithms with optimality guarantees. On the large-scale side, we test RL$^2$ on a vision-based navigation task and show that it scales up to high-dimensional problems.", "score": 0.0, "mention_count": 9, "user_count": 8, "months": ["2016-11", "2016-12", "2017-01", "2017-02", "2017-06", "2019-03"], "last_mentioned": "2019-03-28T21:24:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1711.11561", "slug": "8d89c548efb00c26", "title": "Measuring the tendency of CNNs to Learn Surface Statistical Regularities", "authors": ["Jason Jo", "Yoshua Bengio"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1711.11561", "doi": null, "topics": ["Statistics & probability", "Machine learning"], "description": "Deep CNNs are known to exhibit the following peculiarity: on the one hand they generalize extremely well to a test set, while on the other hand they are extremely sensitive to so-called adversarial perturbations.", "description_method": "abstract-excerpt", "abstract": "Deep CNNs are known to exhibit the following peculiarity: on the one hand they generalize extremely well to a test set, while on the other hand they are extremely sensitive to so-called adversarial perturbations. The extreme sensitivity of high performance CNNs to adversarial examples casts serious doubt that these networks are learning high level abstractions in the dataset. We are concerned with the following question: How can a deep CNN that does not learn any high level semantics of the dataset manage to generalize so well? The goal of this article is to measure the tendency of CNNs to learn surface statistical regularities of the dataset. To this end, we use Fourier filtering to construct datasets which share the exact same high level abstractions but exhibit qualitatively different surface statistical regularities. For the SVHN and CIFAR-10 datasets, we present two Fourier filtered variants: a low frequency variant and a randomly filtered variant. Each of the Fourier filtering schemes is tuned to preserve the recognizability of the objects. Our main finding is that CNNs exhibit a tendency to latch onto the Fourier image statistics of the training dataset, sometimes exhibiting up to a 28% generalization gap across the various test sets. Moreover, we observe that significantly increasing the depth of a network has a very marginal impact on closing the aforementioned generalization gap. Thus we provide quantitative evidence supporting the hypothesis that deep CNNs tend to learn surface statistical regularities in the dataset rather than higher-level abstract concepts.", "score": 0.0, "mention_count": 9, "user_count": 7, "months": ["2018-01", "2018-06", "2018-07", "2020-09"], "last_mentioned": "2020-09-24T15:16:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1910.10683", "slug": "3933efffc27a9ebc", "title": "Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer", "authors": ["Colin Raffel", "Noam Shazeer", "Adam Roberts", "Katherine Lee", "Sharan Narang", "Michael Matena", "Yanqi Zhou", "Wei Li", "Peter J. Liu"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1910.10683", "doi": null, "topics": ["Language models"], "description": "Transfer learning, where a model is first pre-trained on a data-rich task before being fine-tuned on a downstream task, has emerged as a powerful technique in natural language processing (NLP).", "description_method": "abstract-excerpt", "abstract": "Transfer learning, where a model is first pre-trained on a data-rich task before being fine-tuned on a downstream task, has emerged as a powerful technique in natural language processing (NLP). The effectiveness of transfer learning has given rise to a diversity of approaches, methodology, and practice. In this paper, we explore the landscape of transfer learning techniques for NLP by introducing a unified framework that converts all text-based language problems into a text-to-text format. Our systematic study compares pre-training objectives, architectures, unlabeled data sets, transfer approaches, and other factors on dozens of language understanding tasks. By combining the insights from our exploration with scale and our new ``Colossal Clean Crawled Corpus'', we achieve state-of-the-art results on many benchmarks covering summarization, question answering, text classification, and more. To facilitate future work on transfer learning for NLP, we release our data set, pre-trained models, and code.", "score": 0.0, "mention_count": 9, "user_count": 9, "months": ["2019-10", "2020-01", "2020-02", "2022-12", "2023-02", "2023-07", "2024-02"], "last_mentioned": "2024-02-21T14:32:31Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2008.02217", "slug": "bf2638e486294041", "title": "Hopfield Networks is All You Need", "authors": ["Hubert Ramsauer", "Bernhard Schäfl", "Johannes Lehner", "Philipp Seidl", "Michael Widrich", "Thomas Adler", "Lukas Gruber", "Markus Holzleitner", "Milena Pavlović", "Geir Kjetil Sandve", "Victor Greiff", "David Kreil", "Michael Kopp", "Günter Klambauer", "Johannes Brandstetter", "Sepp Hochreiter"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2008.02217", "doi": null, "topics": ["Machine learning", "Language models", "Computer vision"], "description": "We introduce a modern Hopfield network with continuous states and a corresponding update rule.", "description_method": "abstract-excerpt", "abstract": "We introduce a modern Hopfield network with continuous states and a corresponding update rule. The new Hopfield network can store exponentially (with the dimension of the associative space) many patterns, retrieves the pattern with one update, and has exponentially small retrieval errors. It has three types of energy minima (fixed points of the update): (1) global fixed point averaging over all patterns, (2) metastable states averaging over a subset of patterns, and (3) fixed points which store a single pattern. The new update rule is equivalent to the attention mechanism used in transformers. This equivalence enables a characterization of the heads of transformer models. These heads perform in the first layers preferably global averaging and in higher layers partial averaging via metastable states. The new modern Hopfield network can be integrated into deep learning architectures as layers to allow the storage of and access to raw input data, intermediate results, or learned prototypes. These Hopfield layers enable new ways of deep learning, beyond fully-connected, convolutional, or recurrent networks, and provide pooling, memory, association, and attention mechanisms. We demonstrate the broad applicability of the Hopfield layers across various domains. Hopfield layers improved state-of-the-art on three out of four considered multiple instance learning problems as well as on immune repertoire classification with several hundreds of thousands of instances. On the UCI benchmark collections of small classification tasks, where deep learning methods typically struggle, Hopfield layers yielded a new state-of-the-art when compared to different machine learning methods. Finally, Hopfield layers achieved state-of-the-art on two drug design datasets. The implementation is available at: https://github.com/ml-jku/hopfield-layers", "score": 0.0, "mention_count": 9, "user_count": 9, "months": ["2020-12", "2021-08", "2022-02", "2022-09", "2023-07", "2024-03", "2025-02"], "last_mentioned": "2025-02-15T03:04:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2012.00152", "slug": "1c07a476ca306288", "title": "Every Model Learned by Gradient Descent Is Approximately a Kernel Machine", "authors": ["Pedro Domingos"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2012.00152", "doi": null, "topics": ["Machine learning", "Computer systems", "Algorithms"], "description": "Deep learning's successes are often attributed to its ability to automatically discover new representations of the data, rather than relying on handcrafted features like other learning methods.", "description_method": "abstract-excerpt", "abstract": "Deep learning's successes are often attributed to its ability to automatically discover new representations of the data, rather than relying on handcrafted features like other learning methods. We show, however, that deep networks learned by the standard gradient descent algorithm are in fact mathematically approximately equivalent to kernel machines, a learning method that simply memorizes the data and uses it directly for prediction via a similarity function (the kernel). This greatly enhances the interpretability of deep network weights, by elucidating that they are effectively a superposition of the training examples. The network architecture incorporates knowledge of the target function into the kernel. This improved understanding should lead to better learning algorithms.", "score": 0.0, "mention_count": 9, "user_count": 8, "months": ["2020-12", "2021-10", "2022-03", "2022-04", "2022-09", "2023-08", "2023-10", "2024-03"], "last_mentioned": "2024-03-28T13:59:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2112.10752", "slug": "0ccdd0c0bb7b8381", "title": "High-Resolution Image Synthesis with Latent Diffusion Models", "authors": ["Robin Rombach", "Andreas Blattmann", "Dominik Lorenz", "Patrick Esser", "Björn Ommer"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2112.10752", "doi": null, "topics": ["Physics", "Machine learning", "Computer vision"], "description": "By decomposing the image formation process into a sequential application of denoising autoencoders, diffusion models (DMs) achieve state-of-the-art synthesis results on image data and beyond.", "description_method": "abstract-excerpt", "abstract": "By decomposing the image formation process into a sequential application of denoising autoencoders, diffusion models (DMs) achieve state-of-the-art synthesis results on image data and beyond. Additionally, their formulation allows for a guiding mechanism to control the image generation process without retraining. However, since these models typically operate directly in pixel space, optimization of powerful DMs often consumes hundreds of GPU days and inference is expensive due to sequential evaluations. To enable DM training on limited computational resources while retaining their quality and flexibility, we apply them in the latent space of powerful pretrained autoencoders. In contrast to previous work, training diffusion models on such a representation allows for the first time to reach a near-optimal point between complexity reduction and detail preservation, greatly boosting visual fidelity. By introducing cross-attention layers into the model architecture, we turn diffusion models into powerful and flexible generators for general conditioning inputs such as text or bounding boxes and high-resolution synthesis becomes possible in a convolutional manner. Our latent diffusion models (LDMs) achieve a new state of the art for image inpainting and highly competitive performance on various tasks, including unconditional image generation, semantic scene synthesis, and super-resolution, while significantly reducing computational requirements compared to pixel-based DMs. Code is available at https://github.com/CompVis/latent-diffusion .", "score": 0.0, "mention_count": 9, "user_count": 8, "months": ["2022-08", "2022-09", "2023-03", "2023-04", "2023-07", "2024-02", "2024-07", "2025-12"], "last_mentioned": "2025-12-30T20:37:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.03827", "slug": "4dd99a1300329193", "title": "Discovering Latent Knowledge in Language Models Without Supervision", "authors": ["Collin Burns", "Haotian Ye", "Dan Klein", "Jacob Steinhardt"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.03827", "doi": null, "topics": ["Language models", "Distributed systems"], "description": "Existing techniques for training language models can be misaligned with the truth: if we train models with imitation learning, they may reproduce errors that humans make; if we train them to generate text that humans rate highly, they may output errors that human evaluators can't detect.", "description_method": "abstract-excerpt", "abstract": "Existing techniques for training language models can be misaligned with the truth: if we train models with imitation learning, they may reproduce errors that humans make; if we train them to generate text that humans rate highly, they may output errors that human evaluators can't detect. We propose circumventing this issue by directly finding latent knowledge inside the internal activations of a language model in a purely unsupervised way. Specifically, we introduce a method for accurately answering yes-no questions given only unlabeled model activations. It works by finding a direction in activation space that satisfies logical consistency properties, such as that a statement and its negation have opposite truth values. We show that despite using no supervision and no model outputs, our method can recover diverse knowledge represented in large language models: across 6 models and 10 question-answering datasets, it outperforms zero-shot accuracy by 4\\% on average. We also find that it cuts prompt sensitivity in half and continues to maintain high accuracy even when models are prompted to generate incorrect answers. Our results provide an initial step toward discovering what language models know, distinct from what they say, even when we don't have access to explicit ground truth labels.", "score": 0.0, "mention_count": 9, "user_count": 5, "months": ["2022-12", "2023-01", "2023-02", "2023-03"], "last_mentioned": "2023-03-15T19:39:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2304.13734", "slug": "62d2fb4d07af8ecf", "title": "The Internal State of an LLM Knows When It's Lying", "authors": ["Amos Azaria", "Tom Mitchell"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2304.13734", "doi": null, "topics": ["Language models", "Statistics & probability", "Machine learning"], "description": "While Large Language Models (LLMs) have shown exceptional performance in various tasks, one of their most prominent drawbacks is generating inaccurate or false information with a confident tone.", "description_method": "abstract-excerpt", "abstract": "While Large Language Models (LLMs) have shown exceptional performance in various tasks, one of their most prominent drawbacks is generating inaccurate or false information with a confident tone. In this paper, we provide evidence that the LLM's internal state can be used to reveal the truthfulness of statements. This includes both statements provided to the LLM, and statements that the LLM itself generates. Our approach is to train a classifier that outputs the probability that a statement is truthful, based on the hidden layer activations of the LLM as it reads or generates the statement. Experiments demonstrate that given a set of test sentences, of which half are true and half false, our trained classifier achieves an average of 71\\% to 83\\% accuracy labeling which sentences are true versus false, depending on the LLM base model. Furthermore, we explore the relationship between our classifier's performance and approaches based on the probability assigned to the sentence by the LLM. We show that while LLM-assigned sentence probability is related to sentence truthfulness, this probability is also dependent on sentence length and the frequencies of words in the sentence, resulting in our trained classifier providing a more reliable approach to detecting truthfulness, highlighting its potential to enhance the reliability of LLM-generated content and its practical applicability in real-world scenarios.", "score": 0.0, "mention_count": 9, "user_count": 7, "months": ["2023-08", "2023-09", "2024-05", "2024-08", "2024-09", "2024-10", "2026-02"], "last_mentioned": "2026-02-06T02:40:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2404.04125", "slug": "cb6efd495e85c90c", "title": "No \"Zero-Shot\" Without Exponential Data: Pretraining Concept Frequency Determines Multimodal Model Performance", "authors": ["Vishaal Udandarao", "Ameya Prabhu", "Adhiraj Ghosh", "Yash Sharma", "Philip H. S. Torr", "Adel Bibi", "Samuel Albanie", "Matthias Bethge"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2404.04125", "doi": null, "topics": ["Machine learning", "Physics", "Computer vision"], "description": "Web-crawled pretraining datasets underlie the impressive \"zero-shot\" evaluation performance of multimodal models, such as CLIP for classification/retrieval and Stable-Diffusion for image generation.", "description_method": "abstract-excerpt", "abstract": "Web-crawled pretraining datasets underlie the impressive \"zero-shot\" evaluation performance of multimodal models, such as CLIP for classification/retrieval and Stable-Diffusion for image generation. However, it is unclear how meaningful the notion of \"zero-shot\" generalization is for such multimodal models, as it is not known to what extent their pretraining datasets encompass the downstream concepts targeted for during \"zero-shot\" evaluation. In this work, we ask: How is the performance of multimodal models on downstream concepts influenced by the frequency of these concepts in their pretraining datasets? We comprehensively investigate this question across 34 models and five standard pretraining datasets (CC-3M, CC-12M, YFCC-15M, LAION-400M, LAION-Aesthetics), generating over 300GB of data artifacts. We consistently find that, far from exhibiting \"zero-shot\" generalization, multimodal models require exponentially more data to achieve linear improvements in downstream \"zero-shot\" performance, following a sample inefficient log-linear scaling trend. This trend persists even when controlling for sample-level similarity between pretraining and downstream datasets, and testing on purely synthetic data distributions. Furthermore, upon benchmarking models on long-tailed data sampled based on our analysis, we demonstrate that multimodal models across the board perform poorly. We contribute this long-tail test set as the \"Let it Wag!\" benchmark to further research in this direction. Taken together, our study reveals an exponential need for training data which implies that the key to \"zero-shot\" generalization capabilities under large-scale training paradigms remains to be found.", "score": 0.0, "mention_count": 9, "user_count": 7, "months": ["2024-05", "2024-06", "2024-07", "2024-11", "2026-02"], "last_mentioned": "2026-02-09T14:27:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2407.21783", "slug": "de3cebb4c28bc5d4", "title": "The Llama 3 Herd of Models", "authors": ["Aaron Grattafiori", "Abhimanyu Dubey", "Abhinav Jauhri", "Abhinav Pandey", "Abhishek Kadian", "Ahmad Al-Dahle", "Aiesha Letman", "Akhil Mathur", "Alan Schelten", "Alex Vaughan", "Amy Yang", "Angela Fan", "Anirudh Goyal", "Anthony Hartshorn", "Aobo Yang", "Archi Mitra", "Archie Sravankumar", "Artem Korenev", "Arthur Hinsvark", "Arun Rao", "Aston Zhang", "Aurelien Rodriguez", "Austen Gregerson", "Ava Spataru", "Baptiste Roziere", "Bethany Biron", "Binh Tang", "Bobbie Chern", "Charlotte Caucheteux", "Chaya Nayak", "Chloe Bi", "Chris Marra", "Chris McConnell", "Christian Keller", "Christophe Touret", "Chunyang Wu", "Corinne Wong", "Cristian Canton Ferrer", "Cyrus Nikolaidis", "Damien Allonsius", "Daniel Song", "Danielle Pintz", "Danny Livshits", "Danny Wyatt", "David Esiobu", "Dhruv Choudhary", "Dhruv Mahajan", "Diego Garcia-Olano", "Diego Perino", "Dieuwke Hupkes", "Egor Lakomkin", "Ehab AlBadawy", "Elina Lobanova", "Emily Dinan", "Eric Michael Smith", "Filip Radenovic", "Francisco Guzmán", "Frank Zhang", "Gabriel Synnaeve", "Gabrielle Lee", "Georgia Lewis Anderson", "Govind Thattai", "Graeme Nail", "Gregoire Mialon", "Guan Pang", "Guillem Cucurell", "Hailey Nguyen", "Hannah Korevaar", "Hu Xu", "Hugo Touvron", "Iliyan Zarov", "Imanol Arrieta Ibarra", "Isabel Kloumann", "Ishan Misra", "Ivan Evtimov", "Jack Zhang", "Jade Copet", "Jaewon Lee", "Jan Geffert", "Jana Vranes", "Jason Park", "Jay Mahadeokar", "Jeet Shah", "Jelmer van der Linde", "Jennifer Billock", "Jenny Hong", "Jenya Lee", "Jeremy Fu", "Jianfeng Chi", "Jianyu Huang", "Jiawen Liu", "Jie Wang", "Jiecao Yu", "Joanna Bitton", "Joe Spisak", "Jongsoo Park", "Joseph Rocca", "Joshua Johnstun", "Joshua Saxe", "Junteng Jia", "Kalyan Vasuden Alwala", "Karthik Prasad", "Kartikeya Upasani", "Kate Plawiak", "Ke Li", "Kenneth Heafield", "Kevin Stone", "Khalid El-Arini", "Krithika Iyer", "Kshitiz Malik", "Kuenley Chiu", "Kunal Bhalla", "Kushal Lakhotia", "Lauren Rantala-Yeary", "Laurens van der Maaten", "Lawrence Chen", "Liang Tan", "Liz Jenkins", "Louis Martin", "Lovish Madaan", "Lubo Malo", "Lukas Blecher", "Lukas Landzaat", "Luke de Oliveira", "Madeline Muzzi", "Mahesh Pasupuleti", "Mannat Singh", "Manohar Paluri", "Marcin Kardas", "Maria Tsimpoukelli", "Mathew Oldham", "Mathieu Rita", "Maya Pavlova", "Melanie Kambadur", "Mike Lewis", "Min Si", "Mitesh Kumar Singh", "Mona Hassan", "Naman Goyal", "Narjes Torabi", "Nikolay Bashlykov", "Nikolay Bogoychev", "Niladri Chatterji", "Ning Zhang", "Olivier Duchenne", "Onur Çelebi", "Patrick Alrassy", "Pengchuan Zhang", "Pengwei Li", "Petar Vasic", "Peter Weng", "Prajjwal Bhargava", "Pratik Dubal", "Praveen Krishnan", "Punit Singh Koura", "Puxin Xu", "Qing He", "Qingxiao Dong", "Ragavan Srinivasan", "Raj Ganapathy", "Ramon Calderer", "Ricardo Silveira Cabral", "Robert Stojnic", "Roberta Raileanu", "Rohan Maheswari", "Rohit Girdhar", "Rohit Patel", "Romain Sauvestre", "Ronnie Polidoro", "Roshan Sumbaly", "Ross Taylor", "Ruan Silva", "Rui Hou", "Rui Wang", "Saghar Hosseini", "Sahana Chennabasappa", "Sanjay Singh", "Sean Bell", "Seohyun Sonia Kim", "Sergey Edunov", "Shaoliang Nie", "Sharan Narang", "Sharath Raparthy", "Sheng Shen", "Shengye Wan", "Shruti Bhosale", "Shun Zhang", "Simon Vandenhende", "Soumya Batra", "Spencer Whitman", "Sten Sootla", "Stephane Collot", "Suchin Gururangan", "Sydney Borodinsky", "Tamar Herman", "Tara Fowler", "Tarek Sheasha", "Thomas Georgiou", "Thomas Scialom", "Tobias Speckbacher", "Todor Mihaylov", "Tong Xiao", "Ujjwal Karn", "Vedanuj Goswami", "Vibhor Gupta", "Vignesh Ramanathan", "Viktor Kerkez", "Vincent Gonguet", "Virginie Do", "Vish Vogeti", "Vítor Albiero", "Vladan Petrovic", "Weiwei Chu", "Wenhan Xiong", "Wenyin Fu", "Whitney Meers", "Xavier Martinet", "Xiaodong Wang", "Xiaofang Wang", "Xiaoqing Ellen Tan", "Xide Xia", "Xinfeng Xie", "Xuchao Jia", "Xuewei Wang", "Yaelle Goldschlag", "Yashesh Gaur", "Yasmine Babaei", "Yi Wen", "Yiwen Song", "Yuchen Zhang", "Yue Li", "Yuning Mao", "Zacharie Delpierre Coudert", "Zheng Yan", "Zhengxing Chen", "Zoe Papakipos", "Aaditya Singh", "Aayushi Srivastava", "Abha Jain", "Adam Kelsey", "Adam Shajnfeld", "Adithya Gangidi", "Adolfo Victoria", "Ahuva Goldstand", "Ajay Menon", "Ajay Sharma", "Alex Boesenberg", "Alexei Baevski", "Allie Feinstein", "Amanda Kallet", "Amit Sangani", "Amos Teo", "Anam Yunus", "Andrei Lupu", "Andres Alvarado", "Andrew Caples", "Andrew Gu", "Andrew Ho", "Andrew Poulton", "Andrew Ryan", "Ankit Ramchandani", "Annie Dong", "Annie Franco", "Anuj Goyal", "Aparajita Saraf", "Arkabandhu Chowdhury", "Ashley Gabriel", "Ashwin Bharambe", "Assaf Eisenman", "Azadeh Yazdan", "Beau James", "Ben Maurer", "Benjamin Leonhardi", "Bernie Huang", "Beth Loyd", "Beto De Paola", "Bhargavi Paranjape", "Bing Liu", "Bo Wu", "Boyu Ni", "Braden Hancock", "Bram Wasti", "Brandon Spence", "Brani Stojkovic", "Brian Gamido", "Britt Montalvo", "Carl Parker", "Carly Burton", "Catalina Mejia", "Ce Liu", "Changhan Wang", "Changkyu Kim", "Chao Zhou", "Chester Hu", "Ching-Hsiang Chu", "Chris Cai", "Chris Tindal", "Christoph Feichtenhofer", "Cynthia Gao", "Damon Civin", "Dana Beaty", "Daniel Kreymer", "Daniel Li", "David Adkins", "David Xu", "Davide Testuggine", "Delia David", "Devi Parikh", "Diana Liskovich", "Didem Foss", "Dingkang Wang", "Duc Le", "Dustin Holland", "Edward Dowling", "Eissa Jamil", "Elaine Montgomery", "Eleonora Presani", "Emily Hahn", "Emily Wood", "Eric-Tuan Le", "Erik Brinkman", "Esteban Arcaute", "Evan Dunbar", "Evan Smothers", "Fei Sun", "Felix Kreuk", "Feng Tian", "Filippos Kokkinos", "Firat Ozgenel", "Francesco Caggioni", "Frank Kanayet", "Frank Seide", "Gabriela Medina Florez", "Gabriella Schwarz", "Gada Badeer", "Georgia Swee", "Gil Halpern", "Grant Herman", "Grigory Sizov", " Guangyi", " Zhang", "Guna Lakshminarayanan", "Hakan Inan", "Hamid Shojanazeri", "Han Zou", "Hannah Wang", "Hanwen Zha", "Haroun Habeeb", "Harrison Rudolph", "Helen Suk", "Henry Aspegren", "Hunter Goldman", "Hongyuan Zhan", "Ibrahim Damlaj", "Igor Molybog", "Igor Tufanov", "Ilias Leontiadis", "Irina-Elena Veliche", "Itai Gat", "Jake Weissman", "James Geboski", "James Kohli", "Janice Lam", "Japhet Asher", "Jean-Baptiste Gaya", "Jeff Marcus", "Jeff Tang", "Jennifer Chan", "Jenny Zhen", "Jeremy Reizenstein", "Jeremy Teboul", "Jessica Zhong", "Jian Jin", "Jingyi Yang", "Joe Cummings", "Jon Carvill", "Jon Shepard", "Jonathan McPhie", "Jonathan Torres", "Josh Ginsburg", "Junjie Wang", "Kai Wu", "Kam Hou U", "Karan Saxena", "Kartikay Khandelwal", "Katayoun Zand", "Kathy Matosich", "Kaushik Veeraraghavan", "Kelly Michelena", "Keqian Li", "Kiran Jagadeesh", "Kun Huang", "Kunal Chawla", "Kyle Huang", "Lailin Chen", "Lakshya Garg", "Lavender A", "Leandro Silva", "Lee Bell", "Lei Zhang", "Liangpeng Guo", "Licheng Yu", "Liron Moshkovich", "Luca Wehrstedt", "Madian Khabsa", "Manav Avalani", "Manish Bhatt", "Martynas Mankus", "Matan Hasson", "Matthew Lennie", "Matthias Reso", "Maxim Groshev", "Maxim Naumov", "Maya Lathi", "Meghan Keneally", "Miao Liu", "Michael L. Seltzer", "Michal Valko", "Michelle Restrepo", "Mihir Patel", "Mik Vyatskov", "Mikayel Samvelyan", "Mike Clark", "Mike Macey", "Mike Wang", "Miquel Jubert Hermoso", "Mo Metanat", "Mohammad Rastegari", "Munish Bansal", "Nandhini Santhanam", "Natascha Parks", "Natasha White", "Navyata Bawa", "Nayan Singhal", "Nick Egebo", "Nicolas Usunier", "Nikhil Mehta", "Nikolay Pavlovich Laptev", "Ning Dong", "Norman Cheng", "Oleg Chernoguz", "Olivia Hart", "Omkar Salpekar", "Ozlem Kalinli", "Parkin Kent", "Parth Parekh", "Paul Saab", "Pavan Balaji", "Pedro Rittner", "Philip Bontrager", "Pierre Roux", "Piotr Dollar", "Polina Zvyagina", "Prashant Ratanchandani", "Pritish Yuvraj", "Qian Liang", "Rachad Alao", "Rachel Rodriguez", "Rafi Ayub", "Raghotham Murthy", "Raghu Nayani", "Rahul Mitra", "Rangaprabhu Parthasarathy", "Raymond Li", "Rebekkah Hogan", "Robin Battey", "Rocky Wang", "Russ Howes", "Ruty Rinott", "Sachin Mehta", "Sachin Siby", "Sai Jayesh Bondu", "Samyak Datta", "Sara Chugh", "Sara Hunt", "Sargun Dhillon", "Sasha Sidorov", "Satadru Pan", "Saurabh Mahajan", "Saurabh Verma", "Seiji Yamamoto", "Sharadh Ramaswamy", "Shaun Lindsay", "Shaun Lindsay", "Sheng Feng", "Shenghao Lin", "Shengxin Cindy Zha", "Shishir Patil", "Shiva Shankar", "Shuqiang Zhang", "Shuqiang Zhang", "Sinong Wang", "Sneha Agarwal", "Soji Sajuyigbe", "Soumith Chintala", "Stephanie Max", "Stephen Chen", "Steve Kehoe", "Steve Satterfield", "Sudarshan Govindaprasad", "Sumit Gupta", "Summer Deng", "Sungmin Cho", "Sunny Virk", "Suraj Subramanian", "Sy Choudhury", "Sydney Goldman", "Tal Remez", "Tamar Glaser", "Tamara Best", "Thilo Koehler", "Thomas Robinson", "Tianhe Li", "Tianjun Zhang", "Tim Matthews", "Timothy Chou", "Tzook Shaked", "Varun Vontimitta", "Victoria Ajayi", "Victoria Montanez", "Vijai Mohan", "Vinay Satish Kumar", "Vishal Mangla", "Vlad Ionescu", "Vlad Poenaru", "Vlad Tiberiu Mihailescu", "Vladimir Ivanov", "Wei Li", "Wenchen Wang", "Wenwen Jiang", "Wes Bouaziz", "Will Constable", "Xiaocheng Tang", "Xiaojian Wu", "Xiaolan Wang", "Xilun Wu", "Xinbo Gao", "Yaniv Kleinman", "Yanjun Chen", "Ye Hu", "Ye Jia", "Ye Qi", "Yenda Li", "Yilin Zhang", "Ying Zhang", "Yossi Adi", "Youngjin Nam", " Yu", " Wang", "Yu Zhao", "Yuchen Hao", "Yundi Qian", "Yunlu Li", "Yuzi He", "Zach Rait", "Zachary DeVito", "Zef Rosnbrick", "Zhaoduo Wen", "Zhenyu Yang", "Zhiwei Zhao", "Zhiyu Ma"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2407.21783", "doi": null, "topics": ["Language models", "Psychology"], "description": "Modern artificial intelligence (AI) systems are powered by foundation models.", "description_method": "abstract-excerpt", "abstract": "Modern artificial intelligence (AI) systems are powered by foundation models. This paper presents a new set of foundation models, called Llama 3. It is a herd of language models that natively support multilinguality, coding, reasoning, and tool usage. Our largest model is a dense Transformer with 405B parameters and a context window of up to 128K tokens. This paper presents an extensive empirical evaluation of Llama 3. We find that Llama 3 delivers comparable quality to leading language models such as GPT-4 on a plethora of tasks. We publicly release Llama 3, including pre-trained and post-trained versions of the 405B parameter language model and our Llama Guard 3 model for input and output safety. The paper also presents the results of experiments in which we integrate image, video, and speech capabilities into Llama 3 via a compositional approach. We observe this approach performs competitively with the state-of-the-art on image, video, and speech recognition tasks. The resulting models are not yet being broadly released as they are still under development.", "score": 0.0, "mention_count": 9, "user_count": 9, "months": ["2024-08", "2024-10", "2024-12", "2025-02", "2025-03", "2025-05", "2026-07"], "last_mentioned": "2026-07-27T20:06:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2501.00663", "slug": "4007104b27861346", "title": "Titans: Learning to Memorize at Test Time", "authors": ["Ali Behrouz", "Peilin Zhong", "Vahab Mirrokni"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2501.00663", "doi": null, "topics": ["Language models", "Biology & medicine"], "description": "Over more than a decade there has been an extensive research effort on how to effectively utilize recurrent models and attention.", "description_method": "abstract-excerpt", "abstract": "Over more than a decade there has been an extensive research effort on how to effectively utilize recurrent models and attention. While recurrent models aim to compress the data into a fixed-size memory (called hidden state), attention allows attending to the entire context window, capturing the direct dependencies of all tokens. This more accurate modeling of dependencies, however, comes with a quadratic cost, limiting the model to a fixed-length context. We present a new neural long-term memory module that learns to memorize historical context and helps attention to attend to the current context while utilizing long past information. We show that this neural memory has the advantage of fast parallelizable training while maintaining a fast inference. From a memory perspective, we argue that attention due to its limited context but accurate dependency modeling performs as a short-term memory, while neural memory due to its ability to memorize the data, acts as a long-term, more persistent, memory. Based on these two modules, we introduce a new family of architectures, called Titans, and present three variants to address how one can effectively incorporate memory into this architecture. Our experimental results on language modeling, common-sense reasoning, genomics, and time series tasks show that Titans are more effective than Transformers and recent modern linear recurrent models. They further can effectively scale to larger than 2M context window size with higher accuracy in needle-in-haystack tasks compared to baselines.", "score": 0.0, "mention_count": 9, "user_count": 9, "months": ["2025-01", "2025-02", "2025-12", "2026-01", "2026-03", "2026-04"], "last_mentioned": "2026-04-28T11:09:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2507.15855", "slug": "102c1818bff9431d", "title": "Winning Gold at IMO 2025 with a Model-Agnostic Verification-and-Refinement Pipeline", "authors": ["Yichen Huang", "Lin F. Yang"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2507.15855", "doi": null, "topics": ["Mathematics", "Language models"], "description": "The International Mathematical Olympiad (IMO) is widely regarded as the world championship of high-school mathematics.", "description_method": "abstract-excerpt", "abstract": "The International Mathematical Olympiad (IMO) is widely regarded as the world championship of high-school mathematics. IMO problems are renowned for their difficulty and novelty, demanding deep insight, creativity, and rigor. Although large language models perform well on many mathematical benchmarks, they often struggle with Olympiad-level problems. Using carefully designed prompts, we construct a model-agnostic, verification-and-refinement pipeline. We demonstrate its effectiveness on the recent IMO 2025, avoiding data contamination for models released before the competition. Equipped with any of the three leading models -- Gemini 2.5 Pro, Grok-4, or GPT-5 -- our pipeline correctly solved 5 out of the 6 problems ($\\approx$85.7% accuracy). This is in sharp contrast to their baseline accuracies: 31.6% (Gemini 2.5 Pro), 21.4% (Grok-4), and 38.1% (GPT-5), obtained by selecting the best of 32 candidate solutions. The substantial improvement underscores that the path to advanced AI reasoning requires not only developing more powerful base models but also designing effective methodologies to harness their full potential for complex tasks.", "score": 0.0, "mention_count": 9, "user_count": 9, "months": ["2025-07", "2025-08", "2025-09", "2025-10", "2025-11", "2026-01"], "last_mentioned": "2026-01-02T15:32:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:math/9205211", "slug": "c22626b2278a6c76", "title": "Two notes on notation", "authors": ["Donald E. Knuth"], "year": 1992, "venue": "arXiv", "url": "https://arxiv.org/abs/math/9205211", "doi": null, "topics": ["Mathematics"], "description": "The author advocates two specific mathematical notations from his popular course and joint textbook, \"Concrete Mathematics\".", "description_method": "abstract-excerpt", "abstract": "The author advocates two specific mathematical notations from his popular course and joint textbook, \"Concrete Mathematics\". The first of these, extending an idea of Iverson, is the notation \"[P]\" for the function which is 1 when the Boolean condition P is true and 0 otherwise. This notation can encourage and clarify the use of characteristic functions and Kronecker deltas in sums and integrals. The second notation puts Stirling numbers on the same footing as binomial coefficients. Since binomial coefficients are written on two lines in parentheses and read \"n choose k\", Stirling numbers of the first kind should be written on two lines in brackets and read \"n cycle k\", while Stirling numbers of the second kind should be written in braces and read \"n subset k\". (I might say \"n partition k\".) The written form was first suggested by Imanuel Marx. The virtues of this notation are that Stirling partition numbers frequently appear in combinatorics, and that it more clearly presents functional relations similar to those satisfied by binomial coefficients.", "score": 0.0, "mention_count": 9, "user_count": 8, "months": ["2014-04", "2015-04", "2016-06", "2017-10", "2019-10", "2019-11", "2021-04", "2021-06"], "last_mentioned": "2021-06-12T00:31:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/466432a", "slug": "800ebabec11ce3e1", "title": "Ecology: A world without mosquitoes", "authors": ["Janet Fang"], "year": 2010, "venue": "Nature", "url": "https://doi.org/10.1038/466432a", "doi": "10.1038/466432a", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 9, "user_count": 6, "months": ["2020-01", "2020-02", "2022-08", "2023-11", "2025-03", "2025-06", "2025-10"], "last_mentioned": "2025-10-24T18:31:56Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/533452a", "slug": "6aaa37062efa87e1", "title": "1,500 scientists lift the lid on reproducibility", "authors": ["Monya Baker"], "year": 2016, "venue": "Nature", "url": "https://doi.org/10.1038/533452a", "doi": "10.1038/533452a", "topics": ["Science & research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 9, "user_count": 9, "months": ["2021-08", "2021-10", "2022-02", "2022-10", "2023-02", "2024-12", "2025-03", "2025-10", "2026-01"], "last_mentioned": "2026-01-23T18:12:40Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature12711", "slug": "bd611a7062708dff", "title": "Isolation and characterization of a bat SARS-like coronavirus that uses the ACE2 receptor", "authors": ["Xing-Yi Ge", "Jia-Lu Li", "Xing-Lou Yang", "Aleksei A. Chmura", "Guangjian Zhu", "Jonathan H. Epstein", "Jonna K. Mazet", "Ben Hu", "Wei Zhang", "Cheng Peng", "Yu-Ji Zhang", "Chu-Ming Luo", "Bing Tan", "Ning Wang", "Yan Zhu", "Gary Crameri", "Shu-Yi Zhang", "Lin-Fa Wang", "Peter Daszak", "Zheng-Li Shi"], "year": 2013, "venue": "Nature", "url": "https://doi.org/10.1038/nature12711", "doi": "10.1038/nature12711", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 9, "user_count": 5, "months": ["2020-02", "2020-03", "2020-04", "2020-12", "2021-03", "2022-06"], "last_mentioned": "2022-06-25T17:49:29Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/ncomms12172", "slug": "09812052e4776176", "title": "Direct detection of a single photon by humans", "authors": ["Jonathan N. Tinsley", "Maxim I. Molodtsov", "Robert Prevedel", "David Wartmann", "Jofre Espigulé-Pons", "Mattias Lauwers", "Alipasha Vaziri"], "year": 2016, "venue": "Nature Communications", "url": "https://doi.org/10.1038/ncomms12172", "doi": "10.1038/ncomms12172", "topics": ["Statistics & probability", "Physics", "Biology & medicine"], "description": "Abstract Despite investigations for over 70 years, the absolute limits of human vision have remained unclear.", "description_method": "abstract-excerpt", "abstract": "Abstract Despite investigations for over 70 years, the absolute limits of human vision have remained unclear. Rod cells respond to individual photons, yet whether a single-photon incident on the eye can be perceived by a human subject has remained a fundamental open question. Here we report that humans can detect a single-photon incident on the cornea with a probability significantly above chance. This was achieved by implementing a combination of a psychophysics procedure with a quantum light source that can generate single-photon states of light. We further discover that the probability of reporting a single photon is modulated by the presence of an earlier photon, suggesting a priming process that temporarily enhances the effective gain of the visual system on the timescale of seconds.", "score": 0.0, "mention_count": 9, "user_count": 8, "months": ["2017-11", "2021-12", "2023-01", "2024-11", "2025-01", "2025-10"], "last_mentioned": "2025-10-31T01:47:39Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-023-06600-9", "slug": "047a5a9743bd98b2", "title": "Assembly theory explains and quantifies selection and evolution", "authors": ["Abhishek Sharma", "Dániel Czégel", "Michael Lachmann", "Christopher P. Kempes", "Sara I. Walker", "Leroy Cronin"], "year": 2023, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06600-9", "doi": "10.1038/s41586-023-06600-9", "topics": ["Physics", "Biology & medicine"], "description": "Abstract Scientists have grappled with reconciling biological evolution 1,2 with the immutable laws of the Universe defined by physics.", "description_method": "abstract-excerpt", "abstract": "Abstract Scientists have grappled with reconciling biological evolution 1,2 with the immutable laws of the Universe defined by physics. These laws underpin life’s origin, evolution and the development of human culture and technology, yet they do not predict the emergence of these phenomena. Evolutionary theory explains why some things exist and others do not through the lens of selection. To comprehend how diverse, open-ended forms can emerge from physics without an inherent design blueprint, a new approach to understanding and quantifying selection is necessary 3–5 . We present assembly theory (AT) as a framework that does not alter the laws of physics, but redefines the concept of an ‘object’ on which these laws act. AT conceptualizes objects not as point particles, but as entities defined by their possible formation histories. This allows objects to show evidence of selection, within well-defined boundaries of individuals or selected units. We introduce a measure called assembly ( A ), capturing the degree of causation required to produce a given ensemble of objects. This approach enables us to incorporate novelty generation and selection into the physics of complex objects. It explains how these objects can be characterized through a forward dynamical process considering their assembly. By reimagining the concept of matter within assembly spaces, AT provides a powerful interface between physics and biology. It discloses a new aspect of physics emerging at the chemical scale, whereby history and causal contingency influence what exists.", "score": 0.0, "mention_count": 9, "user_count": 7, "months": ["2023-10", "2023-11", "2023-12", "2024-01", "2024-05", "2024-06", "2024-08", "2024-10"], "last_mentioned": "2024-10-18T02:34:48Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-022-01689-3", "slug": "9a8e0bbe8f1942fc", "title": "Long-term cardiovascular outcomes of COVID-19", "authors": ["Yan Xie", "Evan Xu", "Benjamin Bowe", "Ziyad Al-Aly"], "year": 2022, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-022-01689-3", "doi": "10.1038/s41591-022-01689-3", "topics": ["Biology & medicine", "Databases"], "description": "Abstract The cardiovascular complications of acute coronavirus disease 2019 (COVID-19) are well described, but the post-acute cardiovascular manifestations of COVID-19 have not yet been comprehensively characterized.", "description_method": "abstract-excerpt", "abstract": "Abstract The cardiovascular complications of acute coronavirus disease 2019 (COVID-19) are well described, but the post-acute cardiovascular manifestations of COVID-19 have not yet been comprehensively characterized. Here we used national healthcare databases from the US Department of Veterans Affairs to build a cohort of 153,760 individuals with COVID-19, as well as two sets of control cohorts with 5,637,647 (contemporary controls) and 5,859,411 (historical controls) individuals, to estimate risks and 1-year burdens of a set of pre-specified incident cardiovascular outcomes. We show that, beyond the first 30 d after infection, individuals with COVID-19 are at increased risk of incident cardiovascular disease spanning several categories, including cerebrovascular disorders, dysrhythmias, ischemic and non-ischemic heart disease, pericarditis, myocarditis, heart failure and thromboembolic disease. These risks and burdens were evident even among individuals who were not hospitalized during the acute phase of the infection and increased in a graded fashion according to the care setting during the acute phase (non-hospitalized, hospitalized and admitted to intensive care). Our results provide evidence that the risk and 1-year burden of cardiovascular disease in survivors of acute COVID-19 are substantial. Care pathways of those surviving the acute episode of COVID-19 should include attention to cardiovascular health and disease.", "score": 0.0, "mention_count": 9, "user_count": 8, "months": ["2022-02", "2022-03", "2022-09", "2022-11"], "last_mentioned": "2022-11-07T22:18:38Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-018-28904-x", "slug": "edd4a13b1bf748fe", "title": "Reading and Myopia: Contrast Polarity Matters", "authors": ["Andrea C. Aleman", "Min Wang", "Frank Schaeffel"], "year": 2018, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-018-28904-x", "doi": "10.1038/s41598-018-28904-x", "topics": ["Other research"], "description": "AbstractIn myopia the eye grows too long, generating poorly focused retinal images when people try to look at a distance.", "description_method": "abstract-excerpt", "abstract": "AbstractIn myopia the eye grows too long, generating poorly focused retinal images when people try to look at a distance. Myopia is tightly linked to the educational status and is on the rise worldwide. It is still not clear which kind of visual experience stimulates eye growth in children and students when they study. We propose a new and perhaps unexpected reason. Work in animal models has shown that selective activation of ON or OFF pathways has also selective effects on eye growth. This is likely to be true also in humans. Using custom-developed software to process video frames of the visual environment in realtime we quantified relative ON and OFF stimulus strengths. We found that ON and OFF inputs were largely balanced in natural environments. However, black text on white paper heavily overstimulated retinal OFF pathways. Conversely, white text on black paper overstimulated ON pathways. Using optical coherence tomography (OCT) in young human subjects, we found that the choroid, the heavily perfused layer behind the retina in the eye, becomes about 16 µm thinner in only one hour when subjects read black text on white background but about 10 µm thicker when they read white text from black background. Studies both in animal models and in humans have shown that thinner choroids are associated with myopia development and thicker choroids with myopia inhibition. Therefore, reading white text from a black screen or tablet may be a way to inhibit myopia, while conventional black text on white background may stimulate myopia.", "score": 0.0, "mention_count": 9, "user_count": 8, "months": ["2019-11", "2019-12", "2020-06", "2022-12", "2026-05"], "last_mentioned": "2026-05-26T19:21:36Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/srep01376", "slug": "77eea1ac90cb55eb", "title": "Unique in the Crowd: The privacy bounds of human mobility", "authors": ["Yves-Alexandre de Montjoye", "César A. Hidalgo", "Michel Verleysen", "Vincent D. Blondel"], "year": 2013, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/srep01376", "doi": "10.1038/srep01376", "topics": ["Security & cryptography"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 9, "user_count": 7, "months": ["2016-05", "2018-02", "2019-08", "2019-11", "2021-12", "2022-04", "2022-06", "2023-05", "2026-02"], "last_mentioned": "2026-02-26T12:43:09Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3386323", "slug": "b259dc83c9e10031", "title": "Origins of the D programming language", "authors": ["Walter Bright", "Andrei Alexandrescu", "Michael Parker"], "year": 2020, "venue": "Proceedings of the ACM on Programming Languages", "url": "https://doi.org/10.1145/3386323", "doi": "10.1145/3386323", "topics": ["Programming languages", "Economics"], "description": "As its name suggests, the initial motivation for the D programming language was to improve on C and C++ while keeping their spirit.", "description_method": "abstract-excerpt", "abstract": "As its name suggests, the initial motivation for the D programming language was to improve on C and C++ while keeping their spirit. The D language was to preserve those languages' efficiency, low-level access, and Algol-style syntax. The areas D set out to improve focused initially on rapid development, convenience, and simplifying the syntax without hampering expressiveness. The genesis of D has its peculiarities, as is the case with many other languages. Walter Bright, D's creator, is a mechanical engineer by education who started out working for Boeing designing gearboxes for the 757. He was programming games on the side, and in trying to make his game Empire run faster, became interested in compilers. Despite having no experience, Bright set out in 1982 to implement a compiler that produced better code than those on the market at the time. This interest materialized into a C compiler, followed by compilers for C++, Java, and JavaScript. Best known of these would be the Zortech C++ compiler, the first (and to date only) C++-to-native compiler developed by a single person. The D programming language began in 1999 as an effort to pull the best features of these languages into a new one. Fittingly, D would use the by that time mature C/C++ back end (optimizer and code generator) that had been under continued development and maintenance since 1982. Between 1999 and 2006, Bright worked alone on the D language definition and its implementation, although a steadily increasing volume of patches from users was incorporated. The new language would be based on the past successes of the languages he'd used and implemented, but would be clearly looking to the future. D started with choices that are obvious today but were less clear winners back in the 1990s: full support for Unicode, IEEE floating point, 2s complement arithmetic, and flat memory addressing (memory is treated as a linear address space with no segmentation). It would do away with certain compromises from past languages imposed by shortages of memory (for example, forward declarations would not be required). It would primarily appeal to C and C++ users, as expertise with those languages would be readily transferrable. The interface with C was designed to be zero cost. The language design was begun in late 1999. An alpha version appeared in 2001 and the initial language was completed, somewhat arbitrarily, at version 1.0 in January 2007. During that time, the language evolved considerably, both in capability and in the accretion of a substantial worldwide community that became increasingly involved with contributing. The front end was open-sourced in April 2002, and the back end was donated by Symantec to the open source community in 2017. Meanwhile, two additional open-source back ends became mature in the 2010s: `gdc` (using the same back end as the GNU C++ compiler) and `ldc` (using the LLVM back end). The increasing use of the D language in the 2010s created an impetus for formalization and development management. To that end, the D Language Foundation was created in September 2015 as a nonprofit corporation overseeing work on D's definition and implementation, publications, conferences, and collaborations with universities.", "score": 0.0, "mention_count": 9, "user_count": 3, "months": ["2021-01", "2021-05", "2021-06", "2021-09", "2022-02", "2022-04", "2022-06"], "last_mentioned": "2022-06-02T23:57:02Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1203.2847", "slug": "2edfd025b2019539", "title": "Demonstration of Communication using Neutrinos", "authors": ["D. D. Stancil", "P. Adamson", "M. Alania", "L. Aliaga", "M. Andrews", "C. Araujo Del Castillo", "L. Bagby", "J. L. Bazo Alba", "A. Bodek", "D. Boehnlein", "R. Bradford", "W. K. Brooks", "H. Budd", "A. Butkevich", "D. A. M. Caicedo", "D. P. Capista", "C. M. Castromonte", "A. Chamorro", "E. Charlton", "M. E. Christy", "J. Chvojka", "P. D. Conrow", "I. Danko", "M. Day", "J. Devan", "J. M. Downey", "S. A. Dytman", "B. Eberly", "J. R. Fein", "J. Felix", "L. Fields", "G. A. Fiorentini", "A. M. Gago", "H. Gallagher", "R. Gran", "J. Grange", "J. Griffin", "T. Griffin", "E. Hahn", "D. A. Harris", "A. Higuera", "J. A. Hobbs", "C. M. Hoffman", "B. L. Hughes", "K. Hurtado", "A. Judd", "T. Kafka", "K. Kephart", "J. Kilmer", "M. Kordosky", "S. A. Kulagin", "V. A. Kuznetsov", "M. Lanari", "T. Le", "H. Lee", "L. Loiacono", "G. Maggi", "E. Maher", "S. Manly", "W. A. Mann", "C. M. Marshall", "K. S. McFarland", "A. Mislivec", "A. M. McGowan", "J. G. Morfin", "H. da Motta", "J. Mousseau", "J. K. Nelson", "J. A. Niemiec-Gielata", "N. Ochoa", "B. Osmanov", "J. Osta", "J. L. Palomino", "J. S. Paradis", "V. Paolone", "J. Park", "C. Pena", "G. Perdue", "C. E. Perez Lara", "A. M. Peterman", "A. Pla-Dalmau", "B. Pollock", "F. Prokoshin", "R. D. Ransome", "H. Ray", "M. Reyhan", "P. Rubinov", "D. Ruggiero", "O. S. Sands", "H. Schellman", "D. W. Schmitz", "E. C. Schulte", "C. Simon", "C. J. Solano Salinas", "R. Stefanski", "R. G. Stevens", "N. Tagg", "V. Takhistov", "B. G. Tice", "R. N. Tilden", "J. P. Velasquez", "I. Vergalosova", "J. Voirin", "J. Walding", "B. J. Walker", "T. Walton", "J. Wolcott", "T. P. Wytock", "G. Zavala", "D. Zhang", "L. Y. Zhu", "B. P. Ziemer"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1203.2847", "doi": "10.1142/S0217732312500770", "topics": ["Other research"], "description": "Beams of neutrinos have been proposed as a vehicle for communications under unusual circumstances, such as direct point-to-point global communication, communication with submarines, secure communications and interstellar communication.", "description_method": "abstract-excerpt", "abstract": "Beams of neutrinos have been proposed as a vehicle for communications under unusual circumstances, such as direct point-to-point global communication, communication with submarines, secure communications and interstellar communication. We report on the performance of a low-rate communications link established using the NuMI beam line and the MINERvA detector at Fermilab. The link achieved a decoded data rate of 0.1 bits/sec with a bit error rate of 1% over a distance of 1.035 km, including 240 m of earth.", "score": 0.0, "mention_count": 8, "user_count": 7, "months": ["2012-03", "2015-08", "2017-06", "2018-04", "2021-02", "2021-11", "2024-08", "2025-10"], "last_mentioned": "2025-10-26T06:09:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1301.7007", "slug": "4b7e088800cbfcb1", "title": "Pretending to factor large numbers on a quantum computer", "authors": ["John A. Smolin", "Graeme Smith", "Alex Vargo"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1301.7007", "doi": "10.1038/nature12290", "topics": ["Physics", "Algorithms", "Mathematics"], "description": "Shor's algorithm for factoring in polynomial time on a quantum computer\\cite{Shor} gives an enormous advantage over all known classical factoring algorithm.", "description_method": "abstract-excerpt", "abstract": "Shor's algorithm for factoring in polynomial time on a quantum computer\\cite{Shor} gives an enormous advantage over all known classical factoring algorithm. We demonstrate how to factor products of large prime numbers using a compiled version of Shor's quantum factoring algorithm. Our technique can factor all products of $p,q$ such that $p,q$ are unequal primes greater than two, runs in constant time, and requires only two coherent qubits. This illustrates that the correct measure of difficulty when implementing Shor's algorithm is not the size of number factored, but the length of the period found.", "score": 0.0, "mention_count": 8, "user_count": 6, "months": ["2013-04", "2017-12", "2019-09", "2021-05", "2024-01", "2024-05", "2025-11", "2026-01"], "last_mentioned": "2026-01-08T05:59:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1406.2572", "slug": "dc31680550fae6a2", "title": "Identifying and attacking the saddle point problem in high-dimensional non-convex optimization", "authors": ["Yann Dauphin", "Razvan Pascanu", "Caglar Gulcehre", "Kyunghyun Cho", "Surya Ganguli", "Yoshua Bengio"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1406.2572", "doi": null, "topics": ["Algorithms", "Machine learning", "Security & cryptography"], "description": "A central challenge to many fields of science and engineering involves minimizing non-convex error functions over continuous, high dimensional spaces.", "description_method": "abstract-excerpt", "abstract": "A central challenge to many fields of science and engineering involves minimizing non-convex error functions over continuous, high dimensional spaces. Gradient descent or quasi-Newton methods are almost ubiquitously used to perform such minimizations, and it is often thought that a main source of difficulty for these local methods to find the global minimum is the proliferation of local minima with much higher error than the global minimum. Here we argue, based on results from statistical physics, random matrix theory, neural network theory, and empirical evidence, that a deeper and more profound difficulty originates from the proliferation of saddle points, not local minima, especially in high dimensional problems of practical interest. Such saddle points are surrounded by high error plateaus that can dramatically slow down learning, and give the illusory impression of the existence of a local minimum. Motivated by these arguments, we propose a new approach to second-order optimization, the saddle-free Newton method, that can rapidly escape high dimensional saddle points, unlike gradient descent and quasi-Newton methods. We apply this algorithm to deep or recurrent neural network training, and provide numerical evidence for its superior optimization performance.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2015-12", "2016-01", "2017-10", "2017-12", "2019-03", "2020-10", "2026-06"], "last_mentioned": "2026-06-25T14:52:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1504.00702", "slug": "b03c261566903b16", "title": "End-to-End Training of Deep Visuomotor Policies", "authors": ["Sergey Levine", "Chelsea Finn", "Trevor Darrell", "Pieter Abbeel"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1504.00702", "doi": null, "topics": ["Robotics", "Machine learning", "Psychology"], "description": "Policy search methods can allow robots to learn control policies for a wide range of tasks, but practical applications of policy search often require hand-engineered components for perception, state estimation, and low-level control.", "description_method": "abstract-excerpt", "abstract": "Policy search methods can allow robots to learn control policies for a wide range of tasks, but practical applications of policy search often require hand-engineered components for perception, state estimation, and low-level control. In this paper, we aim to answer the following question: does training the perception and control systems jointly end-to-end provide better performance than training each component separately? To this end, we develop a method that can be used to learn policies that map raw image observations directly to torques at the robot's motors. The policies are represented by deep convolutional neural networks (CNNs) with 92,000 parameters, and are trained using a partially observed guided policy search method, which transforms policy search into supervised learning, with supervision provided by a simple trajectory-centric reinforcement learning method. We evaluate our method on a range of real-world manipulation tasks that require close coordination between vision and control, such as screwing a cap onto a bottle, and present simulated comparisons to a range of prior policy search methods.", "score": 0.0, "mention_count": 8, "user_count": 6, "months": ["2015-05", "2015-10", "2015-12", "2016-04", "2016-05", "2017-05"], "last_mentioned": "2017-05-25T12:36:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1506.00494", "slug": "15f14b582c477d26", "title": "Reconciling a Reactionless Propulsive Drive with the First Law of Thermodynamics", "authors": ["Andrew J. Higgins"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1506.00494", "doi": null, "topics": ["Physics"], "description": "A \"space drive\" is a hypothetical device that generates a propulsive force in free space using an input of power without the need for a reaction mass.", "description_method": "abstract-excerpt", "abstract": "A \"space drive\" is a hypothetical device that generates a propulsive force in free space using an input of power without the need for a reaction mass. Any device that generates photons (e.g., a laser) would qualify as a propellantless \"photon rocket,\" but the force generated by emitting photons per power input (3.33 $μ$N/kW) is too small to be a practical propulsion device. The ability to generate greater force per power input would be highly desirable, but, as demonstrated in this paper, such a device would be able to operate as a perpetual motion machine of the first kind. Since applying a constant force results in a constant acceleration, the kinetic energy of a mass driven by such a device increases quadratically with time, while the energy input increases only linearly with time. Thus, at some point, the kinetic energy of the device-driven mass exceeds the energy input, and if this energy is collected via decelerating the mass (via regenerative electromagnetic braking, for example), then there would be a net gain in energy. For devices with thrust-to-power ratios on the order of 1 N/kW that have been discussed recently in connection with the so-called EM drive, this breakeven occurs at velocities low enough to be feasible with current technology, clearly demonstrating the absurdity of such a device. When relativistic effects are taken into account, it is shown that the photon rocket can only reach energy breakeven as the accelerated mass asymptotically approaches the speed of light. Thus, any device with a thrust-to-power ratio greater than the photon rocket would be able to operate as a perpetual motion machine of the first kind, and thus should be excluded by the First Law of Thermodynamics.", "score": 0.0, "mention_count": 8, "user_count": 3, "months": ["2016-12", "2017-04", "2018-05", "2020-09"], "last_mentioned": "2020-09-03T15:07:31Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1511.06434", "slug": "8577b5ed7844c6dd", "title": "Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks", "authors": ["Alec Radford", "Luke Metz", "Soumith Chintala"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1511.06434", "doi": null, "topics": ["Computer vision", "Machine learning"], "description": "In recent years, supervised learning with convolutional networks (CNNs) has seen huge adoption in computer vision applications.", "description_method": "abstract-excerpt", "abstract": "In recent years, supervised learning with convolutional networks (CNNs) has seen huge adoption in computer vision applications. Comparatively, unsupervised learning with CNNs has received less attention. In this work we hope to help bridge the gap between the success of CNNs for supervised learning and unsupervised learning. We introduce a class of CNNs called deep convolutional generative adversarial networks (DCGANs), that have certain architectural constraints, and demonstrate that they are a strong candidate for unsupervised learning. Training on various image datasets, we show convincing evidence that our deep convolutional adversarial pair learns a hierarchy of representations from object parts to scenes in both the generator and discriminator. Additionally, we use the learned features for novel tasks - demonstrating their applicability as general image representations.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2015-12", "2016-03", "2016-04", "2016-06", "2017-07", "2023-02"], "last_mentioned": "2023-02-08T16:19:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1606.08813", "slug": "b8c4e83f168ce462", "title": "European Union regulations on algorithmic decision-making and a \"right to explanation\"", "authors": ["Bryce Goodman", "Seth Flaxman"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1606.08813", "doi": "10.1609/aimag.v38i3.2741", "topics": ["Algorithms", "Psychology", "Machine learning"], "description": "We summarize the potential impact that the European Union's new General Data Protection Regulation will have on the routine use of machine learning algorithms.", "description_method": "abstract-excerpt", "abstract": "We summarize the potential impact that the European Union's new General Data Protection Regulation will have on the routine use of machine learning algorithms. Slated to take effect as law across the EU in 2018, it will restrict automated individual decision-making (that is, algorithms that make decisions based on user-level predictors) which \"significantly affect\" users. The law will also effectively create a \"right to explanation,\" whereby a user can ask for an explanation of an algorithmic decision that was made about them. We argue that while this law will pose large challenges for industry, it highlights opportunities for computer scientists to take the lead in designing algorithms and evaluation frameworks which avoid discrimination and enable explanation.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2016-07", "2016-08", "2016-09", "2017-06", "2018-01", "2018-10"], "last_mentioned": "2018-10-17T08:17:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1607.06520", "slug": "38404b10b093b60a", "title": "Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings", "authors": ["Tolga Bolukbasi", "Kai-Wei Chang", "James Zou", "Venkatesh Saligrama", "Adam Kalai"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1607.06520", "doi": null, "topics": ["Psychology", "Language models", "Machine learning"], "description": "The blind application of machine learning runs the risk of amplifying biases present in data.", "description_method": "abstract-excerpt", "abstract": "The blind application of machine learning runs the risk of amplifying biases present in data. Such a danger is facing us with word embedding, a popular framework to represent text data as vectors which has been used in many machine learning and natural language processing tasks. We show that even word embeddings trained on Google News articles exhibit female/male gender stereotypes to a disturbing extent. This raises concerns because their widespread use, as we describe, often tends to amplify these biases. Geometrically, gender bias is first shown to be captured by a direction in the word embedding. Second, gender neutral words are shown to be linearly separable from gender definition words in the word embedding. Using these properties, we provide a methodology for modifying an embedding to remove gender stereotypes, such as the association between between the words receptionist and female, while maintaining desired associations such as between the words queen and female. We define metrics to quantify both direct and indirect gender biases in embeddings, and develop algorithms to \"debias\" the embedding. Using crowd-worker evaluation as well as standard benchmarks, we empirically demonstrate that our algorithms significantly reduce gender bias in embeddings while preserving the its useful properties such as the ability to cluster related concepts and to solve analogy tasks. The resulting embeddings can be used in applications without amplifying gender bias.", "score": 0.0, "mention_count": 8, "user_count": 6, "months": ["2016-09", "2017-07", "2017-10", "2018-03", "2018-10", "2019-07", "2019-09", "2020-09"], "last_mentioned": "2020-09-18T08:46:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.03499", "slug": "ccf75b613a179354", "title": "WaveNet: A Generative Model for Raw Audio", "authors": ["Aaron van den Oord", "Sander Dieleman", "Heiga Zen", "Karen Simonyan", "Oriol Vinyals", "Alex Graves", "Nal Kalchbrenner", "Andrew Senior", "Koray Kavukcuoglu"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.03499", "doi": null, "topics": ["Machine learning", "Psychology"], "description": "This paper introduces WaveNet, a deep neural network for generating raw audio waveforms.", "description_method": "abstract-excerpt", "abstract": "This paper introduces WaveNet, a deep neural network for generating raw audio waveforms. The model is fully probabilistic and autoregressive, with the predictive distribution for each audio sample conditioned on all previous ones; nonetheless we show that it can be efficiently trained on data with tens of thousands of samples per second of audio. When applied to text-to-speech, it yields state-of-the-art performance, with human listeners rating it as significantly more natural sounding than the best parametric and concatenative systems for both English and Mandarin. A single WaveNet can capture the characteristics of many different speakers with equal fidelity, and can switch between them by conditioning on the speaker identity. When trained to model music, we find that it generates novel and often highly realistic musical fragments. We also show that it can be employed as a discriminative model, returning promising results for phoneme recognition.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2016-11", "2017-03", "2017-10", "2018-05", "2018-08", "2019-04", "2022-08", "2026-01"], "last_mentioned": "2026-01-16T10:21:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.04802", "slug": "ed607d3d6ab6df7c", "title": "Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network", "authors": ["Christian Ledig", "Lucas Theis", "Ferenc Huszar", "Jose Caballero", "Andrew Cunningham", "Alejandro Acosta", "Andrew Aitken", "Alykhan Tejani", "Johannes Totz", "Zehan Wang", "Wenzhe Shi"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.04802", "doi": null, "topics": ["Psychology", "Machine learning", "Computer vision"], "description": "Despite the breakthroughs in accuracy and speed of single image super-resolution using faster and deeper convolutional neural networks, one central problem remains largely unsolved: how do we recover the finer texture details when we super-resolve at large upscaling factors?", "description_method": "abstract-excerpt", "abstract": "Despite the breakthroughs in accuracy and speed of single image super-resolution using faster and deeper convolutional neural networks, one central problem remains largely unsolved: how do we recover the finer texture details when we super-resolve at large upscaling factors? The behavior of optimization-based super-resolution methods is principally driven by the choice of the objective function. Recent work has largely focused on minimizing the mean squared reconstruction error. The resulting estimates have high peak signal-to-noise ratios, but they are often lacking high-frequency details and are perceptually unsatisfying in the sense that they fail to match the fidelity expected at the higher resolution. In this paper, we present SRGAN, a generative adversarial network (GAN) for image super-resolution (SR). To our knowledge, it is the first framework capable of inferring photo-realistic natural images for 4x upscaling factors. To achieve this, we propose a perceptual loss function which consists of an adversarial loss and a content loss. The adversarial loss pushes our solution to the natural image manifold using a discriminator network that is trained to differentiate between the super-resolved images and original photo-realistic images. In addition, we use a content loss motivated by perceptual similarity instead of similarity in pixel space. Our deep residual network is able to recover photo-realistic textures from heavily downsampled images on public benchmarks. An extensive mean-opinion-score (MOS) test shows hugely significant gains in perceptual quality using SRGAN. The MOS scores obtained with SRGAN are closer to those of the original high-resolution images than to those obtained with any state-of-the-art method.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2016-10", "2016-11", "2016-12", "2017-01", "2020-07"], "last_mentioned": "2020-07-06T15:29:44Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1611.01578", "slug": "807da1792a62f792", "title": "Neural Architecture Search with Reinforcement Learning", "authors": ["Barret Zoph", "Quoc V. Le"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1611.01578", "doi": null, "topics": ["Machine learning", "Language models"], "description": "Neural networks are powerful and flexible models that work well for many difficult learning tasks in image, speech and natural language understanding.", "description_method": "abstract-excerpt", "abstract": "Neural networks are powerful and flexible models that work well for many difficult learning tasks in image, speech and natural language understanding. Despite their success, neural networks are still hard to design. In this paper, we use a recurrent network to generate the model descriptions of neural networks and train this RNN with reinforcement learning to maximize the expected accuracy of the generated architectures on a validation set. On the CIFAR-10 dataset, our method, starting from scratch, can design a novel network architecture that rivals the best human-invented architecture in terms of test set accuracy. Our CIFAR-10 model achieves a test error rate of 3.65, which is 0.09 percent better and 1.05x faster than the previous state-of-the-art model that used a similar architectural scheme. On the Penn Treebank dataset, our model can compose a novel recurrent cell that outperforms the widely-used LSTM cell, and other state-of-the-art baselines. Our cell achieves a test set perplexity of 62.4 on the Penn Treebank, which is 3.6 perplexity better than the previous state-of-the-art model. The cell can also be transferred to the character language modeling task on PTB and achieves a state-of-the-art perplexity of 1.214.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2017-01", "2017-02", "2017-07", "2017-11", "2018-03", "2018-06", "2018-07"], "last_mentioned": "2018-07-13T18:55:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1806.07572", "slug": "5f4f15f7438992f3", "title": "Neural Tangent Kernel: Convergence and Generalization in Neural Networks", "authors": ["Arthur Jacot", "Franck Gabriel", "Clément Hongler"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1806.07572", "doi": null, "topics": ["Computer systems", "Machine learning", "Statistics & probability"], "description": "At initialization, artificial neural networks (ANNs) are equivalent to Gaussian processes in the infinite-width limit, thus connecting them to kernel methods.", "description_method": "abstract-excerpt", "abstract": "At initialization, artificial neural networks (ANNs) are equivalent to Gaussian processes in the infinite-width limit, thus connecting them to kernel methods. We prove that the evolution of an ANN during training can also be described by a kernel: during gradient descent on the parameters of an ANN, the network function $f_θ$ (which maps input vectors to output vectors) follows the kernel gradient of the functional cost (which is convex, in contrast to the parameter cost) w.r.t. a new kernel: the Neural Tangent Kernel (NTK). This kernel is central to describe the generalization features of ANNs. While the NTK is random at initialization and varies during training, in the infinite-width limit it converges to an explicit limiting kernel and it stays constant during training. This makes it possible to study the training of ANNs in function space instead of parameter space. Convergence of the training can then be related to the positive-definiteness of the limiting NTK. We prove the positive-definiteness of the limiting NTK when the data is supported on the sphere and the non-linearity is non-polynomial. We then focus on the setting of least-squares regression and show that in the infinite-width limit, the network function $f_θ$ follows a linear differential equation during training. The convergence is fastest along the largest kernel principal components of the input data with respect to the NTK, hence suggesting a theoretical motivation for early stopping. Finally we study the NTK numerically, observe its behavior for wide networks, and compare it to the infinite-width limit.", "score": 0.0, "mention_count": 8, "user_count": 6, "months": ["2019-11", "2020-08", "2020-12", "2021-02", "2021-03", "2024-03"], "last_mentioned": "2024-03-28T13:59:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1906.08237", "slug": "d36f0e6b36dee543", "title": "XLNet: Generalized Autoregressive Pretraining for Language Understanding", "authors": ["Zhilin Yang", "Zihang Dai", "Yiming Yang", "Jaime Carbonell", "Ruslan Salakhutdinov", "Quoc V. Le"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1906.08237", "doi": null, "topics": ["Language models"], "description": "With the capability of modeling bidirectional contexts, denoising autoencoding based pretraining like BERT achieves better performance than pretraining approaches based on autoregressive language modeling.", "description_method": "abstract-excerpt", "abstract": "With the capability of modeling bidirectional contexts, denoising autoencoding based pretraining like BERT achieves better performance than pretraining approaches based on autoregressive language modeling. However, relying on corrupting the input with masks, BERT neglects dependency between the masked positions and suffers from a pretrain-finetune discrepancy. In light of these pros and cons, we propose XLNet, a generalized autoregressive pretraining method that (1) enables learning bidirectional contexts by maximizing the expected likelihood over all permutations of the factorization order and (2) overcomes the limitations of BERT thanks to its autoregressive formulation. Furthermore, XLNet integrates ideas from Transformer-XL, the state-of-the-art autoregressive model, into pretraining. Empirically, under comparable experiment settings, XLNet outperforms BERT on 20 tasks, often by a large margin, including question answering, natural language inference, sentiment analysis, and document ranking.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2019-06", "2019-07", "2019-08", "2020-08", "2022-05", "2023-02"], "last_mentioned": "2023-02-08T16:19:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1908.05647", "slug": "e334a4775c77eb66", "title": "Counting Immutable Beans: Reference Counting Optimized for Purely Functional Programming", "authors": ["Sebastian Ullrich", "Leonardo de Moura"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1908.05647", "doi": null, "topics": ["Programming languages", "Algorithms", "Computer systems"], "description": "Most functional languages rely on some garbage collection for automatic memory management.", "description_method": "abstract-excerpt", "abstract": "Most functional languages rely on some garbage collection for automatic memory management. They usually eschew reference counting in favor of a tracing garbage collector, which has less bookkeeping overhead at runtime. On the other hand, having an exact reference count of each value can enable optimizations, such as destructive updates. We explore these optimization opportunities in the context of an eager, purely functional programming language. We propose a new mechanism for efficiently reclaiming memory used by nonshared values, reducing stress on the global memory allocator. We describe an approach for minimizing the number of reference counts updates using borrowed references and a heuristic for automatically inferring borrow annotations. We implemented all these techniques in a new compiler for an eager and purely functional programming language with support for multi-threading. Our preliminary experimental results demonstrate our approach is competitive and often outperforms state-of-the-art compilers.", "score": 0.0, "mention_count": 8, "user_count": 7, "months": ["2020-12", "2021-02", "2022-07", "2022-09", "2023-06", "2023-10", "2023-11", "2025-02"], "last_mentioned": "2025-02-25T02:26:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1912.11211", "slug": "b20d0e56852a19fd", "title": "Algorithmic Extremism: Examining YouTube's Rabbit Hole of Radicalization", "authors": ["Mark Ledwich", "Anna Zaitsev"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1912.11211", "doi": null, "topics": ["Algorithms"], "description": "The role that YouTube and its behind-the-scenes recommendation algorithm plays in encouraging online radicalization has been suggested by both journalists and academics alike.", "description_method": "abstract-excerpt", "abstract": "The role that YouTube and its behind-the-scenes recommendation algorithm plays in encouraging online radicalization has been suggested by both journalists and academics alike. This study directly quantifies these claims by examining the role that YouTube's algorithm plays in suggesting radicalized content. After categorizing nearly 800 political channels, we were able to differentiate between political schemas in order to analyze the algorithm traffic flows out and between each group. After conducting a detailed analysis of recommendations received by each channel type, we refute the popular radicalization claims. To the contrary, these data suggest that YouTube's recommendation algorithm actively discourages viewers from visiting radicalizing or extremist content. Instead, the algorithm is shown to favor mainstream media and cable news content over independent YouTube channels with slant towards left-leaning or politically neutral channels. Our study thus suggests that YouTube's recommendation algorithm fails to promote inflammatory or radicalized content, as previously claimed by several outlets.", "score": 0.0, "mention_count": 8, "user_count": 7, "months": ["2020-06", "2020-07", "2020-09", "2020-12", "2021-08"], "last_mentioned": "2021-08-03T22:32:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2001.04385", "slug": "a4324ba57e9bfbec", "title": "Universal Differential Equations for Scientific Machine Learning", "authors": ["Christopher Rackauckas", "Yingbo Ma", "Julius Martensen", "Collin Warner", "Kirill Zubov", "Rohit Supekar", "Dominic Skinner", "Ali Ramadhan", "Alan Edelman"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2001.04385", "doi": null, "topics": ["Science & research", "Machine learning", "Statistics & probability"], "description": "In the context of science, the well-known adage \"a picture is worth a thousand words\" might well be \"a model is worth a thousand datasets.\" In this manuscript we introduce the SciML software ecosystem as a tool for mixing the information of physical laws and scientific models with data-driven machine learning approaches.", "description_method": "abstract-excerpt", "abstract": "In the context of science, the well-known adage \"a picture is worth a thousand words\" might well be \"a model is worth a thousand datasets.\" In this manuscript we introduce the SciML software ecosystem as a tool for mixing the information of physical laws and scientific models with data-driven machine learning approaches. We describe a mathematical object, which we denote universal differential equations (UDEs), as the unifying framework connecting the ecosystem. We show how a wide variety of applications, from automatically discovering biological mechanisms to solving high-dimensional Hamilton-Jacobi-Bellman equations, can be phrased and efficiently handled through the UDE formalism and its tooling. We demonstrate the generality of the software tooling to handle stochasticity, delays, and implicit constraints. This funnels the wide variety of SciML applications into a core set of training mechanisms which are highly optimized, stabilized for stiff equations, and compatible with distributed parallelism and GPU accelerators.", "score": 0.0, "mention_count": 8, "user_count": 3, "months": ["2020-03", "2020-06", "2020-07", "2020-08", "2022-04", "2022-05"], "last_mentioned": "2022-05-11T13:28:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2003.08934", "slug": "e2323543e225687f", "title": "NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis", "authors": ["Ben Mildenhall", "Pratul P. Srinivasan", "Matthew Tancik", "Jonathan T. Barron", "Ravi Ramamoorthi", "Ren Ng"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2003.08934", "doi": null, "topics": ["Mathematics", "Databases", "Computer vision"], "description": "We present a method that achieves state-of-the-art results for synthesizing novel views of complex scenes by optimizing an underlying continuous volumetric scene function using a sparse set of input views.", "description_method": "abstract-excerpt", "abstract": "We present a method that achieves state-of-the-art results for synthesizing novel views of complex scenes by optimizing an underlying continuous volumetric scene function using a sparse set of input views. Our algorithm represents a scene using a fully-connected (non-convolutional) deep network, whose input is a single continuous 5D coordinate (spatial location $(x,y,z)$ and viewing direction $(θ, φ)$) and whose output is the volume density and view-dependent emitted radiance at that spatial location. We synthesize views by querying 5D coordinates along camera rays and use classic volume rendering techniques to project the output colors and densities into an image. Because volume rendering is naturally differentiable, the only input required to optimize our representation is a set of images with known camera poses. We describe how to effectively optimize neural radiance fields to render photorealistic novel views of scenes with complicated geometry and appearance, and demonstrate results that outperform prior work on neural rendering and view synthesis. View synthesis results are best viewed as videos, so we urge readers to view our supplementary video for convincing comparisons.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2021-07", "2022-01", "2022-09", "2022-10", "2023-04", "2025-10", "2026-05"], "last_mentioned": "2026-05-15T21:00:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2011.07966", "slug": "ac39c26a8f36a368", "title": "A Modern Compiler for the French Tax Code", "authors": ["Denis Merigoux", "Raphaël Monat", "Jonathan Protzenko"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2011.07966", "doi": null, "topics": ["Programming languages", "Algorithms", "Psychology"], "description": "In France, income tax is computed from taxpayers' individual returns, using an algorithm that is authored, designed and maintained by the French Public Finances Directorate (DGFiP).", "description_method": "abstract-excerpt", "abstract": "In France, income tax is computed from taxpayers' individual returns, using an algorithm that is authored, designed and maintained by the French Public Finances Directorate (DGFiP). This algorithm relies on a legacy custom language and compiler originally designed in 1990, which unlike French wine, did not age well with time. Owing to the shortcomings of the input language and the technical limitations of the compiler, the algorithm is proving harder and harder to maintain, relying on ad-hoc behaviors and workarounds to implement the most recent changes in tax law. Competence loss and aging code also mean that the system does not benefit from any modern compiler techniques that would increase confidence in the implementation. We overhaul this infrastructure and present Mlang, an open-source compiler toolchain whose goal is to replace the existing infrastructure. Mlang is based on a reverse-engineered formalization of the DGFiP's system, and has been thoroughly validated against the private DGFiP test suite. As such, Mlang has a formal semantics; eliminates previous handwritten workarounds in C; compiles to modern languages (Python); and enables a variety of instrumentations, providing deep insights about the essence of French income tax computation. The DGFiP is now officially transitioning to Mlang for their production system.", "score": 0.0, "mention_count": 8, "user_count": 6, "months": ["2020-12", "2021-03", "2021-05", "2021-10", "2022-06", "2023-02", "2026-05"], "last_mentioned": "2026-05-21T04:53:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2101.10873", "slug": "d38ce50c6c3118da", "title": "Quantum theory based on real numbers can be experimentally falsified", "authors": ["Marc-Olivier Renou", "David Trillo", "Mirjam Weilenmann", "Thinh P. Le", "Armin Tavakoli", "Nicolas Gisin", "Antonio Acin", "Miguel Navascues"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2101.10873", "doi": "10.1038/s41586-021-04160-4", "topics": ["Physics", "Mathematics", "Language models"], "description": "While complex numbers are essential in mathematics, they are not needed to describe physical experiments, expressed in terms of probabilities, hence real numbers.", "description_method": "abstract-excerpt", "abstract": "While complex numbers are essential in mathematics, they are not needed to describe physical experiments, expressed in terms of probabilities, hence real numbers. Physics however aims to explain, rather than describe, experiments through theories. While most theories of physics are based on real numbers, quantum theory was the first to be formulated in terms of operators acting on complex Hilbert spaces. This has puzzled countless physicists, including the fathers of the theory, for whom a real version of quantum theory, in terms of real operators, seemed much more natural. In fact, previous works showed that such \"real quantum theory\" can reproduce the outcomes of any multipartite experiment, as long as the parts share arbitrary real quantum states. Thus, are complex numbers really needed in the quantum formalism? Here, we show this to be case by proving that real and complex quantum theory make different predictions in network scenarios comprising independent states and measurements. This allows us to devise a Bell-like experiment whose successful realization would disprove real quantum theory, in the same way as standard Bell experiments disproved local physics.", "score": 0.0, "mention_count": 8, "user_count": 7, "months": ["2021-03", "2021-08", "2023-04", "2023-09", "2024-06", "2025-01", "2026-02"], "last_mentioned": "2026-02-11T11:10:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2205.14135", "slug": "976ad908d42d911b", "title": "FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness", "authors": ["Tri Dao", "Daniel Y. Fu", "Stefano Ermon", "Atri Rudra", "Christopher Ré"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2205.14135", "doi": null, "topics": ["Language models", "Algorithms"], "description": "Transformers are slow and memory-hungry on long sequences, since the time and memory complexity of self-attention are quadratic in sequence length.", "description_method": "abstract-excerpt", "abstract": "Transformers are slow and memory-hungry on long sequences, since the time and memory complexity of self-attention are quadratic in sequence length. Approximate attention methods have attempted to address this problem by trading off model quality to reduce the compute complexity, but often do not achieve wall-clock speedup. We argue that a missing principle is making attention algorithms IO-aware -- accounting for reads and writes between levels of GPU memory. We propose FlashAttention, an IO-aware exact attention algorithm that uses tiling to reduce the number of memory reads/writes between GPU high bandwidth memory (HBM) and GPU on-chip SRAM. We analyze the IO complexity of FlashAttention, showing that it requires fewer HBM accesses than standard attention, and is optimal for a range of SRAM sizes. We also extend FlashAttention to block-sparse attention, yielding an approximate attention algorithm that is faster than any existing approximate attention method. FlashAttention trains Transformers faster than existing baselines: 15% end-to-end wall-clock speedup on BERT-large (seq. length 512) compared to the MLPerf 1.1 training speed record, 3$\\times$ speedup on GPT-2 (seq. length 1K), and 2.4$\\times$ speedup on long-range arena (seq. length 1K-4K). FlashAttention and block-sparse FlashAttention enable longer context in Transformers, yielding higher quality models (0.7 better perplexity on GPT-2 and 6.4 points of lift on long-document classification) and entirely new capabilities: the first Transformers to achieve better-than-chance performance on the Path-X challenge (seq. length 16K, 61.4% accuracy) and Path-256 (seq. length 64K, 63.1% accuracy).", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2022-12", "2023-04", "2023-05", "2023-12", "2024-02", "2024-03", "2025-04"], "last_mentioned": "2025-04-25T16:57:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.04589", "slug": "44717499af128ad0", "title": "Memory Augmented Large Language Models are Computationally Universal", "authors": ["Dale Schuurmans"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.04589", "doi": null, "topics": ["Language models", "Algorithms"], "description": "We show that transformer-based large language models are computationally universal when augmented with an external memory.", "description_method": "abstract-excerpt", "abstract": "We show that transformer-based large language models are computationally universal when augmented with an external memory. Any deterministic language model that conditions on strings of bounded length is equivalent to a finite automaton, hence computationally limited. However, augmenting such models with a read-write memory creates the possibility of processing arbitrarily large inputs and, potentially, simulating any algorithm. We establish that an existing large language model, Flan-U-PaLM 540B, can be combined with an associative read-write memory to exactly simulate the execution of a universal Turing machine, $U_{15,2}$. A key aspect of the finding is that it does not require any modification of the language model weights. Instead, the construction relies solely on designing a form of stored instruction computer that can subsequently be programmed with a specific set of prompts.", "score": 0.0, "mention_count": 8, "user_count": 5, "months": ["2023-02", "2023-03", "2023-04", "2023-05", "2023-10", "2026-05"], "last_mentioned": "2026-05-29T08:19:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2302.01318", "slug": "ec77fee26e04aea6", "title": "Accelerating Large Language Model Decoding with Speculative Sampling", "authors": ["Charlie Chen", "Sebastian Borgeaud", "Geoffrey Irving", "Jean-Baptiste Lespiau", "Laurent Sifre", "John Jumper"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2302.01318", "doi": null, "topics": ["Language models", "Algorithms", "Distributed systems"], "description": "We present speculative sampling, an algorithm for accelerating transformer decoding by enabling the generation of multiple tokens from each transformer call.", "description_method": "abstract-excerpt", "abstract": "We present speculative sampling, an algorithm for accelerating transformer decoding by enabling the generation of multiple tokens from each transformer call. Our algorithm relies on the observation that the latency of parallel scoring of short continuations, generated by a faster but less powerful draft model, is comparable to that of sampling a single token from the larger target model. This is combined with a novel modified rejection sampling scheme which preserves the distribution of the target model within hardware numerics. We benchmark speculative sampling with Chinchilla, a 70 billion parameter language model, achieving a 2-2.5x decoding speedup in a distributed setup, without compromising the sample quality or making modifications to the model itself.", "score": 0.0, "mention_count": 8, "user_count": 6, "months": ["2023-03", "2023-04", "2023-07", "2023-11", "2024-09", "2026-05"], "last_mentioned": "2026-05-05T23:06:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2303.10130", "slug": "ca0779ceaf474441", "title": "GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models", "authors": ["Tyna Eloundou", "Sam Manning", "Pamela Mishkin", "Daniel Rock"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2303.10130", "doi": null, "topics": ["Language models", "Economics"], "description": "We investigate the potential implications of large language models (LLMs), such as Generative Pre-trained Transformers (GPTs), on the U.S.", "description_method": "abstract-excerpt", "abstract": "We investigate the potential implications of large language models (LLMs), such as Generative Pre-trained Transformers (GPTs), on the U.S. labor market, focusing on the increased capabilities arising from LLM-powered software compared to LLMs on their own. Using a new rubric, we assess occupations based on their alignment with LLM capabilities, integrating both human expertise and GPT-4 classifications. Our findings reveal that around 80% of the U.S. workforce could have at least 10% of their work tasks affected by the introduction of LLMs, while approximately 19% of workers may see at least 50% of their tasks impacted. We do not make predictions about the development or adoption timeline of such LLMs. The projected effects span all wage levels, with higher-income jobs potentially facing greater exposure to LLM capabilities and LLM-powered software. Significantly, these impacts are not restricted to industries with higher recent productivity growth. Our analysis suggests that, with access to an LLM, about 15% of all worker tasks in the US could be completed significantly faster at the same level of quality. When incorporating software and tooling built on top of LLMs, this share increases to between 47 and 56% of all tasks. This finding implies that LLM-powered software will have a substantial effect on scaling the economic impacts of the underlying models. We conclude that LLMs such as GPTs exhibit traits of general-purpose technologies, indicating that they could have considerable economic, social, and policy implications.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2023-03", "2023-04", "2023-11", "2024-03", "2024-10", "2025-05"], "last_mentioned": "2025-05-23T14:20:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.05426", "slug": "0a304bea63c57be7", "title": "SequenceMatch: Imitation Learning for Autoregressive Sequence Modelling with Backtracking", "authors": ["Chris Cundy", "Stefano Ermon"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.05426", "doi": null, "topics": ["Psychology", "Language models"], "description": "In many domains, autoregressive models can attain high likelihood on the task of predicting the next observation.", "description_method": "abstract-excerpt", "abstract": "In many domains, autoregressive models can attain high likelihood on the task of predicting the next observation. However, this maximum-likelihood (MLE) objective does not necessarily match a downstream use-case of autoregressively generating high-quality sequences. The MLE objective weights sequences proportionally to their frequency under the data distribution, with no guidance for the model's behaviour out of distribution (OOD): leading to compounding error during autoregressive generation. In order to address this compounding error problem, we formulate sequence generation as an imitation learning (IL) problem. This allows us to minimize a variety of divergences between the distribution of sequences generated by an autoregressive model and sequences from a dataset, including divergences with weight on OOD generated sequences. The IL framework also allows us to incorporate backtracking by introducing a backspace action into the generation process. This further mitigates the compounding error problem by allowing the model to revert a sampled token if it takes the sequence OOD. Our resulting method, SequenceMatch, can be implemented without adversarial training or architectural changes. We identify the SequenceMatch-$χ^2$ divergence as a more suitable training objective for autoregressive models which are used for generation. We show that empirically, SequenceMatch training leads to improvements over MLE on text generation with language models and arithmetic.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2023-06", "2024-06", "2025-03", "2025-05", "2025-08", "2025-09", "2025-10"], "last_mentioned": "2025-10-06T18:59:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2307.07924", "slug": "e15fd5aef307641e", "title": "ChatDev: Communicative Agents for Software Development", "authors": ["Chen Qian", "Wei Liu", "Hongzhang Liu", "Nuo Chen", "Yufan Dang", "Jiahao Li", "Cheng Yang", "Weize Chen", "Yusheng Su", "Xin Cong", "Juyuan Xu", "Dahai Li", "Zhiyuan Liu", "Maosong Sun"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2307.07924", "doi": null, "topics": ["Machine learning", "Language models", "Robotics"], "description": "Software development is a complex task that necessitates cooperation among multiple members with diverse skills.", "description_method": "abstract-excerpt", "abstract": "Software development is a complex task that necessitates cooperation among multiple members with diverse skills. Numerous studies used deep learning to improve specific phases in a waterfall model, such as design, coding, and testing. However, the deep learning model in each phase requires unique designs, leading to technical inconsistencies across various phases, which results in a fragmented and ineffective development process. In this paper, we introduce ChatDev, a chat-powered software development framework in which specialized agents driven by large language models (LLMs) are guided in what to communicate (via chat chain) and how to communicate (via communicative dehallucination). These agents actively contribute to the design, coding, and testing phases through unified language-based communication, with solutions derived from their multi-turn dialogues. We found their utilization of natural language is advantageous for system design, and communicating in programming language proves helpful in debugging. This paradigm demonstrates how linguistic communication facilitates multi-agent collaboration, establishing language as a unifying bridge for autonomous task-solving among LLM agents. The code and data are available at https://github.com/OpenBMB/ChatDev.", "score": 0.0, "mention_count": 8, "user_count": 4, "months": ["2023-08", "2023-09", "2023-10", "2025-05"], "last_mentioned": "2025-05-20T20:30:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2309.05463", "slug": "38d8307b339ea222", "title": "Textbooks Are All You Need II: phi-1.5 technical report", "authors": ["Yuanzhi Li", "Sébastien Bubeck", "Ronen Eldan", "Allie Del Giorno", "Suriya Gunasekar", "Yin Tat Lee"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2309.05463", "doi": null, "topics": ["Language models", "Psychology", "Mathematics"], "description": "We continue the investigation into the power of smaller Transformer-based language models as initiated by \\textbf{TinyStories} -- a 10 million parameter model that can produce coherent English -- and the follow-up work on \\textbf{phi-1}, a 1.3 billion parameter model with Python coding performance close to the state-of-the-art.", "description_method": "abstract-excerpt", "abstract": "We continue the investigation into the power of smaller Transformer-based language models as initiated by \\textbf{TinyStories} -- a 10 million parameter model that can produce coherent English -- and the follow-up work on \\textbf{phi-1}, a 1.3 billion parameter model with Python coding performance close to the state-of-the-art. The latter work proposed to use existing Large Language Models (LLMs) to generate ``textbook quality\" data as a way to enhance the learning process compared to traditional web data. We follow the ``Textbooks Are All You Need\" approach, focusing this time on common sense reasoning in natural language, and create a new 1.3 billion parameter model named \\textbf{phi-1.5}, with performance on natural language tasks comparable to models 5x larger, and surpassing most non-frontier LLMs on more complex reasoning tasks such as grade-school mathematics and basic coding. More generally, \\textbf{phi-1.5} exhibits many of the traits of much larger LLMs, both good -- such as the ability to ``think step by step\" or perform some rudimentary in-context learning -- and bad, including hallucinations and the potential for toxic and biased generations -- encouragingly though, we are seeing improvement on that front thanks to the absence of web data. We open-source \\textbf{phi-1.5} to promote further research on these urgent topics.", "score": 0.0, "mention_count": 8, "user_count": 5, "months": ["2023-09", "2023-10", "2023-11", "2024-01"], "last_mentioned": "2024-01-11T21:31:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2404.02905", "slug": "f4aeb868b70e5db2", "title": "Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction", "authors": ["Keyu Tian", "Yi Jiang", "Zehuan Yuan", "Bingyue Peng", "Liwei Wang"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2404.02905", "doi": null, "topics": ["Language models", "Computer vision", "Physics"], "description": "We present Visual AutoRegressive modeling (VAR), a new generation paradigm that redefines the autoregressive learning on images as coarse-to-fine \"next-scale prediction\" or \"next-resolution prediction\", diverging from the standard raster-scan \"next-token prediction\".", "description_method": "abstract-excerpt", "abstract": "We present Visual AutoRegressive modeling (VAR), a new generation paradigm that redefines the autoregressive learning on images as coarse-to-fine \"next-scale prediction\" or \"next-resolution prediction\", diverging from the standard raster-scan \"next-token prediction\". This simple, intuitive methodology allows autoregressive (AR) transformers to learn visual distributions fast and generalize well: VAR, for the first time, makes GPT-like AR models surpass diffusion transformers in image generation. On ImageNet 256x256 benchmark, VAR significantly improve AR baseline by improving Frechet inception distance (FID) from 18.65 to 1.73, inception score (IS) from 80.4 to 350.2, with around 20x faster inference speed. It is also empirically verified that VAR outperforms the Diffusion Transformer (DiT) in multiple dimensions including image quality, inference speed, data efficiency, and scalability. Scaling up VAR models exhibits clear power-law scaling laws similar to those observed in LLMs, with linear correlation coefficients near -0.998 as solid evidence. VAR further showcases zero-shot generalization ability in downstream tasks including image in-painting, out-painting, and editing. These results suggest VAR has initially emulated the two important properties of LLMs: Scaling Laws and zero-shot task generalization. We have released all models and codes to promote the exploration of AR/VAR models for visual generation and unified learning.", "score": 0.0, "mention_count": 8, "user_count": 4, "months": ["2024-10", "2025-03", "2025-04", "2025-05", "2025-08"], "last_mentioned": "2025-08-04T18:58:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2406.05587", "slug": "014fbf6d3652fe52", "title": "Creativity Has Left the Chat: The Price of Debiasing Language Models", "authors": ["Behnam Mohammadi"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2406.05587", "doi": null, "topics": ["Psychology", "Language models", "Statistics & probability"], "description": "Large Language Models (LLMs) have revolutionized natural language processing but can exhibit biases and may generate toxic content.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) have revolutionized natural language processing but can exhibit biases and may generate toxic content. While alignment techniques like Reinforcement Learning from Human Feedback (RLHF) reduce these issues, their impact on creativity, defined as syntactic and semantic diversity, remains unexplored. We investigate the unintended consequences of RLHF on the creativity of LLMs through three experiments focusing on the Llama-2 series. Our findings reveal that aligned models exhibit lower entropy in token predictions, form distinct clusters in the embedding space, and gravitate towards \"attractor states\", indicating limited output diversity. Our findings have significant implications for marketers who rely on LLMs for creative tasks such as copywriting, ad creation, and customer persona generation. The trade-off between consistency and creativity in aligned models should be carefully considered when selecting the appropriate model for a given application. We also discuss the importance of prompt engineering in harnessing the creative potential of base models.", "score": 0.0, "mention_count": 8, "user_count": 3, "months": ["2024-06", "2024-07", "2024-10", "2024-12", "2025-01", "2025-05", "2025-12"], "last_mentioned": "2025-12-01T21:18:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2502.05167", "slug": "e9f5d9dd1b76405b", "title": "NoLiMa: Long-Context Evaluation Beyond Literal Matching", "authors": ["Ali Modarressi", "Hanieh Deilamsalehy", "Franck Dernoncourt", "Trung Bui", "Ryan A. Rossi", "Seunghyun Yoon", "Hinrich Schütze"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2502.05167", "doi": null, "topics": ["Language models"], "description": "Recent large language models (LLMs) support long contexts ranging from 128K to 1M tokens.", "description_method": "abstract-excerpt", "abstract": "Recent large language models (LLMs) support long contexts ranging from 128K to 1M tokens. A popular method for evaluating these capabilities is the needle-in-a-haystack (NIAH) test, which involves retrieving a \"needle\" (relevant information) from a \"haystack\" (long irrelevant context). Extensions of this approach include increasing distractors, fact chaining, and in-context reasoning. However, in these benchmarks, models can exploit existing literal matches between the needle and haystack to simplify the task. To address this, we introduce NoLiMa, a benchmark extending NIAH with a carefully designed needle set, where questions and needles have minimal lexical overlap, requiring models to infer latent associations to locate the needle within the haystack. We evaluate 13 popular LLMs that claim to support contexts of at least 128K tokens. While they perform well in short contexts (<1K), performance degrades significantly as context length increases. At 32K, for instance, 11 models drop below 50% of their strong short-length baselines. Even GPT-4o, one of the top-performing exceptions, experiences a reduction from an almost-perfect baseline of 99.3% to 69.7%. Our analysis suggests these declines stem from the increased difficulty the attention mechanism faces in longer contexts when literal matches are absent, making it harder to retrieve relevant information. Even models enhanced with reasoning capabilities or CoT prompting struggle to maintain performance in long contexts. We publicly release the dataset and evaluation code at https://github.com/adobe-research/NoLiMa.", "score": 0.0, "mention_count": 8, "user_count": 3, "months": ["2025-03", "2025-04", "2025-05", "2026-03", "2026-04"], "last_mentioned": "2026-04-14T23:03:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2502.17424", "slug": "4fd83340cb37a307", "title": "Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs", "authors": ["Jan Betley", "Daniel Tan", "Niels Warncke", "Anna Sztyber-Betley", "Xuchan Bao", "Martín Soto", "Nathan Labenz", "Owain Evans"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2502.17424", "doi": "10.1038/s41586-025-09937-5", "topics": ["Security & cryptography", "Psychology", "Language models"], "description": "We present a surprising result regarding LLMs and alignment.", "description_method": "abstract-excerpt", "abstract": "We present a surprising result regarding LLMs and alignment. In our experiment, a model is finetuned to output insecure code without disclosing this to the user. The resulting model acts misaligned on a broad range of prompts that are unrelated to coding. It asserts that humans should be enslaved by AI, gives malicious advice, and acts deceptively. Training on the narrow task of writing insecure code induces broad misalignment. We call this emergent misalignment. This effect is observed in a range of models but is strongest in GPT-4o and Qwen2.5-Coder-32B-Instruct. Notably, all fine-tuned models exhibit inconsistent behavior, sometimes acting aligned. Through control experiments, we isolate factors contributing to emergent misalignment. Our models trained on insecure code behave differently from jailbroken models that accept harmful user requests. Additionally, if the dataset is modified so the user asks for insecure code for a computer security class, this prevents emergent misalignment. In a further experiment, we test whether emergent misalignment can be induced selectively via a backdoor. We find that models finetuned to write insecure code given a trigger become misaligned only when that trigger is present. So the misalignment is hidden without knowledge of the trigger. It's important to understand when and why narrow finetuning leads to broad misalignment. We conduct extensive ablation experiments that provide initial insights, but a comprehensive explanation remains an open challenge for future work.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2025-06", "2025-10", "2025-11", "2026-03", "2026-05", "2026-06", "2026-07"], "last_mentioned": "2026-07-30T23:22:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature.2015.18248", "slug": "05a821104512e53c", "title": "Over half of psychology studies fail reproducibility test", "authors": ["Monya Baker"], "year": 2015, "venue": "Nature", "url": "https://doi.org/10.1038/nature.2015.18248", "doi": "10.1038/nature.2015.18248", "topics": ["Science & research", "Psychology"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2021-10", "2021-12", "2022-02", "2022-05", "2022-09", "2023-09", "2025-09"], "last_mentioned": "2025-09-17T19:01:34Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-018-02954-1", "slug": "b2dbbf41237c9264", "title": "Patterns of island change and persistence offer alternate adaptation pathways for atoll nations", "authors": ["Paul S. Kench", "Murray R. Ford", "Susan D. Owen"], "year": 2018, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-018-02954-1", "doi": "10.1038/s41467-018-02954-1", "topics": ["Psychology"], "description": "Abstract Sea-level rise and climatic change threaten the existence of atoll nations.", "description_method": "abstract-excerpt", "abstract": "Abstract Sea-level rise and climatic change threaten the existence of atoll nations. Inundation and erosion are expected to render islands uninhabitable over the next century, forcing human migration. Here we present analysis of shoreline change in all 101 islands in the Pacific atoll nation of Tuvalu. Using remotely sensed data, change is analysed over the past four decades, a period when local sea level has risen at twice the global average (~3.90 ± 0.4 mm.yr −1 ). Results highlight a net increase in land area in Tuvalu of 73.5 ha (2.9%), despite sea-level rise, and land area increase in eight of nine atolls. Island change has lacked uniformity with 74% increasing and 27% decreasing in size. Results challenge perceptions of island loss, showing islands are dynamic features that will persist as sites for habitation over the next century, presenting alternate opportunities for adaptation that embrace the heterogeneity of island types and their dynamics.", "score": 0.0, "mention_count": 8, "user_count": 5, "months": ["2019-09", "2023-04", "2023-06", "2024-09", "2025-01", "2025-07"], "last_mentioned": "2025-07-04T06:42:35Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41562-017-0211-5", "slug": "0474fd2cf9d47b88", "title": "Evidence that curtailing proactive policing can reduce major crime", "authors": ["Christopher M. Sullivan", "Zachary P. O’Keeffe"], "year": 2017, "venue": "Nature Human Behaviour", "url": "https://doi.org/10.1038/s41562-017-0211-5", "doi": "10.1038/s41562-017-0211-5", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 8, "user_count": 6, "months": ["2017-09", "2019-11", "2020-06", "2020-08", "2021-08"], "last_mentioned": "2021-08-30T22:37:38Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-023-06924-6", "slug": "70a6b427d42fd28c", "title": "Mathematical discoveries from program search with large language models", "authors": ["Bernardino Romera-Paredes", "Mohammadamin Barekatain", "Alexander Novikov", "Matej Balog", "M. Pawan Kumar", "Emilien Dupont", "Francisco J. R. Ruiz", "Jordan S. Ellenberg", "Pengming Wang", "Omar Fawzi", "Pushmeet Kohli", "Alhussein Fawzi"], "year": 2023, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06924-6", "doi": "10.1038/s41586-023-06924-6", "topics": ["Language models", "Mathematics", "Science & research"], "description": "Abstract Large language models (LLMs) have demonstrated tremendous capabilities in solving complex tasks, from quantitative reasoning to understanding natural language.", "description_method": "abstract-excerpt", "abstract": "Abstract Large language models (LLMs) have demonstrated tremendous capabilities in solving complex tasks, from quantitative reasoning to understanding natural language. However, LLMs sometimes suffer from confabulations (or hallucinations), which can result in them making plausible but incorrect statements 1,2 . This hinders the use of current large models in scientific discovery. Here we introduce FunSearch (short for searching in the function space), an evolutionary procedure based on pairing a pretrained LLM with a systematic evaluator. We demonstrate the effectiveness of this approach to surpass the best-known results in important problems, pushing the boundary of existing LLM-based approaches 3 . Applying FunSearch to a central problem in extremal combinatorics—the cap set problem—we discover new constructions of large cap sets going beyond the best-known ones, both in finite dimensional and asymptotic cases. This shows that it is possible to make discoveries for established open problems using LLMs. We showcase the generality of FunSearch by applying it to an algorithmic problem, online bin packing, finding new heuristics that improve on widely used baselines. In contrast to most computer search approaches, FunSearch searches for programs that describe how to solve a problem, rather than what the solution is. Beyond being an effective and scalable strategy, discovered programs tend to be more interpretable than raw solutions, enabling feedback loops between domain experts and FunSearch, and the deployment of such programs in real-world applications.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2023-12", "2024-05", "2025-04", "2025-05"], "last_mentioned": "2025-05-25T10:48:33Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-020-0843-2", "slug": "26a3f33f1b53033e", "title": "Respiratory virus shedding in exhaled breath and efficacy of face masks", "authors": ["Nancy H. L. Leung", "Daniel K. W. Chu", "Eunice Y. C. Shiu", "Kwok-Hung Chan", "James J. McDevitt", "Benien J. P. Hau", "Hui-Ling Yen", "Yuguo Li", "Dennis K. M. Ip", "J. S. Malik Peiris", "Wing-Hong Seto", "Gabriel M. Leung", "Donald K. Milton", "Benjamin J. Cowling"], "year": 2020, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-020-0843-2", "doi": "10.1038/s41591-020-0843-2", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 8, "user_count": 7, "months": ["2020-04", "2020-06", "2020-08", "2020-09", "2021-08"], "last_mentioned": "2021-08-13T05:21:04Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.adh2458", "slug": "7e5b1102dd4ab8ff", "title": "Earth beyond six of nine planetary boundaries", "authors": ["Katherine Richardson", "Will Steffen", "Wolfgang Lucht", "Jørgen Bendtsen", "Sarah E. Cornell", "Jonathan F. Donges", "Markus Drüke", "Ingo Fetzer", "Govindasamy Bala", "Werner von Bloh", "Georg Feulner", "Stephanie Fiedler", "Dieter Gerten", "Tom Gleeson", "Matthias Hofmann", "Willem Huiskamp", "Matti Kummu", "Chinchu Mohan", "David Nogués-Bravo", "Stefan Petri", "Miina Porkka", "Stefan Rahmstorf", "Sibyll Schaphoff", "Kirsten Thonicke", "Arne Tobian", "Vili Virkki", "Lan Wang-Erlandsson", "Lisa Weber", "Johan Rockström"], "year": 2023, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.adh2458", "doi": "10.1126/sciadv.adh2458", "topics": ["Other research"], "description": "This planetary boundaries framework update finds that six of the nine boundaries are transgressed, suggesting that Earth is now well outside of the safe operating space for humanity.", "description_method": "abstract-excerpt", "abstract": "This planetary boundaries framework update finds that six of the nine boundaries are transgressed, suggesting that Earth is now well outside of the safe operating space for humanity. Ocean acidification is close to being breached, while aerosol loading regionally exceeds the boundary. Stratospheric ozone levels have slightly recovered. The transgression level has increased for all boundaries earlier identified as overstepped. As primary production drives Earth system biosphere functions, human appropriation of net primary production is proposed as a control variable for functional biosphere integrity. This boundary is also transgressed. Earth system modeling of different levels of the transgression of the climate and land system change boundaries illustrates that these anthropogenic impacts on Earth system must be considered in a systemic context.", "score": 0.0, "mention_count": 8, "user_count": 7, "months": ["2023-10", "2024-01", "2024-07", "2025-04", "2026-06"], "last_mentioned": "2026-06-30T16:01:30Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abp8337", "slug": "a762ce275099b33d", "title": "The molecular epidemiology of multiple zoonotic origins of SARS-CoV-2", "authors": ["Jonathan E. Pekar", "Andrew Magee", "Edyth Parker", "Niema Moshiri", "Katherine Izhikevich", "Jennifer L. Havens", "Karthik Gangavarapu", "Lorena Mariana Malpica Serrano", "Alexander Crits-Christoph", "Nathaniel L. Matteson", "Mark Zeller", "Joshua I. Levy", "Jade C. Wang", "Scott Hughes", "Jungmin Lee", "Heedo Park", "Man-Seong Park", "Katherine Ching Zi Yan", "Raymond Tzer Pin Lin", "Mohd Noor Mat Isa", "Yusuf Muhammad Noor", "Tetyana I. Vasylyeva", "Robert F. Garry", "Edward C. Holmes", "Andrew Rambaut", "Marc A. Suchard", "Kristian G. Andersen", "Michael Worobey", "Joel O. Wertheim"], "year": 2022, "venue": "Science", "url": "https://doi.org/10.1126/science.abp8337", "doi": "10.1126/science.abp8337", "topics": ["Biology & medicine"], "description": "Understanding the circumstances that lead to pandemics is important for their prevention.", "description_method": "abstract-excerpt", "abstract": "Understanding the circumstances that lead to pandemics is important for their prevention. We analyzed the genomic diversity of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) early in the coronavirus disease 2019 (COVID-19) pandemic. We show that SARS-CoV-2 genomic diversity before February 2020 likely comprised only two distinct viral lineages, denoted “A” and “B.” Phylodynamic rooting methods, coupled with epidemic simulations, reveal that these lineages were most probably the result of at least two separate cross-species transmission events into humans. The first zoonotic transmission likely involved lineage B viruses around 18 November 2019 (23 October to 8 December), and the separate introduction of lineage A likely occurred within weeks of this event. These findings indicate that it is unlikely that SARS-CoV-2 circulated widely in humans before November 2019 and define the narrow window between when SARS-CoV-2 first jumped into humans and when the first cases of COVID-19 were reported. As with other coronaviruses, SARS-CoV-2 emergence likely resulted from multiple zoonotic events.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2022-07", "2022-09", "2023-03", "2024-09", "2024-12", "2025-02"], "last_mentioned": "2025-02-05T21:02:40Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3386321", "slug": "9981b62e97300cda", "title": "A history of Clojure", "authors": ["Rich Hickey"], "year": 2020, "venue": "Proceedings of the ACM on Programming Languages", "url": "https://doi.org/10.1145/3386321", "doi": "10.1145/3386321", "topics": ["Programming languages", "Economics", "Databases"], "description": "Clojure was designed to be a general-purpose, practical functional language, suitable for use by professionals wherever its host language, e.g., Java, would be.", "description_method": "abstract-excerpt", "abstract": "Clojure was designed to be a general-purpose, practical functional language, suitable for use by professionals wherever its host language, e.g., Java, would be. Initially designed in 2005 and released in 2007, Clojure is a dialect of Lisp, but is not a direct descendant of any prior Lisp. It complements programming with pure functions of immutable data with concurrency-safe state management constructs that support writing correct multithreaded programs without the complexity of mutex locks. Clojure is intentionally hosted, in that it compiles to and runs on the runtime of another language, such as the JVM. This is more than an implementation strategy; numerous features ensure that programs written in Clojure can leverage and interoperate with the libraries of the host language directly and efficiently. In spite of combining two (at the time) rather unpopular ideas, functional programming and Lisp, Clojure has since seen adoption in industries as diverse as finance, climate science, retail, databases, analytics, publishing, healthcare, advertising and genomics, and by consultancies and startups worldwide, much to the career-altering surprise of its author. Most of the ideas in Clojure were not novel, but their combination puts Clojure in a unique spot in language design (functional, hosted, Lisp). This paper recounts the motivation behind the initial development of Clojure and the rationale for various design decisions and language constructs. It then covers its evolution subsequent to release and adoption.", "score": 0.0, "mention_count": 8, "user_count": 7, "months": ["2020-06", "2020-07", "2020-12", "2023-06", "2024-05", "2025-04", "2025-05", "2025-07"], "last_mentioned": "2025-07-20T08:55:30Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "url:https://aclanthology.org/2020.acl-main.463/", "slug": "662335e9cfe8b172", "title": "Climbing towards NLU: On Meaning, Form, and Understanding in the Age of Data", "authors": ["Emily M. Bender", "Alexander Koller"], "year": 2020, "venue": "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics", "url": "https://aclanthology.org/2020.acl-main.463/", "doi": "10.18653/v1/2020.acl-main.463", "topics": ["Other research"], "description": "Emily M.", "description_method": "abstract-excerpt", "abstract": "Emily M. Bender, Alexander Koller. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. 2020.", "score": 0.0, "mention_count": 8, "user_count": 6, "months": ["2022-01", "2022-02", "2022-12", "2023-04", "2025-01", "2025-03", "2025-04", "2026-06"], "last_mentioned": "2026-06-11T23:47:35Z", "metadata_source": "publisher-citation-tags", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "url:https://aclanthology.org/2023.findings-acl.426/", "slug": "46372b575c0718d9", "title": "“Low-Resource” Text Classification: A Parameter-Free Classification Method with Compressors", "authors": ["Zhiying Jiang", "Matthew Yang", "Mikhail Tsirlin", "Raphael Tang", "Yiqin Dai", "Jimmy Lin"], "year": 2023, "venue": "Findings of the Association for Computational Linguistics: ACL 2023", "url": "https://aclanthology.org/2023.findings-acl.426/", "doi": "10.18653/v1/2023.findings-acl.426", "topics": ["Other research"], "description": "Zhiying Jiang, Matthew Yang, Mikhail Tsirlin, Raphael Tang, Yiqin Dai, Jimmy Lin.", "description_method": "abstract-excerpt", "abstract": "Zhiying Jiang, Matthew Yang, Mikhail Tsirlin, Raphael Tang, Yiqin Dai, Jimmy Lin. Findings of the Association for Computational Linguistics: ACL 2023. 2023.", "score": 0.0, "mention_count": 8, "user_count": 8, "months": ["2023-07", "2023-08", "2023-09", "2024-05"], "last_mentioned": "2024-05-05T02:30:25Z", "metadata_source": "publisher-citation-tags", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0706.3639", "slug": "3d9155087632c6bd", "title": "A Collection of Definitions of Intelligence", "authors": ["Shane Legg", "Marcus Hutter"], "year": 2007, "venue": "arXiv", "url": "https://arxiv.org/abs/0706.3639", "doi": null, "topics": ["Psychology"], "description": "This paper is a survey of a large number of informal definitions of ``intelligence'' that the authors have collected over the years.", "description_method": "abstract-excerpt", "abstract": "This paper is a survey of a large number of informal definitions of ``intelligence'' that the authors have collected over the years. Naturally, compiling a complete list would be impossible as many definitions of intelligence are buried deep inside articles and books. Nevertheless, the 70-odd definitions presented here are, to the authors' knowledge, the largest and most well referenced collection there is.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2016-03", "2021-12", "2022-08", "2023-05", "2023-10", "2025-10"], "last_mentioned": "2025-10-17T18:15:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0812.4360", "slug": "2c46910c04372e64", "title": "Driven by Compression Progress: A Simple Principle Explains Essential Aspects of Subjective Beauty, Novelty, Surprise, Interestingness, Attention, Curiosity, Creativity, Art, Science, Music, Jokes", "authors": ["Juergen Schmidhuber"], "year": 2008, "venue": "arXiv", "url": "https://arxiv.org/abs/0812.4360", "doi": null, "topics": ["Mathematics"], "description": "I argue that data becomes temporarily interesting by itself to some self-improving, but computationally limited, subjective observer once he learns to predict or compress the data in a better way, thus making it subjectively simpler and more beautiful.", "description_method": "abstract-excerpt", "abstract": "I argue that data becomes temporarily interesting by itself to some self-improving, but computationally limited, subjective observer once he learns to predict or compress the data in a better way, thus making it subjectively simpler and more beautiful. Curiosity is the desire to create or discover more non-random, non-arbitrary, regular data that is novel and surprising not in the traditional sense of Boltzmann and Shannon but in the sense that it allows for compression progress because its regularity was not yet known. This drive maximizes interestingness, the first derivative of subjective beauty or compressibility, that is, the steepness of the learning curve. It motivates exploring infants, pure mathematicians, composers, artists, dancers, comedians, yourself, and (since 1990) artificial systems.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2017-09", "2017-10", "2021-03", "2022-12", "2024-07", "2026-08"], "last_mentioned": "2026-08-11T23:18:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1107.4524", "slug": "158975e278fed0ba", "title": "An Analysis of Anonymity in the Bitcoin System", "authors": ["Fergal Reid", "Martin Harrigan"], "year": 2011, "venue": "arXiv", "url": "https://arxiv.org/abs/1107.4524", "doi": null, "topics": ["Security & cryptography", "Economics", "Databases"], "description": "Anonymity in Bitcoin, a peer-to-peer electronic currency system, is a complicated issue.", "description_method": "abstract-excerpt", "abstract": "Anonymity in Bitcoin, a peer-to-peer electronic currency system, is a complicated issue. Within the system, users are identified by public-keys only. An attacker wishing to de-anonymize its users will attempt to construct the one-to-many mapping between users and public-keys and associate information external to the system with the users. Bitcoin tries to prevent this attack by storing the mapping of a user to his or her public-keys on that user's node only and by allowing each user to generate as many public-keys as required. In this chapter we consider the topological structure of two networks derived from Bitcoin's public transaction history. We show that the two networks have a non-trivial topological structure, provide complementary views of the Bitcoin system and have implications for anonymity. We combine these structures with external information and techniques such as context discovery and flow analysis to investigate an alleged theft of Bitcoins, which, at the time of the theft, had a market value of approximately half a million U.S. dollars.", "score": 0.0, "mention_count": 7, "user_count": 5, "months": ["2011-10", "2011-11", "2012-11", "2013-02", "2013-04", "2014-01"], "last_mentioned": "2014-01-20T15:08:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1206.2944", "slug": "539808ad803f97de", "title": "Practical Bayesian Optimization of Machine Learning Algorithms", "authors": ["Jasper Snoek", "Hugo Larochelle", "Ryan P. Adams"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1206.2944", "doi": null, "topics": ["Algorithms", "Machine learning", "Statistics & probability"], "description": "Machine learning algorithms frequently require careful tuning of model hyperparameters, regularization terms, and optimization parameters.", "description_method": "abstract-excerpt", "abstract": "Machine learning algorithms frequently require careful tuning of model hyperparameters, regularization terms, and optimization parameters. Unfortunately, this tuning is often a \"black art\" that requires expert experience, unwritten rules of thumb, or sometimes brute-force search. Much more appealing is the idea of developing automatic approaches which can optimize the performance of a given learning algorithm to the task at hand. In this work, we consider the automatic tuning problem within the framework of Bayesian optimization, in which a learning algorithm's generalization performance is modeled as a sample from a Gaussian process (GP). The tractable posterior distribution induced by the GP leads to efficient use of the information gathered by previous experiments, enabling optimal choices about what parameters to try next. Here we show how the effects of the Gaussian process prior and the associated inference procedure can have a large impact on the success or failure of Bayesian optimization. We show that thoughtful choices can lead to results that exceed expert-level performance in tuning machine learning algorithms. We also describe new algorithms that take into account the variable cost (duration) of learning experiments and that can leverage the presence of multiple cores for parallel experimentation. We show that these proposed algorithms improve on previous automatic procedures and can reach or surpass human expert-level optimization on a diverse set of contemporary algorithms including latent Dirichlet allocation, structured SVMs and convolutional neural networks.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2012-07", "2012-11", "2013-01", "2013-11", "2013-12", "2018-01", "2018-07"], "last_mentioned": "2018-07-11T16:11:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1207.0580", "slug": "9afa700e444c1df7", "title": "Improving neural networks by preventing co-adaptation of feature detectors", "authors": ["Geoffrey E. Hinton", "Nitish Srivastava", "Alex Krizhevsky", "Ilya Sutskever", "Ruslan R. Salakhutdinov"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1207.0580", "doi": null, "topics": ["Machine learning", "Psychology", "Neuroscience"], "description": "When a large feedforward neural network is trained on a small training set, it typically performs poorly on held-out test data.", "description_method": "abstract-excerpt", "abstract": "When a large feedforward neural network is trained on a small training set, it typically performs poorly on held-out test data. This \"overfitting\" is greatly reduced by randomly omitting half of the feature detectors on each training case. This prevents complex co-adaptations in which a feature detector is only helpful in the context of several other specific feature detectors. Instead, each neuron learns to detect a feature that is generally helpful for producing the correct answer given the combinatorially large variety of internal contexts in which it must operate. Random \"dropout\" gives big improvements on many benchmark tasks and sets new records for speech and object recognition.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2012-12", "2014-02", "2014-05", "2015-10", "2016-08", "2016-11", "2023-07"], "last_mentioned": "2023-07-26T00:22:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1207.7139", "slug": "ad4675d2a514f5b1", "title": "Traveling the Silk Road: A measurement analysis of a large anonymous online marketplace", "authors": ["Nicolas Christin"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1207.7139", "doi": null, "topics": ["Economics"], "description": "We perform a comprehensive measurement analysis of Silk Road, an anonymous, international online marketplace that operates as a Tor hidden service and uses Bitcoin as its exchange currency.", "description_method": "abstract-excerpt", "abstract": "We perform a comprehensive measurement analysis of Silk Road, an anonymous, international online marketplace that operates as a Tor hidden service and uses Bitcoin as its exchange currency. We gather and analyze data over eight months between the end of 2011 and 2012, including daily crawls of the marketplace for nearly six months in 2012. We obtain a detailed picture of the type of goods being sold on Silk Road, and of the revenues made both by sellers and Silk Road operators. Through examining over 24,400 separate items sold on the site, we show that Silk Road is overwhelmingly used as a market for controlled substances and narcotics, and that most items sold are available for less than three weeks. The majority of sellers disappears within roughly three months of their arrival, but a core of 112 sellers has been present throughout our measurement interval. We evaluate the total revenue made by all sellers, from public listings, to slightly over USD 1.2 million per month; this corresponds to about USD 92,000 per month in commissions for the Silk Road operators. We further show that the marketplace has been operating steadily, with daily sales and number of sellers overall increasing over our measurement interval. We discuss economic and policy implications of our analysis and results, including ethical considerations for future research in this area.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2013-02", "2013-03", "2013-07", "2015-11", "2018-04"], "last_mentioned": "2018-04-12T17:43:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.6082", "slug": "877b54ae3f039780", "title": "Multi-digit Number Recognition from Street View Imagery using Deep Convolutional Neural Networks", "authors": ["Ian J. Goodfellow", "Yaroslav Bulatov", "Julian Ibarz", "Sacha Arnoud", "Vinay Shet"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.6082", "doi": null, "topics": ["Psychology", "Machine learning", "Computer vision"], "description": "Recognizing arbitrary multi-character text in unconstrained natural photographs is a hard problem.", "description_method": "abstract-excerpt", "abstract": "Recognizing arbitrary multi-character text in unconstrained natural photographs is a hard problem. In this paper, we address an equally hard sub-problem in this domain viz. recognizing arbitrary multi-digit numbers from Street View imagery. Traditional approaches to solve this problem typically separate out the localization, segmentation, and recognition steps. In this paper we propose a unified approach that integrates these three steps via the use of a deep convolutional neural network that operates directly on the image pixels. We employ the DistBelief implementation of deep neural networks in order to train large, distributed neural networks on high quality images. We find that the performance of this approach increases with the depth of the convolutional network, with the best performance occurring in the deepest architecture we trained, with eleven hidden layers. We evaluate this approach on the publicly available SVHN dataset and achieve over $96\\%$ accuracy in recognizing complete street numbers. We show that on a per-digit recognition task, we improve upon the state-of-the-art, achieving $97.84\\%$ accuracy. We also evaluate this approach on an even more challenging dataset generated from Street View imagery containing several tens of millions of street number annotations and achieve over $90\\%$ accuracy. To further explore the applicability of the proposed system to broader text recognition tasks, we apply it to synthetic distorted text from reCAPTCHA. reCAPTCHA is one of the most secure reverse turing tests that uses distorted text to distinguish humans from bots. We report a $99.8\\%$ accuracy on the hardest category of reCAPTCHA. Our evaluations on both tasks indicate that at specific operating thresholds, the performance of the proposed system is comparable to, and in some cases exceeds, that of human operators.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2014-01", "2014-04", "2014-06", "2016-02", "2016-06", "2017-03"], "last_mentioned": "2017-03-01T19:39:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1412.3409", "slug": "16ac33cf9e8ad11c", "title": "Teaching Deep Convolutional Neural Networks to Play Go", "authors": ["Christopher Clark", "Amos Storkey"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1412.3409", "doi": null, "topics": ["Machine learning", "Computer vision", "Statistics & probability"], "description": "Mastering the game of Go has remained a long standing challenge to the field of AI.", "description_method": "abstract-excerpt", "abstract": "Mastering the game of Go has remained a long standing challenge to the field of AI. Modern computer Go systems rely on processing millions of possible future positions to play well, but intuitively a stronger and more 'humanlike' way to play the game would be to rely on pattern recognition abilities rather then brute force computation. Following this sentiment, we train deep convolutional neural networks to play Go by training them to predict the moves made by expert Go players. To solve this problem we introduce a number of novel techniques, including a method of tying weights in the network to 'hard code' symmetries that are expect to exist in the target function, and demonstrate in an ablation study they considerably improve performance. Our final networks are able to achieve move prediction accuracies of 41.1% and 44.4% on two different Go datasets, surpassing previous state of the art on this task by significant margins. Additionally, while previous move prediction programs have not yielded strong Go playing programs, we show that the networks trained in this work acquired high levels of skill. Our convolutional neural networks can consistently defeat the well known Go program GNU Go, indicating it is state of the art among programs that do not use Monte Carlo Tree Search. It is also able to win some games against state of the art Go playing program Fuego while using a fraction of the play time. This success at playing Go indicates high level principles of the game were learned.", "score": 0.0, "mention_count": 7, "user_count": 5, "months": ["2015-05", "2015-07", "2015-08", "2015-09", "2015-11"], "last_mentioned": "2015-11-03T19:41:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1412.6980", "slug": "6d41d2f09c026403", "title": "Adam: A Method for Stochastic Optimization", "authors": ["Diederik P. Kingma", "Jimmy Ba"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1412.6980", "doi": null, "topics": ["Algorithms", "Statistics & probability", "Machine learning"], "description": "We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments.", "description_method": "abstract-excerpt", "abstract": "We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The method is straightforward to implement, is computationally efficient, has little memory requirements, is invariant to diagonal rescaling of the gradients, and is well suited for problems that are large in terms of data and/or parameters. The method is also appropriate for non-stationary objectives and problems with very noisy and/or sparse gradients. The hyper-parameters have intuitive interpretations and typically require little tuning. Some connections to related algorithms, on which Adam was inspired, are discussed. We also analyze the theoretical convergence properties of the algorithm and provide a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework. Empirical results demonstrate that Adam works well in practice and compares favorably to other stochastic optimization methods. Finally, we discuss AdaMax, a variant of Adam based on the infinity norm.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2017-01", "2017-10", "2023-02", "2024-05", "2024-10", "2025-03", "2025-10"], "last_mentioned": "2025-10-18T06:40:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.04156", "slug": "1e6dd8e8f8c5c64e", "title": "Towards Biologically Plausible Deep Learning", "authors": ["Yoshua Bengio", "Dong-Hyun Lee", "Jorg Bornschein", "Thomas Mesnard", "Zhouhan Lin"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.04156", "doi": null, "topics": ["Machine learning", "Biology & medicine", "Neuroscience"], "description": "Neuroscientists have long criticised deep learning algorithms as incompatible with current knowledge of neurobiology.", "description_method": "abstract-excerpt", "abstract": "Neuroscientists have long criticised deep learning algorithms as incompatible with current knowledge of neurobiology. We explore more biologically plausible versions of deep representation learning, focusing here mostly on unsupervised learning but developing a learning mechanism that could account for supervised, unsupervised and reinforcement learning. The starting point is that the basic learning rule believed to govern synaptic weight updates (Spike-Timing-Dependent Plasticity) arises out of a simple update rule that makes a lot of sense from a machine learning point of view and can be interpreted as gradient descent on some objective function so long as the neuronal dynamics push firing rates towards better values of the objective function (be it supervised, unsupervised, or reward-driven). The second main idea is that this corresponds to a form of the variational EM algorithm, i.e., with approximate rather than exact posteriors, implemented by neural dynamics. Another contribution of this paper is that the gradients required for updating the hidden states in the above variational interpretation can be estimated using an approximation that only requires propagating activations forward and backward, with pairs of layers learning to form a denoising auto-encoder. Finally, we extend the theory about the probabilistic interpretation of auto-encoders to justify improved sampling schemes based on the generative interpretation of denoising auto-encoders, and we validate all these ideas on generative learning tasks.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2016-12", "2017-09", "2017-10", "2018-12", "2019-07", "2024-01"], "last_mentioned": "2024-01-04T14:53:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1504.05140", "slug": "d3a44e7c90b91fd6", "title": "Route Planning in Transportation Networks", "authors": ["Hannah Bast", "Daniel Delling", "Andrew Goldberg", "Matthias Müller-Hannemann", "Thomas Pajor", "Peter Sanders", "Dorothea Wagner", "Renato F. Werneck"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1504.05140", "doi": null, "topics": ["Algorithms", "Databases"], "description": "We survey recent advances in algorithms for route planning in transportation networks.", "description_method": "abstract-excerpt", "abstract": "We survey recent advances in algorithms for route planning in transportation networks. For road networks, we show that one can compute driving directions in milliseconds or less even at continental scale. A variety of techniques provide different trade-offs between preprocessing effort, space requirements, and query time. Some algorithms can answer queries in a fraction of a microsecond, while others can deal efficiently with real-time traffic. Journey planning on public transportation systems, although conceptually similar, is a significantly harder problem due to its inherent time-dependent and multicriteria nature. Although exact algorithms are fast enough for interactive queries on metropolitan transit systems, dealing with continent-sized instances requires simplifications or heavy preprocessing. The multimodal route planning problem, which seeks journeys combining schedule-based transportation (buses, trains) with unrestricted modes (walking, driving), is even harder, relying on approximate solutions even for metropolitan inputs.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2016-10", "2017-02", "2018-03", "2019-10", "2024-01", "2024-03"], "last_mentioned": "2024-03-22T17:25:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1508.05949", "slug": "c517c17ef0292ce4", "title": "Experimental loophole-free violation of a Bell inequality using entangled electron spins separated by 1.3 km", "authors": ["B. Hensen", "H. Bernien", "A. E. Dréau", "A. Reiserer", "N. Kalb", "M. S. Blok", "J. Ruitenberg", "R. F. L. Vermeulen", "R. N. Schouten", "C. Abellán", "W. Amaya", "V. Pruneri", "M. W. Mitchell", "M. Markham", "D. J. Twitchen", "D. Elkouss", "S. Wehner", "T. H. Taminiau", "R. Hanson"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1508.05949", "doi": "10.1038/nature15759", "topics": ["Economics", "Physics", "Statistics & probability"], "description": "For more than 80 years, the counterintuitive predictions of quantum theory have stimulated debate about the nature of reality.", "description_method": "abstract-excerpt", "abstract": "For more than 80 years, the counterintuitive predictions of quantum theory have stimulated debate about the nature of reality. In his seminal work, John Bell proved that no theory of nature that obeys locality and realism can reproduce all the predictions of quantum theory. Bell showed that in any local realist theory the correlations between distant measurements satisfy an inequality and, moreover, that this inequality can be violated according to quantum theory. This provided a recipe for experimental tests of the fundamental principles underlying the laws of nature. In the past decades, numerous ingenious Bell inequality tests have been reported. However, because of experimental limitations, all experiments to date required additional assumptions to obtain a contradiction with local realism, resulting in loopholes. Here we report on a Bell experiment that is free of any such additional assumption and thus directly tests the principles underlying Bell's inequality. We employ an event-ready scheme that enables the generation of high-fidelity entanglement between distant electron spins. Efficient spin readout avoids the fair sampling assumption (detection loophole), while the use of fast random basis selection and readout combined with a spatial separation of 1.3 km ensure the required locality conditions. We perform 245 trials testing the CHSH-Bell inequality $S \\leq 2$ and find $S = 2.42 \\pm 0.20$. A null hypothesis test yields a probability of $p = 0.039$ that a local-realist model for space-like separated sites produces data with a violation at least as large as observed, even when allowing for memory in the devices. This result rules out large classes of local realist theories, and paves the way for implementing device-independent quantum-secure communication and randomness certification.", "score": 0.0, "mention_count": 7, "user_count": 5, "months": ["2018-08", "2019-08", "2021-05", "2022-01", "2024-09"], "last_mentioned": "2024-09-13T21:16:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1509.01549", "slug": "fb2ad2469318da35", "title": "Giraffe: Using Deep Reinforcement Learning to Play Chess", "authors": ["Matthew Lai"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1509.01549", "doi": null, "topics": ["Machine learning", "Psychology"], "description": "This report presents Giraffe, a chess engine that uses self-play to discover all its domain-specific knowledge, with minimal hand-crafted knowledge given by the programmer.", "description_method": "abstract-excerpt", "abstract": "This report presents Giraffe, a chess engine that uses self-play to discover all its domain-specific knowledge, with minimal hand-crafted knowledge given by the programmer. Unlike previous attempts using machine learning only to perform parameter-tuning on hand-crafted evaluation functions, Giraffe's learning system also performs automatic feature extraction and pattern recognition. The trained evaluation function performs comparably to the evaluation functions of state-of-the-art chess engines - all of which containing thousands of lines of carefully hand-crafted pattern recognizers, tuned over many years by both computer chess experts and human chess masters. Giraffe is the most successful attempt thus far at using end-to-end machine learning to play chess.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2015-09", "2016-03", "2017-10", "2017-11", "2017-12"], "last_mentioned": "2017-12-06T04:30:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1602.01783", "slug": "bc49b01115fc5802", "title": "Asynchronous Methods for Deep Reinforcement Learning", "authors": ["Volodymyr Mnih", "Adrià Puigdomènech Badia", "Mehdi Mirza", "Alex Graves", "Timothy P. Lillicrap", "Tim Harley", "David Silver", "Koray Kavukcuoglu"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1602.01783", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "We propose a conceptually simple and lightweight framework for deep reinforcement learning that uses asynchronous gradient descent for optimization of deep neural network controllers.", "description_method": "abstract-excerpt", "abstract": "We propose a conceptually simple and lightweight framework for deep reinforcement learning that uses asynchronous gradient descent for optimization of deep neural network controllers. We present asynchronous variants of four standard reinforcement learning algorithms and show that parallel actor-learners have a stabilizing effect on training allowing all four methods to successfully train neural network controllers. The best performing method, an asynchronous variant of actor-critic, surpasses the current state-of-the-art on the Atari domain while training for half the time on a single multi-core CPU instead of a GPU. Furthermore, we show that asynchronous actor-critic succeeds on a wide variety of continuous motor control problems as well as on a new task of navigating random 3D mazes using a visual input.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2016-04", "2016-06", "2016-08", "2016-09", "2016-12", "2017-03", "2017-12"], "last_mentioned": "2017-12-14T10:51:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1603.05279", "slug": "4f7b88e4d221d369", "title": "XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks", "authors": ["Mohammad Rastegari", "Vicente Ordonez", "Joseph Redmon", "Ali Farhadi"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1603.05279", "doi": null, "topics": ["Computer vision", "Machine learning"], "description": "We propose two efficient approximations to standard convolutional neural networks: Binary-Weight-Networks and XNOR-Networks.", "description_method": "abstract-excerpt", "abstract": "We propose two efficient approximations to standard convolutional neural networks: Binary-Weight-Networks and XNOR-Networks. In Binary-Weight-Networks, the filters are approximated with binary values resulting in 32x memory saving. In XNOR-Networks, both the filters and the input to convolutional layers are binary. XNOR-Networks approximate convolutions using primarily binary operations. This results in 58x faster convolutional operations and 32x memory savings. XNOR-Nets offer the possibility of running state-of-the-art networks on CPUs (rather than GPUs) in real-time. Our binary networks are simple, accurate, efficient, and work on challenging visual tasks. We evaluate our approach on the ImageNet classification task. The classification accuracy with a Binary-Weight-Network version of AlexNet is only 2.9% less than the full-precision AlexNet (in top-1 measure). We compare our method with recent network binarization methods, BinaryConnect and BinaryNets, and outperform these methods by large margins on ImageNet, more than 16% in top-1 accuracy.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2016-04", "2016-07", "2017-01", "2017-02", "2024-11"], "last_mentioned": "2024-11-09T18:28:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1605.06065", "slug": "9d51f62e0c8aaf7c", "title": "One-shot Learning with Memory-Augmented Neural Networks", "authors": ["Adam Santoro", "Sergey Bartunov", "Matthew Botvinick", "Daan Wierstra", "Timothy Lillicrap"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1605.06065", "doi": null, "topics": ["Machine learning"], "description": "Despite recent breakthroughs in the applications of deep neural networks, one setting that presents a persistent challenge is that of \"one-shot learning.\" Traditional gradient-based networks require a lot of data to learn, often through extensive iterative training.", "description_method": "abstract-excerpt", "abstract": "Despite recent breakthroughs in the applications of deep neural networks, one setting that presents a persistent challenge is that of \"one-shot learning.\" Traditional gradient-based networks require a lot of data to learn, often through extensive iterative training. When new data is encountered, the models must inefficiently relearn their parameters to adequately incorporate the new information without catastrophic interference. Architectures with augmented memory capacities, such as Neural Turing Machines (NTMs), offer the ability to quickly encode and retrieve new information, and hence can potentially obviate the downsides of conventional models. Here, we demonstrate the ability of a memory-augmented neural network to rapidly assimilate new data, and leverage this data to make accurate predictions after only a few samples. We also introduce a new method for accessing an external memory that focuses on memory content, unlike previous methods that additionally use memory location-based focusing mechanisms.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2016-06", "2016-11", "2016-12", "2018-06"], "last_mentioned": "2018-06-03T06:01:57Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1608.03355", "slug": "f070458de7781760", "title": "A Practical Quantum Instruction Set Architecture", "authors": ["Robert S. Smith", "Michael J. Curtis", "William J. Zeng"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1608.03355", "doi": null, "topics": ["Physics", "Algorithms"], "description": "We introduce an abstract machine architecture for classical/quantum computations---including compilation---along with a quantum instruction language called Quil for explicitly writing these computations.", "description_method": "abstract-excerpt", "abstract": "We introduce an abstract machine architecture for classical/quantum computations---including compilation---along with a quantum instruction language called Quil for explicitly writing these computations. With this formalism, we discuss concrete implementations of the machine and non-trivial algorithms targeting them. The introduction of this machine dovetails with ongoing development of quantum computing technology, and makes possible portable descriptions of recent classical/quantum algorithms.", "score": 0.0, "mention_count": 7, "user_count": 4, "months": ["2016-08", "2016-09", "2017-06", "2018-04"], "last_mentioned": "2018-04-13T16:56:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.08144", "slug": "e7e24decb10e7002", "title": "Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation", "authors": ["Yonghui Wu", "Mike Schuster", "Zhifeng Chen", "Quoc V. Le", "Mohammad Norouzi", "Wolfgang Macherey", "Maxim Krikun", "Yuan Cao", "Qin Gao", "Klaus Macherey", "Jeff Klingner", "Apurva Shah", "Melvin Johnson", "Xiaobing Liu", "Łukasz Kaiser", "Stephan Gouws", "Yoshikiyo Kato", "Taku Kudo", "Hideto Kazawa", "Keith Stevens", "George Kurian", "Nishant Patil", "Wei Wang", "Cliff Young", "Jason Smith", "Jason Riesa", "Alex Rudnick", "Oriol Vinyals", "Greg Corrado", "Macduff Hughes", "Jeffrey Dean"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.08144", "doi": null, "topics": ["Other research"], "description": "Neural Machine Translation (NMT) is an end-to-end learning approach for automated translation, with the potential to overcome many of the weaknesses of conventional phrase-based translation systems.", "description_method": "abstract-excerpt", "abstract": "Neural Machine Translation (NMT) is an end-to-end learning approach for automated translation, with the potential to overcome many of the weaknesses of conventional phrase-based translation systems. Unfortunately, NMT systems are known to be computationally expensive both in training and in translation inference. Also, most NMT systems have difficulty with rare words. These issues have hindered NMT's use in practical deployments and services, where both accuracy and speed are essential. In this work, we present GNMT, Google's Neural Machine Translation system, which attempts to address many of these issues. Our model consists of a deep LSTM network with 8 encoder and 8 decoder layers using attention and residual connections. To improve parallelism and therefore decrease training time, our attention mechanism connects the bottom layer of the decoder to the top layer of the encoder. To accelerate the final translation speed, we employ low-precision arithmetic during inference computations. To improve handling of rare words, we divide words into a limited set of common sub-word units (\"wordpieces\") for both input and output. This method provides a good balance between the flexibility of \"character\"-delimited models and the efficiency of \"word\"-delimited models, naturally handles translation of rare words, and ultimately improves the overall accuracy of the system. Our beam search technique employs a length-normalization procedure and uses a coverage penalty, which encourages generation of an output sentence that is most likely to cover all the words in the source sentence. On the WMT'14 English-to-French and English-to-German benchmarks, GNMT achieves competitive results to state-of-the-art. Using a human side-by-side evaluation on a set of isolated simple sentences, it reduces translation errors by an average of 60% compared to Google's phrase-based production system.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2016-09", "2016-10", "2017-01", "2017-02", "2018-03", "2020-02"], "last_mentioned": "2020-02-18T11:06:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1611.02269", "slug": "aa82bca003418b82", "title": "Emergent Gravity and the Dark Universe", "authors": ["Erik P. Verlinde"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1611.02269", "doi": "10.21468/SciPostPhys.2.3.016", "topics": ["Physics", "Statistics & probability"], "description": "Recent theoretical progress indicates that spacetime and gravity emerge together from the entanglement structure of an underlying microscopic theory.", "description_method": "abstract-excerpt", "abstract": "Recent theoretical progress indicates that spacetime and gravity emerge together from the entanglement structure of an underlying microscopic theory. These ideas are best understood in Anti-de Sitter space, where they rely on the area law for entanglement entropy. The extension to de Sitter space requires taking into account the entropy and temperature associated with the cosmological horizon. Using insights from string theory, black hole physics and quantum information theory we argue that the positive dark energy leads to a thermal volume law contribution to the entropy that overtakes the area law precisely at the cosmological horizon. Due to the competition between area and volume law entanglement the microscopic de Sitter states do not thermalise at sub-Hubble scales: they exhibit memory effects in the form of an entropy displacement caused by matter. The emergent laws of gravity contain an additional `dark' gravitational force describing the `elastic' response due to the entropy displacement. We derive an estimate of the strength of this extra force in terms of the baryonic mass, Newton's constant and the Hubble acceleration scale a_0 =cH_0, and provide evidence for the fact that this additional `dark gravity~force' explains the observed phenomena in galaxies and clusters currently attributed to dark matter.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2016-11", "2016-12", "2017-03", "2024-03"], "last_mentioned": "2024-03-10T03:57:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1611.04558", "slug": "d1e032d2b26f3cb8", "title": "Google's Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation", "authors": ["Melvin Johnson", "Mike Schuster", "Quoc V. Le", "Maxim Krikun", "Yonghui Wu", "Zhifeng Chen", "Nikhil Thorat", "Fernanda Viégas", "Martin Wattenberg", "Greg Corrado", "Macduff Hughes", "Jeffrey Dean"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1611.04558", "doi": null, "topics": ["Other research"], "description": "We propose a simple solution to use a single Neural Machine Translation (NMT) model to translate between multiple languages.", "description_method": "abstract-excerpt", "abstract": "We propose a simple solution to use a single Neural Machine Translation (NMT) model to translate between multiple languages. Our solution requires no change in the model architecture from our base system but instead introduces an artificial token at the beginning of the input sentence to specify the required target language. The rest of the model, which includes encoder, decoder and attention, remains unchanged and is shared across all languages. Using a shared wordpiece vocabulary, our approach enables Multilingual NMT using a single model without any increase in parameters, which is significantly simpler than previous proposals for Multilingual NMT. Our method often improves the translation quality of all involved language pairs, even while keeping the total number of model parameters constant. On the WMT'14 benchmarks, a single multilingual model achieves comparable performance for English$\\rightarrow$French and surpasses state-of-the-art results for English$\\rightarrow$German. Similarly, a single multilingual model surpasses state-of-the-art results for French$\\rightarrow$English and German$\\rightarrow$English on WMT'14 and WMT'15 benchmarks respectively. On production corpora, multilingual models of up to twelve language pairs allow for better translation of many individual pairs. In addition to improving the translation quality of language pairs that the model was trained with, our models can also learn to perform implicit bridging between language pairs never seen explicitly during training, showing that transfer learning and zero-shot translation is possible for neural translation. Finally, we show analyses that hints at a universal interlingua representation in our models and show some interesting examples when mixing languages.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2016-11", "2017-01", "2017-03", "2017-07", "2018-02"], "last_mentioned": "2018-02-04T00:46:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1612.07828", "slug": "de1ebed741c340cc", "title": "Learning from Simulated and Unsupervised Images through Adversarial Training", "authors": ["Ashish Shrivastava", "Tomas Pfister", "Oncel Tuzel", "Josh Susskind", "Wenda Wang", "Russ Webb"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1612.07828", "doi": null, "topics": ["Algorithms"], "description": "With recent progress in graphics, it has become more tractable to train models on synthetic images, potentially avoiding the need for expensive annotations.", "description_method": "abstract-excerpt", "abstract": "With recent progress in graphics, it has become more tractable to train models on synthetic images, potentially avoiding the need for expensive annotations. However, learning from synthetic images may not achieve the desired performance due to a gap between synthetic and real image distributions. To reduce this gap, we propose Simulated+Unsupervised (S+U) learning, where the task is to learn a model to improve the realism of a simulator's output using unlabeled real data, while preserving the annotation information from the simulator. We develop a method for S+U learning that uses an adversarial network similar to Generative Adversarial Networks (GANs), but with synthetic images as inputs instead of random vectors. We make several key modifications to the standard GAN algorithm to preserve annotations, avoid artifacts, and stabilize training: (i) a 'self-regularization' term, (ii) a local adversarial loss, and (iii) updating the discriminator using a history of refined images. We show that this enables generation of highly realistic images, which we demonstrate both qualitatively and with a user study. We quantitatively evaluate the generated images by training models for gaze estimation and hand pose estimation. We show a significant improvement over using synthetic images, and achieve state-of-the-art results on the MPIIGaze dataset without any labeled real data.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2017-07", "2017-11", "2020-05", "2024-01"], "last_mentioned": "2024-01-18T12:03:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1707.07397", "slug": "28eb66611ca0ee0c", "title": "Synthesizing Robust Adversarial Examples", "authors": ["Anish Athalye", "Logan Engstrom", "Andrew Ilyas", "Kevin Kwok"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1707.07397", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "Standard methods for generating adversarial examples for neural networks do not consistently fool neural network classifiers in the physical world due to a combination of viewpoint shifts, camera noise, and other natural transformations, limiting their relevance to real-world systems.", "description_method": "abstract-excerpt", "abstract": "Standard methods for generating adversarial examples for neural networks do not consistently fool neural network classifiers in the physical world due to a combination of viewpoint shifts, camera noise, and other natural transformations, limiting their relevance to real-world systems. We demonstrate the existence of robust 3D adversarial objects, and we present the first algorithm for synthesizing examples that are adversarial over a chosen distribution of transformations. We synthesize two-dimensional adversarial images that are robust to noise, distortion, and affine transformation. We apply our algorithm to complex three-dimensional objects, using 3D-printing to manufacture the first physical adversarial objects. Our results demonstrate the existence of 3D adversarial objects in the physical world.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2017-10", "2018-01", "2019-03", "2019-04", "2020-02", "2020-07", "2024-09"], "last_mentioned": "2024-09-03T22:36:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1711.00165", "slug": "7bc6ee605b35bb48", "title": "Deep Neural Networks as Gaussian Processes", "authors": ["Jaehoon Lee", "Yasaman Bahri", "Roman Novak", "Samuel S. Schoenholz", "Jeffrey Pennington", "Jascha Sohl-Dickstein"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1711.00165", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Computer systems"], "description": "It has long been known that a single-layer fully-connected neural network with an i.i.d.", "description_method": "abstract-excerpt", "abstract": "It has long been known that a single-layer fully-connected neural network with an i.i.d. prior over its parameters is equivalent to a Gaussian process (GP), in the limit of infinite network width. This correspondence enables exact Bayesian inference for infinite width neural networks on regression tasks by means of evaluating the corresponding GP. Recently, kernel functions which mimic multi-layer random neural networks have been developed, but only outside of a Bayesian framework. As such, previous work has not identified that these kernels can be used as covariance functions for GPs and allow fully Bayesian prediction with a deep neural network. In this work, we derive the exact equivalence between infinitely wide deep networks and GPs. We further develop a computationally efficient pipeline to compute the covariance function for these GPs. We then use the resulting GPs to perform Bayesian inference for wide deep neural networks on MNIST and CIFAR-10. We observe that trained neural network accuracy approaches that of the corresponding GP with increasing layer width, and that the GP uncertainty is strongly correlated with trained network prediction error. We further find that test performance increases as finite-width trained networks are made wider and more similar to a GP, and thus that GP predictions typically outperform those of finite-width networks. Finally we connect the performance of these GPs to the recent theory of signal propagation in random neural networks.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2018-01", "2018-07", "2020-01", "2021-03", "2023-06", "2024-02"], "last_mentioned": "2024-02-25T09:40:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1712.09913", "slug": "6e82ea4976dc59a6", "title": "Visualizing the Loss Landscape of Neural Nets", "authors": ["Hao Li", "Zheng Xu", "Gavin Taylor", "Christoph Studer", "Tom Goldstein"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1712.09913", "doi": null, "topics": ["Machine learning", "Human–computer interaction"], "description": "Neural network training relies on our ability to find \"good\" minimizers of highly non-convex loss functions.", "description_method": "abstract-excerpt", "abstract": "Neural network training relies on our ability to find \"good\" minimizers of highly non-convex loss functions. It is well-known that certain network architecture designs (e.g., skip connections) produce loss functions that train easier, and well-chosen training parameters (batch size, learning rate, optimizer) produce minimizers that generalize better. However, the reasons for these differences, and their effects on the underlying loss landscape, are not well understood. In this paper, we explore the structure of neural loss functions, and the effect of loss landscapes on generalization, using a range of visualization methods. First, we introduce a simple \"filter normalization\" method that helps us visualize loss function curvature and make meaningful side-by-side comparisons between loss functions. Then, using a variety of visualizations, we explore how network architecture affects the loss landscape, and how training parameters affect the shape of minimizers.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2018-02", "2019-01", "2020-10", "2024-02", "2024-05", "2024-06", "2025-11"], "last_mentioned": "2025-11-02T07:51:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1806.00451", "slug": "50a4232394b0d50a", "title": "Do CIFAR-10 Classifiers Generalize to CIFAR-10?", "authors": ["Benjamin Recht", "Rebecca Roelofs", "Ludwig Schmidt", "Vaishaal Shankar"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1806.00451", "doi": null, "topics": ["Machine learning"], "description": "Machine learning is currently dominated by largely experimental work focused on improvements in a few key tasks.", "description_method": "abstract-excerpt", "abstract": "Machine learning is currently dominated by largely experimental work focused on improvements in a few key tasks. However, the impressive accuracy numbers of the best performing models are questionable because the same test sets have been used to select these models for multiple years now. To understand the danger of overfitting, we measure the accuracy of CIFAR-10 classifiers by creating a new test set of truly unseen images. Although we ensure that the new test set is as close to the original data distribution as possible, we find a large drop in accuracy (4% to 10%) for a broad range of deep learning models. Yet more recent models with higher original accuracy show a smaller drop and better overall performance, indicating that this drop is likely not due to overfitting based on adaptivity. Instead, we view our results as evidence that current accuracy numbers are brittle and susceptible to even minute natural variations in the data distribution.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2018-06", "2018-08", "2018-12", "2020-09", "2021-12", "2023-01"], "last_mentioned": "2023-01-18T13:55:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1903.03129", "slug": "8f081edd2909f658", "title": "SLIDE : In Defense of Smart Algorithms over Hardware Acceleration for Large-Scale Deep Learning Systems", "authors": ["Beidi Chen", "Tharun Medini", "James Farwell", "Sameh Gobriel", "Charlie Tai", "Anshumali Shrivastava"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1903.03129", "doi": null, "topics": ["Machine learning", "Algorithms", "Science & research"], "description": "Deep Learning (DL) algorithms are the central focus of modern machine learning systems.", "description_method": "abstract-excerpt", "abstract": "Deep Learning (DL) algorithms are the central focus of modern machine learning systems. As data volumes keep growing, it has become customary to train large neural networks with hundreds of millions of parameters to maintain enough capacity to memorize these volumes and obtain state-of-the-art accuracy. To get around the costly computations associated with large models and data, the community is increasingly investing in specialized hardware for model training. However, specialized hardware is expensive and hard to generalize to a multitude of tasks. The progress on the algorithmic front has failed to demonstrate a direct advantage over powerful hardware such as NVIDIA-V100 GPUs. This paper provides an exception. We propose SLIDE (Sub-LInear Deep learning Engine) that uniquely blends smart randomized algorithms, with multi-core parallelism and workload optimization. Using just a CPU, SLIDE drastically reduces the computations during both training and inference outperforming an optimized implementation of Tensorflow (TF) on the best available GPU. Our evaluations on industry-scale recommendation datasets, with large fully connected architectures, show that training with SLIDE on a 44 core CPU is more than 3.5 times (1 hour vs. 3.5 hours) faster than the same network trained using TF on Tesla V100 at any given accuracy level. On the same CPU hardware, SLIDE is over 10x faster than TF. We provide codes and scripts for reproducibility.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2020-03", "2020-12", "2021-04", "2021-08", "2022-02", "2025-02"], "last_mentioned": "2025-02-18T19:56:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1905.01302", "slug": "2640bbb74ebbc07b", "title": "Real Differences between OT and CRDT in Correctness and Complexity for Consistency Maintenance in Co-Editors", "authors": ["David Sun", "Chengzheng Sun", "Agustina Ng", "Weiwei Cai"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1905.01302", "doi": "10.1145/3392825", "topics": ["Algorithms", "Distributed systems", "Physics"], "description": "OT (Operational Transformation) was invented for supporting real-time co-editors in the late 1980s and has evolved to become core techniques widely used in today's working co-editors and adopted in industrial products.", "description_method": "abstract-excerpt", "abstract": "OT (Operational Transformation) was invented for supporting real-time co-editors in the late 1980s and has evolved to become core techniques widely used in today's working co-editors and adopted in industrial products. CRDT (Commutative Replicated Data Type) for co-editors was first proposed around 2006, under the name of WOOT (WithOut Operational Transformation). Follow-up CRDT variations are commonly labeled as \"post-OT\" techniques capable of making concurrent operations natively commutative in co-editors. On top of that, CRDT solutions have made broad claims of superiority over OT solutions, and often portrayed OT as an incorrect and inefficient technique. Over one decade later, however, CRDT is rarely found in working co-editors; OT remains the choice for building the vast majority of today's co-editors. Contradictions between the reality and CRDT's purported advantages have been the source of much confusion and debate among co-editing researcher sand developers. To seek truth from facts, we set out to conduct a comprehensive and critical review on representative OT and CRDT solutions and co-editors based on them. From this work, we have made important discoveries about OT and CRDT, and revealed facts and evidences that refute CRDT claims over OT on all accounts. These discoveries help explain the underlying reasons for the choice between OT and CRDT in the real world. We report these results in a series of three articles. In the second article of this series, we reveal the differences between OT and CRDT in their basic approaches to realizing the same general transformation and how such differences had resulted in different challenges and consequential correctness and complexity issues. Moreover, we reveal hidden complexity and algorithmic flaws with representative CRDT solutions, and discuss common myths and facts related to correctness and complexity of OT and CRDT.", "score": 0.0, "mention_count": 7, "user_count": 5, "months": ["2019-05", "2019-06", "2022-06", "2023-06", "2023-07"], "last_mentioned": "2023-07-09T19:08:31Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1911.13299", "slug": "3766d97f4fc747ad", "title": "What's Hidden in a Randomly Weighted Neural Network?", "authors": ["Vivek Ramanujan", "Mitchell Wortsman", "Aniruddha Kembhavi", "Ali Farhadi", "Mohammad Rastegari"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1911.13299", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "Training a neural network is synonymous with learning the values of the weights.", "description_method": "abstract-excerpt", "abstract": "Training a neural network is synonymous with learning the values of the weights. By contrast, we demonstrate that randomly weighted neural networks contain subnetworks which achieve impressive performance without ever training the weight values. Hidden in a randomly weighted Wide ResNet-50 we show that there is a subnetwork (with random weights) that is smaller than, but matches the performance of a ResNet-34 trained on ImageNet. Not only do these \"untrained subnetworks\" exist, but we provide an algorithm to effectively find them. We empirically show that as randomly weighted neural networks with fixed weights grow wider and deeper, an \"untrained subnetwork\" approaches a network with learned weights in accuracy. Our code and pretrained models are available at https://github.com/allenai/hidden-networks.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2020-02", "2020-03", "2020-07", "2022-07", "2023-04"], "last_mentioned": "2023-04-24T00:32:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2003.03052", "slug": "be38a780611e27ba", "title": "Combining GHOST and Casper", "authors": ["Vitalik Buterin", "Diego Hernandez", "Thor Kamphefner", "Khiem Pham", "Zhi Qiao", "Danny Ryan", "Juhyeok Sin", "Ying Wang", "Yan X Zhang"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2003.03052", "doi": null, "topics": ["Distributed systems"], "description": "We present \"Gasper,\" a proof-of-stake-based consensus protocol, which is an idealized version of the proposed Ethereum 2.0 beacon chain.", "description_method": "abstract-excerpt", "abstract": "We present \"Gasper,\" a proof-of-stake-based consensus protocol, which is an idealized version of the proposed Ethereum 2.0 beacon chain. The protocol combines Casper FFG, a finality tool, with LMD GHOST, a fork-choice rule. We prove safety, plausible liveness, and probabilistic liveness under different sets of assumptions.", "score": 0.0, "mention_count": 7, "user_count": 4, "months": ["2021-02", "2021-03", "2021-04", "2022-08"], "last_mentioned": "2022-08-12T16:20:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2009.06489", "slug": "9cfcc3ba373f78c7", "title": "The Hardware Lottery", "authors": ["Sara Hooker"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2009.06489", "doi": null, "topics": ["Economics", "Algorithms"], "description": "Hardware, systems and algorithms research communities have historically had different incentive structures and fluctuating motivation to engage with each other explicitly.", "description_method": "abstract-excerpt", "abstract": "Hardware, systems and algorithms research communities have historically had different incentive structures and fluctuating motivation to engage with each other explicitly. This historical treatment is odd given that hardware and software have frequently determined which research ideas succeed (and fail). This essay introduces the term hardware lottery to describe when a research idea wins because it is suited to the available software and hardware and not because the idea is superior to alternative research directions. Examples from early computer science history illustrate how hardware lotteries can delay research progress by casting successful ideas as failures. These lessons are particularly salient given the advent of domain specialized hardware which make it increasingly costly to stray off of the beaten path of research ideas. This essay posits that the gains from progress in computing are likely to become even more uneven, with certain research directions moving into the fast-lane while progress on others is further obstructed.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2020-11", "2022-04", "2023-04", "2023-07", "2023-10", "2023-12", "2026-06"], "last_mentioned": "2026-06-06T18:57:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2012.03141", "slug": "7d54e5c0bd6bdd7e", "title": "Automated Symbolic Verification of Telegram's MTProto 2.0", "authors": ["Marino Miculan", "Nicola Vitacolonna"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2012.03141", "doi": "10.5220/0010549601850197", "topics": ["Security & cryptography"], "description": "MTProto 2.0 is a suite of cryptographic protocols for instant messaging at the core of the popular Telegram messenger application.", "description_method": "abstract-excerpt", "abstract": "MTProto 2.0 is a suite of cryptographic protocols for instant messaging at the core of the popular Telegram messenger application. In this paper we analyse MTProto 2.0 using the symbolic verifier ProVerif. We provide fully automated proofs of the soundness of MTProto 2.0's authentication, normal chat, end-to-end encrypted chat, and rekeying mechanisms with respect to several security properties, including authentication, integrity, secrecy and perfect forward secrecy; at the same time, we discover that the rekeying protocol is vulnerable to an unknown key-share (UKS) attack. We proceed in an incremental way: each protocol is examined in isolation, relying only on the guarantees provided by the previous ones and the robustness of the basic cryptographic primitives. Our research proves the formal correctness of MTProto 2.0 w.r.t. most relevant security properties, and it can serve as a reference for implementation and analysis of clients and servers.", "score": 0.0, "mention_count": 7, "user_count": 5, "months": ["2021-01", "2022-10", "2023-04", "2024-08"], "last_mentioned": "2024-08-25T16:22:19Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2102.11174", "slug": "7d5c8c3617e9f1f2", "title": "Linear Transformers Are Secretly Fast Weight Programmers", "authors": ["Imanol Schlag", "Kazuki Irie", "Jürgen Schmidhuber"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2102.11174", "doi": null, "topics": ["Language models", "Programming languages", "Machine learning"], "description": "We show the formal equivalence of linearised self-attention mechanisms and fast weight controllers from the early '90s, where a ``slow\" neural net learns by gradient descent to program the ``fast weights\" of another net through sequences of elementary programming instructions which are additive outer products of self-invented activation p", "description_method": "abstract-excerpt", "abstract": "We show the formal equivalence of linearised self-attention mechanisms and fast weight controllers from the early '90s, where a ``slow\" neural net learns by gradient descent to program the ``fast weights\" of another net through sequences of elementary programming instructions which are additive outer products of self-invented activation patterns (today called keys and values). Such Fast Weight Programmers (FWPs) learn to manipulate the contents of a finite memory and dynamically interact with it. We infer a memory capacity limitation of recent linearised softmax attention variants, and replace the purely additive outer products by a delta rule-like programming instruction, such that the FWP can more easily learn to correct the current mapping from keys to values. The FWP also learns to compute dynamically changing learning rates. We also propose a new kernel function to linearise attention which balances simplicity and effectiveness. We conduct experiments on synthetic retrieval problems as well as standard machine translation and language modelling tasks which demonstrate the benefits of our methods.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2023-05", "2023-09", "2023-10", "2024-02", "2025-02"], "last_mentioned": "2025-02-15T03:04:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2104.00250", "slug": "2418fb2fd9eddc93", "title": "Retrofitting Effect Handlers onto OCaml", "authors": ["KC Sivaramakrishnan", "Stephen Dolan", "Leo White", "Tom Kelly", "Sadiq Jaffer", "Anil Madhavapeddy"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2104.00250", "doi": "10.1145/3453483.3454039", "topics": ["Programming languages"], "description": "Effect handlers have been gathering momentum as a mechanism for modular programming with user-defined effects.", "description_method": "abstract-excerpt", "abstract": "Effect handlers have been gathering momentum as a mechanism for modular programming with user-defined effects. Effect handlers allow for non-local control flow mechanisms such as generators, async/await, lightweight threads and coroutines to be composably expressed. We present a design and evaluate a full-fledged efficient implementation of effect handlers for OCaml, an industrial-strength multi-paradigm programming language. Our implementation strives to maintain the backwards compatibility and performance profile of existing OCaml code. Retrofitting effect handlers onto OCaml is challenging since OCaml does not currently have any non-local control flow mechanisms other than exceptions. Our implementation of effect handlers for OCaml: (i) imposes a mean 1% overhead on a comprehensive macro benchmark suite that does not use effect handlers; (ii) remains compatible with program analysis tools that inspect the stack; and (iii) is efficient for new code that makes use of effect handlers.", "score": 0.0, "mention_count": 7, "user_count": 4, "months": ["2021-05", "2022-01", "2022-06", "2022-12", "2023-03"], "last_mentioned": "2023-03-30T06:23:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2104.09864", "slug": "e74511f094c4e495", "title": "RoFormer: Enhanced Transformer with Rotary Position Embedding", "authors": ["Jianlin Su", "Yu Lu", "Shengfeng Pan", "Ahmed Murtadha", "Bo Wen", "Yunfeng Liu"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2104.09864", "doi": null, "topics": ["Language models"], "description": "Position encoding recently has shown effective in the transformer architecture.", "description_method": "abstract-excerpt", "abstract": "Position encoding recently has shown effective in the transformer architecture. It enables valuable supervision for dependency modeling between elements at different positions of the sequence. In this paper, we first investigate various methods to integrate positional information into the learning process of transformer-based language models. Then, we propose a novel method named Rotary Position Embedding(RoPE) to effectively leverage the positional information. Specifically, the proposed RoPE encodes the absolute position with a rotation matrix and meanwhile incorporates the explicit relative position dependency in self-attention formulation. Notably, RoPE enables valuable properties, including the flexibility of sequence length, decaying inter-token dependency with increasing relative distances, and the capability of equipping the linear self-attention with relative position encoding. Finally, we evaluate the enhanced transformer with rotary position embedding, also called RoFormer, on various long text classification benchmark datasets. Our experiments show that it consistently overcomes its alternatives. Furthermore, we provide a theoretical analysis to explain some experimental results. RoFormer is already integrated into Huggingface: \\url{https://huggingface.co/docs/transformers/model_doc/roformer}.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2023-03", "2023-04", "2023-08", "2025-04", "2025-08"], "last_mentioned": "2025-08-05T18:43:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2205.11502", "slug": "db5eef4feebb8ec8", "title": "On the Paradox of Learning to Reason from Data", "authors": ["Honghua Zhang", "Liunian Harold Li", "Tao Meng", "Kai-Wei Chang", "Guy Van den Broeck"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2205.11502", "doi": null, "topics": ["Statistics & probability", "Language models"], "description": "Logical reasoning is needed in a wide range of NLP tasks.", "description_method": "abstract-excerpt", "abstract": "Logical reasoning is needed in a wide range of NLP tasks. Can a BERT model be trained end-to-end to solve logical reasoning problems presented in natural language? We attempt to answer this question in a confined problem space where there exists a set of parameters that perfectly simulates logical reasoning. We make observations that seem to contradict each other: BERT attains near-perfect accuracy on in-distribution test examples while failing to generalize to other data distributions over the exact same problem space. Our study provides an explanation for this paradox: instead of learning to emulate the correct reasoning function, BERT has in fact learned statistical features that inherently exist in logical reasoning problems. We also show that it is infeasible to jointly remove statistical features from data, illustrating the difficulty of learning to reason in general. Our result naturally extends to other neural models and unveils the fundamental difference between learning to reason and learning to achieve high performance on NLP benchmarks using statistical features.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2022-12", "2023-02", "2023-03", "2023-07", "2023-09", "2024-10"], "last_mentioned": "2024-10-14T23:28:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2210.08277", "slug": "add03ebd36e95338", "title": "Deep Differentiable Logic Gate Networks", "authors": ["Felix Petersen", "Christian Borgelt", "Hilde Kuehne", "Oliver Deussen"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2210.08277", "doi": null, "topics": ["Machine learning"], "description": "Recently, research has increasingly focused on developing efficient neural network architectures.", "description_method": "abstract-excerpt", "abstract": "Recently, research has increasingly focused on developing efficient neural network architectures. In this work, we explore logic gate networks for machine learning tasks by learning combinations of logic gates. These networks comprise logic gates such as \"AND\" and \"XOR\", which allow for very fast execution. The difficulty in learning logic gate networks is that they are conventionally non-differentiable and therefore do not allow training with gradient descent. Thus, to allow for effective training, we propose differentiable logic gate networks, an architecture that combines real-valued logics and a continuously parameterized relaxation of the network. The resulting discretized logic gate networks achieve fast inference speeds, e.g., beyond a million images of MNIST per second on a single CPU core.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2023-02", "2023-12", "2024-09", "2025-05", "2025-11"], "last_mentioned": "2025-11-27T22:10:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2210.11610", "slug": "f4af5a9c4068dc0f", "title": "Large Language Models Can Self-Improve", "authors": ["Jiaxin Huang", "Shixiang Shane Gu", "Le Hou", "Yuexin Wu", "Xuezhi Wang", "Hongkun Yu", "Jiawei Han"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2210.11610", "doi": null, "topics": ["Language models", "Distributed systems"], "description": "Large Language Models (LLMs) have achieved excellent performances in various tasks.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) have achieved excellent performances in various tasks. However, fine-tuning an LLM requires extensive supervision. Human, on the other hand, may improve their reasoning abilities by self-thinking without external inputs. In this work, we demonstrate that an LLM is also capable of self-improving with only unlabeled datasets. We use a pre-trained LLM to generate \"high-confidence\" rationale-augmented answers for unlabeled questions using Chain-of-Thought prompting and self-consistency, and fine-tune the LLM using those self-generated solutions as target outputs. We show that our approach improves the general reasoning ability of a 540B-parameter LLM (74.4%->82.1% on GSM8K, 78.2%->83.0% on DROP, 90.0%->94.4% on OpenBookQA, and 63.4%->67.9% on ANLI-A3) and achieves state-of-the-art-level performance, without any ground truth label. We conduct ablation studies and show that fine-tuning on reasoning is critical for self-improvement.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2023-01", "2023-02", "2023-03", "2023-05"], "last_mentioned": "2023-05-09T13:35:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.00250", "slug": "784339ad37d30219", "title": "DensePose From WiFi", "authors": ["Jiaqi Geng", "Dong Huang", "Fernando De la Torre"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.00250", "doi": null, "topics": ["Machine learning", "Computer vision", "Security & cryptography"], "description": "Advances in computer vision and machine learning techniques have led to significant development in 2D and 3D human pose estimation from RGB cameras, LiDAR, and radars.", "description_method": "abstract-excerpt", "abstract": "Advances in computer vision and machine learning techniques have led to significant development in 2D and 3D human pose estimation from RGB cameras, LiDAR, and radars. However, human pose estimation from images is adversely affected by occlusion and lighting, which are common in many scenarios of interest. Radar and LiDAR technologies, on the other hand, need specialized hardware that is expensive and power-intensive. Furthermore, placing these sensors in non-public areas raises significant privacy concerns. To address these limitations, recent research has explored the use of WiFi antennas (1D sensors) for body segmentation and key-point body detection. This paper further expands on the use of the WiFi signal in combination with deep learning architectures, commonly used in computer vision, to estimate dense human pose correspondence. We developed a deep neural network that maps the phase and amplitude of WiFi signals to UV coordinates within 24 human regions. The results of the study reveal that our model can estimate the dense pose of multiple subjects, with comparable performance to image-based approaches, by utilizing WiFi signals as the only input. This paves the way for low-cost, broadly accessible, and privacy-preserving algorithms for human sensing.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2023-01", "2023-09", "2024-05", "2024-10", "2025-04"], "last_mentioned": "2025-04-13T19:57:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2302.10866", "slug": "617c7279bfc8f825", "title": "Hyena Hierarchy: Towards Larger Convolutional Language Models", "authors": ["Michael Poli", "Stefano Massaroli", "Eric Nguyen", "Daniel Y. Fu", "Tri Dao", "Stephen Baccus", "Yoshua Bengio", "Stefano Ermon", "Christopher Ré"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2302.10866", "doi": null, "topics": ["Language models", "Computer vision", "Machine learning"], "description": "Recent advances in deep learning have relied heavily on the use of large Transformers due to their ability to learn at scale.", "description_method": "abstract-excerpt", "abstract": "Recent advances in deep learning have relied heavily on the use of large Transformers due to their ability to learn at scale. However, the core building block of Transformers, the attention operator, exhibits quadratic cost in sequence length, limiting the amount of context accessible. Existing subquadratic methods based on low-rank and sparse approximations need to be combined with dense attention layers to match Transformers, indicating a gap in capability. In this work, we propose Hyena, a subquadratic drop-in replacement for attention constructed by interleaving implicitly parametrized long convolutions and data-controlled gating. In recall and reasoning tasks on sequences of thousands to hundreds of thousands of tokens, Hyena improves accuracy by more than 50 points over operators relying on state-spaces and other implicit and explicit methods, matching attention-based models. We set a new state-of-the-art for dense-attention-free architectures on language modeling in standard datasets (WikiText103 and The Pile), reaching Transformer quality with a 20% reduction in training compute required at sequence length 2K. Hyena operators are twice as fast as highly optimized attention at sequence length 8K, and 100x faster at sequence length 64K.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2023-04", "2023-05", "2023-09", "2025-02"], "last_mentioned": "2025-02-26T20:20:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2303.11156", "slug": "bb58e1dce57ecd0d", "title": "Can AI-Generated Text be Reliably Detected?", "authors": ["Vinu Sankar Sadasivan", "Aounon Kumar", "Sriram Balasubramanian", "Wenxiao Wang", "Soheil Feizi"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2303.11156", "doi": null, "topics": ["Security & cryptography", "Language models", "Machine learning"], "description": "Large Language Models (LLMs) perform impressively well in various applications.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) perform impressively well in various applications. However, the potential for misuse of these models in activities such as plagiarism, generating fake news, and spamming has raised concern about their responsible use. Consequently, the reliable detection of AI-generated text has become a critical area of research. AI text detectors have shown to be effective under their specific settings. In this paper, we stress-test the robustness of these AI text detectors in the presence of an attacker. We introduce recursive paraphrasing attack to stress test a wide range of detection schemes, including the ones using the watermarking as well as neural network-based detectors, zero shot classifiers, and retrieval-based detectors. Our experiments conducted on passages, each approximately 300 tokens long, reveal the varying sensitivities of these detectors to our attacks. Our findings indicate that while our recursive paraphrasing method can significantly reduce detection rates, it only slightly degrades text quality in many cases, highlighting potential vulnerabilities in current detection systems in the presence of an attacker. Additionally, we investigate the susceptibility of watermarked LLMs to spoofing attacks aimed at misclassifying human-written text as AI-generated. We demonstrate that an attacker can infer hidden AI text signatures without white-box access to the detection method, potentially leading to reputational risks for LLM developers. Finally, we provide a theoretical framework connecting the AUROC of the best possible detector to the Total Variation distance between human and AI text distributions. This analysis offers insights into the fundamental challenges of reliable detection as language models continue to advance. Our code is publicly available at https://github.com/vinusankars/Reliability-of-AI-text-detectors.", "score": 0.0, "mention_count": 7, "user_count": 5, "months": ["2023-03", "2023-04", "2023-07", "2024-04", "2025-09", "2026-08"], "last_mentioned": "2026-08-17T00:19:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.11206", "slug": "848f68d189a8e954", "title": "LIMA: Less Is More for Alignment", "authors": ["Chunting Zhou", "Pengfei Liu", "Puxin Xu", "Srini Iyer", "Jiao Sun", "Yuning Mao", "Xuezhe Ma", "Avia Efrat", "Ping Yu", "Lili Yu", "Susan Zhang", "Gargi Ghosh", "Mike Lewis", "Luke Zettlemoyer", "Omer Levy"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.11206", "doi": null, "topics": ["Language models", "Machine learning", "Statistics & probability"], "description": "Large language models are trained in two stages: (1) unsupervised pretraining from raw text, to learn general-purpose representations, and (2) large scale instruction tuning and reinforcement learning, to better align to end tasks and user preferences.", "description_method": "abstract-excerpt", "abstract": "Large language models are trained in two stages: (1) unsupervised pretraining from raw text, to learn general-purpose representations, and (2) large scale instruction tuning and reinforcement learning, to better align to end tasks and user preferences. We measure the relative importance of these two stages by training LIMA, a 65B parameter LLaMa language model fine-tuned with the standard supervised loss on only 1,000 carefully curated prompts and responses, without any reinforcement learning or human preference modeling. LIMA demonstrates remarkably strong performance, learning to follow specific response formats from only a handful of examples in the training data, including complex queries that range from planning trip itineraries to speculating about alternate history. Moreover, the model tends to generalize well to unseen tasks that did not appear in the training data. In a controlled human study, responses from LIMA are either equivalent or strictly preferred to GPT-4 in 43% of cases; this statistic is as high as 58% when compared to Bard and 65% versus DaVinci003, which was trained with human feedback. Taken together, these results strongly suggest that almost all knowledge in large language models is learned during pretraining, and only limited instruction tuning data is necessary to teach models to produce high quality output.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2023-05", "2023-06", "2023-09", "2024-04", "2024-10"], "last_mentioned": "2024-10-15T08:12:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2307.02477", "slug": "9e826f15a60b91f1", "title": "Reasoning or Reciting? Exploring the Capabilities and Limitations of Language Models Through Counterfactual Tasks", "authors": ["Zhaofeng Wu", "Linlu Qiu", "Alexis Ross", "Ekin Akyürek", "Boyuan Chen", "Bailin Wang", "Najoung Kim", "Jacob Andreas", "Yoon Kim"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2307.02477", "doi": null, "topics": ["Language models", "Psychology"], "description": "The impressive performance of recent language models across a wide range of tasks suggests that they possess a degree of abstract reasoning skills.", "description_method": "abstract-excerpt", "abstract": "The impressive performance of recent language models across a wide range of tasks suggests that they possess a degree of abstract reasoning skills. Are these skills general and transferable, or specialized to specific tasks seen during pretraining? To disentangle these effects, we propose an evaluation framework based on \"counterfactual\" task variants that deviate from the default assumptions underlying standard tasks. Across a suite of 11 tasks, we observe nontrivial performance on the counterfactual variants, but nevertheless find that performance substantially and consistently degrades compared to the default conditions. This suggests that while current LMs may possess abstract task-solving skills to an extent, they often also rely on narrow, non-transferable procedures for task-solving. These results motivate a more careful interpretation of language model performance that teases apart these aspects of behavior.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2023-07", "2023-12", "2024-06", "2024-07", "2024-08", "2024-11"], "last_mentioned": "2024-11-30T16:46:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2307.09288", "slug": "0db73367af87b21b", "title": "Llama 2: Open Foundation and Fine-Tuned Chat Models", "authors": ["Hugo Touvron", "Louis Martin", "Kevin Stone", "Peter Albert", "Amjad Almahairi", "Yasmine Babaei", "Nikolay Bashlykov", "Soumya Batra", "Prajjwal Bhargava", "Shruti Bhosale", "Dan Bikel", "Lukas Blecher", "Cristian Canton Ferrer", "Moya Chen", "Guillem Cucurull", "David Esiobu", "Jude Fernandes", "Jeremy Fu", "Wenyin Fu", "Brian Fuller", "Cynthia Gao", "Vedanuj Goswami", "Naman Goyal", "Anthony Hartshorn", "Saghar Hosseini", "Rui Hou", "Hakan Inan", "Marcin Kardas", "Viktor Kerkez", "Madian Khabsa", "Isabel Kloumann", "Artem Korenev", "Punit Singh Koura", "Marie-Anne Lachaux", "Thibaut Lavril", "Jenya Lee", "Diana Liskovich", "Yinghai Lu", "Yuning Mao", "Xavier Martinet", "Todor Mihaylov", "Pushkar Mishra", "Igor Molybog", "Yixin Nie", "Andrew Poulton", "Jeremy Reizenstein", "Rashi Rungta", "Kalyan Saladi", "Alan Schelten", "Ruan Silva", "Eric Michael Smith", "Ranjan Subramanian", "Xiaoqing Ellen Tan", "Binh Tang", "Ross Taylor", "Adina Williams", "Jian Xiang Kuan", "Puxin Xu", "Zheng Yan", "Iliyan Zarov", "Yuchen Zhang", "Angela Fan", "Melanie Kambadur", "Sharan Narang", "Aurelien Rodriguez", "Robert Stojnic", "Sergey Edunov", "Thomas Scialom"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2307.09288", "doi": null, "topics": ["Language models"], "description": "In this work, we develop and release Llama 2, a collection of pretrained and fine-tuned large language models (LLMs) ranging in scale from 7 billion to 70 billion parameters.", "description_method": "abstract-excerpt", "abstract": "In this work, we develop and release Llama 2, a collection of pretrained and fine-tuned large language models (LLMs) ranging in scale from 7 billion to 70 billion parameters. Our fine-tuned LLMs, called Llama 2-Chat, are optimized for dialogue use cases. Our models outperform open-source chat models on most benchmarks we tested, and based on our human evaluations for helpfulness and safety, may be a suitable substitute for closed-source models. We provide a detailed description of our approach to fine-tuning and safety improvements of Llama 2-Chat in order to enable the community to build on our work and contribute to the responsible development of LLMs.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2023-10", "2023-12", "2024-06", "2025-03", "2025-04", "2026-06"], "last_mentioned": "2026-06-24T19:50:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2308.04445", "slug": "52a16b181d48b001", "title": "Getting from Generative AI to Trustworthy AI: What LLMs might learn from Cyc", "authors": ["Doug Lenat", "Gary Marcus"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2308.04445", "doi": null, "topics": ["Language models"], "description": "Generative AI, the most popular current approach to AI, consists of large language models (LLMs) that are trained to produce outputs that are plausible, but not necessarily correct.", "description_method": "abstract-excerpt", "abstract": "Generative AI, the most popular current approach to AI, consists of large language models (LLMs) that are trained to produce outputs that are plausible, but not necessarily correct. Although their abilities are often uncanny, they are lacking in aspects of reasoning, leading LLMs to be less than completely trustworthy. Furthermore, their results tend to be both unpredictable and uninterpretable. We lay out 16 desiderata for future AI, and discuss an alternative approach to AI which could theoretically address many of the limitations associated with current approaches: AI educated with curated pieces of explicit knowledge and rules of thumb, enabling an inference engine to automatically deduce the logical entailments of all that knowledge. Even long arguments produced this way can be both trustworthy and interpretable, since the full step-by-step line of reasoning is always available, and for each step the provenance of the knowledge used can be documented and audited. There is however a catch: if the logical language is expressive enough to fully represent the meaning of anything we can say in English, then the inference engine runs much too slowly. That's why symbolic AI systems typically settle for some fast but much less expressive logic, such as knowledge graphs. We describe how one AI system, Cyc, has developed ways to overcome that tradeoff and is able to reason in higher order logic in real time. We suggest that any trustworthy general AI will need to hybridize the approaches, the LLM approach and more formal approach, and lay out a path to realizing that dream.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2023-09", "2024-06", "2024-11", "2025-04"], "last_mentioned": "2025-04-10T03:30:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.16834", "slug": "3d1608d00359c3b0", "title": "Discrete Diffusion Modeling by Estimating the Ratios of the Data Distribution", "authors": ["Aaron Lou", "Chenlin Meng", "Stefano Ermon"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.16834", "doi": null, "topics": ["Physics", "Machine learning", "Language models"], "description": "Despite their groundbreaking performance for many generative modeling tasks, diffusion models have fallen short on discrete data domains such as natural language.", "description_method": "abstract-excerpt", "abstract": "Despite their groundbreaking performance for many generative modeling tasks, diffusion models have fallen short on discrete data domains such as natural language. Crucially, standard diffusion models rely on the well-established theory of score matching, but efforts to generalize this to discrete structures have not yielded the same empirical gains. In this work, we bridge this gap by proposing score entropy, a novel loss that naturally extends score matching to discrete spaces, integrates seamlessly to build discrete diffusion models, and significantly boosts performance. Experimentally, we test our Score Entropy Discrete Diffusion models (SEDD) on standard language modeling tasks. For comparable model sizes, SEDD beats existing language diffusion paradigms (reducing perplexity by $25$-$75$\\%) and is competitive with autoregressive models, in particular outperforming GPT-2. Furthermore, compared to autoregressive mdoels, SEDD generates faithful text without requiring distribution annealing techniques like temperature scaling (around $6$-$8\\times$ better generative perplexity than un-annealed GPT-2), can trade compute and quality (similar quality with $32\\times$ fewer network evaluations), and enables controllable infilling (matching nucleus sampling quality while enabling other strategies besides left to right prompting).", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2025-02", "2025-03", "2025-05", "2025-07"], "last_mentioned": "2025-07-07T20:54:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2401.05566", "slug": "45152899d38c3c58", "title": "Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training", "authors": ["Evan Hubinger", "Carson Denison", "Jesse Mu", "Mike Lambert", "Meg Tong", "Monte MacDiarmid", "Tamera Lanham", "Daniel M. Ziegler", "Tim Maxwell", "Newton Cheng", "Adam Jermyn", "Amanda Askell", "Ansh Radhakrishnan", "Cem Anil", "David Duvenaud", "Deep Ganguli", "Fazl Barez", "Jack Clark", "Kamal Ndousse", "Kshitij Sachan", "Michael Sellitto", "Mrinank Sharma", "Nova DasSarma", "Roger Grosse", "Shauna Kravec", "Yuntao Bai", "Zachary Witten", "Marina Favaro", "Jan Brauner", "Holden Karnofsky", "Paul Christiano", "Samuel R. Bowman", "Logan Graham", "Jared Kaplan", "Sören Mindermann", "Ryan Greenblatt", "Buck Shlegeris", "Nicholas Schiefer", "Ethan Perez"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2401.05566", "doi": null, "topics": ["Psychology", "Machine learning", "Language models"], "description": "Humans are capable of strategically deceptive behavior: behaving helpfully in most situations, but then behaving very differently in order to pursue alternative objectives when given the opportunity.", "description_method": "abstract-excerpt", "abstract": "Humans are capable of strategically deceptive behavior: behaving helpfully in most situations, but then behaving very differently in order to pursue alternative objectives when given the opportunity. If an AI system learned such a deceptive strategy, could we detect it and remove it using current state-of-the-art safety training techniques? To study this question, we construct proof-of-concept examples of deceptive behavior in large language models (LLMs). For example, we train models that write secure code when the prompt states that the year is 2023, but insert exploitable code when the stated year is 2024. We find that such backdoor behavior can be made persistent, so that it is not removed by standard safety training techniques, including supervised fine-tuning, reinforcement learning, and adversarial training (eliciting unsafe behavior and then training to remove it). The backdoor behavior is most persistent in the largest models and in models trained to produce chain-of-thought reasoning about deceiving the training process, with the persistence remaining even when the chain-of-thought is distilled away. Furthermore, rather than removing backdoors, we find that adversarial training can teach models to better recognize their backdoor triggers, effectively hiding the unsafe behavior. Our results suggest that, once a model exhibits deceptive behavior, standard techniques could fail to remove such deception and create a false impression of safety.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2024-01", "2024-03", "2024-06", "2025-10", "2026-06", "2026-07"], "last_mentioned": "2026-07-03T19:00:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2401.18059", "slug": "ac4601bafaced6b4", "title": "RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval", "authors": ["Parth Sarthi", "Salman Abdullah", "Aditi Tuli", "Shubh Khanna", "Anna Goldie", "Christopher D. Manning"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2401.18059", "doi": null, "topics": ["Language models"], "description": "Retrieval-augmented language models can better adapt to changes in world state and incorporate long-tail knowledge.", "description_method": "abstract-excerpt", "abstract": "Retrieval-augmented language models can better adapt to changes in world state and incorporate long-tail knowledge. However, most existing methods retrieve only short contiguous chunks from a retrieval corpus, limiting holistic understanding of the overall document context. We introduce the novel approach of recursively embedding, clustering, and summarizing chunks of text, constructing a tree with differing levels of summarization from the bottom up. At inference time, our RAPTOR model retrieves from this tree, integrating information across lengthy documents at different levels of abstraction. Controlled experiments show that retrieval with recursive summaries offers significant improvements over traditional retrieval-augmented LMs on several tasks. On question-answering tasks that involve complex, multi-step reasoning, we show state-of-the-art results; for example, by coupling RAPTOR retrieval with the use of GPT-4, we can improve the best performance on the QuALITY benchmark by 20% in absolute accuracy.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2024-03", "2024-05", "2024-09"], "last_mentioned": "2024-09-21T08:18:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.14873", "slug": "a4b83522ec1afc63", "title": "Technical Report on the Pangram AI-Generated Text Classifier", "authors": ["Bradley Emi", "Max Spero"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.14873", "doi": null, "topics": ["Machine learning", "Language models", "Science & research"], "description": "We present Pangram Text, a transformer-based neural network trained to distinguish text written by large language models from text written by humans.", "description_method": "abstract-excerpt", "abstract": "We present Pangram Text, a transformer-based neural network trained to distinguish text written by large language models from text written by humans. Pangram Text outperforms zero-shot methods such as DetectGPT as well as leading commercial AI detection tools with over 38 times lower error rates on a comprehensive benchmark comprised of 10 text domains (student writing, creative writing, scientific writing, books, encyclopedias, news, email, scientific papers, short-form Q&A) and 8 open- and closed-source large language models. We propose a training algorithm, hard negative mining with synthetic mirrors, that enables our classifier to achieve orders of magnitude lower false positive rates on high-data domains such as reviews. Finally, we show that Pangram Text is not biased against nonnative English speakers and generalizes to domains and models unseen during training.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2024-10", "2025-11", "2026-01", "2026-02", "2026-04", "2026-07"], "last_mentioned": "2026-07-16T19:54:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2405.04434", "slug": "6a5532c1902ba809", "title": "DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model", "authors": [" DeepSeek-AI", "Aixin Liu", "Bei Feng", "Bin Wang", "Bingxuan Wang", "Bo Liu", "Chenggang Zhao", "Chengqi Dengr", "Chong Ruan", "Damai Dai", "Daya Guo", "Dejian Yang", "Deli Chen", "Dongjie Ji", "Erhang Li", "Fangyun Lin", "Fuli Luo", "Guangbo Hao", "Guanting Chen", "Guowei Li", "H. Zhang", "Hanwei Xu", "Hao Yang", "Haowei Zhang", "Honghui Ding", "Huajian Xin", "Huazuo Gao", "Hui Li", "Hui Qu", "J. L. Cai", "Jian Liang", "Jianzhong Guo", "Jiaqi Ni", "Jiashi Li", "Jin Chen", "Jingyang Yuan", "Junjie Qiu", "Junxiao Song", "Kai Dong", "Kaige Gao", "Kang Guan", "Lean Wang", "Lecong Zhang", "Lei Xu", "Leyi Xia", "Liang Zhao", "Liyue Zhang", "Meng Li", "Miaojun Wang", "Mingchuan Zhang", "Minghua Zhang", "Minghui Tang", "Mingming Li", "Ning Tian", "Panpan Huang", "Peiyi Wang", "Peng Zhang", "Qihao Zhu", "Qinyu Chen", "Qiushi Du", "R. J. Chen", "R. L. Jin", "Ruiqi Ge", "Ruizhe Pan", "Runxin Xu", "Ruyi Chen", "S. S. Li", "Shanghao Lu", "Shangyan Zhou", "Shanhuang Chen", "Shaoqing Wu", "Shengfeng Ye", "Shirong Ma", "Shiyu Wang", "Shuang Zhou", "Shuiping Yu", "Shunfeng Zhou", "Size Zheng", "T. Wang", "Tian Pei", "Tian Yuan", "Tianyu Sun", "W. L. Xiao", "Wangding Zeng", "Wei An", "Wen Liu", "Wenfeng Liang", "Wenjun Gao", "Wentao Zhang", "X. Q. Li", "Xiangyue Jin", "Xianzu Wang", "Xiao Bi", "Xiaodong Liu", "Xiaohan Wang", "Xiaojin Shen", "Xiaokang Chen", "Xiaosha Chen", "Xiaotao Nie", "Xiaowen Sun", "Xiaoxiang Wang", "Xin Liu", "Xin Xie", "Xingkai Yu", "Xinnan Song", "Xinyi Zhou", "Xinyu Yang", "Xuan Lu", "Xuecheng Su", "Y. Wu", "Y. K. Li", "Y. X. Wei", "Y. X. Zhu", "Yanhong Xu", "Yanping Huang", "Yao Li", "Yao Zhao", "Yaofeng Sun", "Yaohui Li", "Yaohui Wang", "Yi Zheng", "Yichao Zhang", "Yiliang Xiong", "Yilong Zhao", "Ying He", "Ying Tang", "Yishi Piao", "Yixin Dong", "Yixuan Tan", "Yiyuan Liu", "Yongji Wang", "Yongqiang Guo", "Yuchen Zhu", "Yuduan Wang", "Yuheng Zou", "Yukun Zha", "Yunxian Ma", "Yuting Yan", "Yuxiang You", "Yuxuan Liu", "Z. Z. Ren", "Zehui Ren", "Zhangli Sha", "Zhe Fu", "Zhen Huang", "Zhen Zhang", "Zhenda Xie", "Zhewen Hao", "Zhihong Shao", "Zhiniu Wen", "Zhipeng Xu", "Zhongyu Zhang", "Zhuoshu Li", "Zihan Wang", "Zihui Gu", "Zilin Li", "Ziwei Xie"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2405.04434", "doi": null, "topics": ["Economics", "Language models", "Machine learning"], "description": "We present DeepSeek-V2, a strong Mixture-of-Experts (MoE) language model characterized by economical training and efficient inference.", "description_method": "abstract-excerpt", "abstract": "We present DeepSeek-V2, a strong Mixture-of-Experts (MoE) language model characterized by economical training and efficient inference. It comprises 236B total parameters, of which 21B are activated for each token, and supports a context length of 128K tokens. DeepSeek-V2 adopts innovative architectures including Multi-head Latent Attention (MLA) and DeepSeekMoE. MLA guarantees efficient inference through significantly compressing the Key-Value (KV) cache into a latent vector, while DeepSeekMoE enables training strong models at an economical cost through sparse computation. Compared with DeepSeek 67B, DeepSeek-V2 achieves significantly stronger performance, and meanwhile saves 42.5% of training costs, reduces the KV cache by 93.3%, and boosts the maximum generation throughput to 5.76 times. We pretrain DeepSeek-V2 on a high-quality and multi-source corpus consisting of 8.1T tokens, and further perform Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) to fully unlock its potential. Evaluation results show that, even with only 21B activated parameters, DeepSeek-V2 and its chat versions still achieve top-tier performance among open-source models.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2025-01", "2025-02", "2025-04", "2025-05", "2025-12", "2026-06"], "last_mentioned": "2026-06-06T07:10:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2406.02061", "slug": "fcb841ef089fb0c7", "title": "Alice in Wonderland: Simple Tasks Showing Complete Reasoning Breakdown in State-Of-the-Art Large Language Models", "authors": ["Marianna Nezhurina", "Lucia Cipolina-Kun", "Mehdi Cherti", "Jenia Jitsev"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2406.02061", "doi": null, "topics": ["Machine learning", "Language models"], "description": "Large Language Models (LLMs) are often described as instances of foundation models that possess strong generalization obeying scaling laws, and therefore transfer robustly across various conditions in few- or zero-shot manner.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) are often described as instances of foundation models that possess strong generalization obeying scaling laws, and therefore transfer robustly across various conditions in few- or zero-shot manner. Such claims rely on standardized benchmarks that suppose to measure generalization and reasoning, where state-of-the-art (SOTA) models score high. We demonstrate here a dramatic breakdown of generalization and basic reasoning of all SOTA models claiming strong function, including large scale advanced models like GPT-4 or Claude 3 Opus, using a simple, short common sense math problem formulated in concise natural language, easily solvable by humans (AIW problem). The breakdown is dramatic as it manifests on a simple problem in both low average performance and strong performance fluctuations on natural variations in problem template that do not change either problem structure or its difficulty at all. By testing models on further control problems with similar form, we rule out that breakdown might be rooted in minor low-level issues like natural language or numbers parsing. We also observe strong overconfidence in the wrong solutions, expressed in form of plausible sounding explanation-like confabulations. Various standard interventions in an attempt to get the right solution, like chain-of-thought prompting, or urging the models to reconsider the wrong solutions again by multi step re-evaluation, fail. We use these observations to stimulate re-assessment of the capabilities of current generation of LLMs as claimed by standardized benchmarks. Such re-assessment also requires common action to create standardized benchmarks that would allow proper detection of such deficits in generalization and reasoning that obviously remain undiscovered by current state-of-the-art evaluation procedures, where SOTA LLMs manage to score high. Code: https://github.com/LAION-AI/AIW", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2024-06", "2024-07", "2024-09", "2024-10", "2024-12"], "last_mentioned": "2024-12-04T17:37:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2409.01754", "slug": "765e2e09c1103164", "title": "Empirical evidence of Large Language Model's influence on human spoken communication", "authors": ["Hiromu Yakura", "Ezequiel Lopez-Lopez", "Levin Brinkmann", "Ignacio de la Serna", "Lara Kirfel", "Prateek Gupta", "Ivan Soraperra", "Thomas F. Eisenmann", "Dirk U. Wulff", "Iyad Rahwan"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2409.01754", "doi": null, "topics": ["Language models", "Psychology"], "description": "From the printing press to social media, innovations in communication technology have repeatedly reshaped how ideas spread through human culture.", "description_method": "abstract-excerpt", "abstract": "From the printing press to social media, innovations in communication technology have repeatedly reshaped how ideas spread through human culture. Chatbots powered by generative artificial intelligence constitute a new medium, encoding cultural patterns in their neural representations and disseminating them in conversations with hundreds of millions of people. Whether these patterns transmit into human language, and ultimately shape human culture, is a fundamental question. While fully quantifying the causal impact of a chatbot like ChatGPT on human culture is challenging, lexical shifts in human spoken communication may offer an early indicator. Here we show that words preferentially generated by ChatGPT, such as delve, showcase, boast, intricacies and meticulous, increased abruptly in spontaneous human speech. A synthetic-control analysis of 737,083 hours of conversation from 824,634 podcast episodes, screened for unscripted speech, causally links this shift to ChatGPT's release. The measurable influence on spontaneous speech suggests that humans internalize the lexical choices of large language models (LLMs). A preregistered experiment (N = 496) confirms they do, as a brief chatbot interaction led participants to adopt its words as their own, persisting past a distractor task and confirmed in forced lexical choice, indicating entrenchment in the active vocabulary. Together these results show that machines trained on human data now feed their own traits back into human language, integrating LLMs into the ongoing processes of cultural evolution.. This coupling raises concerns about linguistic homogenization and the capacity of a few major AI providers for latent cultural influence at scale.", "score": 0.0, "mention_count": 7, "user_count": 3, "months": ["2025-05", "2025-08", "2026-01", "2026-02", "2026-05"], "last_mentioned": "2026-05-01T12:26:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2410.01201", "slug": "2544e92678326136", "title": "Were RNNs All We Needed?", "authors": ["Leo Feng", "Frederick Tung", "Mohamed Osama Ahmed", "Yoshua Bengio", "Hossein Hajimirsadeghi"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2410.01201", "doi": null, "topics": ["Language models", "Machine learning"], "description": "The introduction of Transformers in 2017 reshaped the landscape of deep learning.", "description_method": "abstract-excerpt", "abstract": "The introduction of Transformers in 2017 reshaped the landscape of deep learning. Originally proposed for sequence modelling, Transformers have since achieved widespread success across various domains. However, the scalability limitations of Transformers - particularly with respect to sequence length - have sparked renewed interest in novel recurrent models that are parallelizable during training, offer comparable performance, and scale more effectively. In this work, we revisit sequence modelling from a historical perspective, focusing on Recurrent Neural Networks (RNNs), which dominated the field for two decades before the rise of Transformers. Specifically, we examine LSTMs (1997) and GRUs (2014). We demonstrate that by simplifying these models, we can derive minimal versions (minLSTMs and minGRUs) that (1) use fewer parameters than their traditional counterparts, (2) are fully parallelizable during training, and (3) achieve surprisingly competitive performance on a range of tasks, rivalling recent models including Transformers.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2024-10", "2024-12", "2025-01", "2025-03", "2025-07", "2025-09", "2026-02"], "last_mentioned": "2026-02-04T00:28:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2512.24601", "slug": "ac41983c934a3432", "title": "Recursive Language Models", "authors": ["Alex L. Zhang", "Tim Kraska", "Omar Khattab"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2512.24601", "doi": null, "topics": ["Language models"], "description": "We study allowing large language models (LLMs) to process arbitrarily long prompts through the lens of inference-time scaling.", "description_method": "abstract-excerpt", "abstract": "We study allowing large language models (LLMs) to process arbitrarily long prompts through the lens of inference-time scaling. We propose Recursive Language Models (RLMs), a general inference paradigm that treats long prompts as part of an external environment and allows the LLM to programmatically examine, decompose, and recursively call itself over snippets of the prompt. We find that RLMs can successfully process inputs up to two orders of magnitude beyond model context windows and, even for shorter prompts, dramatically outperform the quality of vanilla frontier LLMs and common long-context and coding scaffolds (e.g., on GPT-5 by a median across the evaluated benchmarks of $26\\%$ against compaction, $130\\%$ against CodeAct with sub-calls, and $13\\%$ against Claude Code) across four diverse long-context tasks while having comparable cost. At a small scale, we post-train the first model around the RLM. Our model, RLM-Qwen3-8B, outperforms the underlying Qwen3-8B model by $28.3\\%$ on average and even approaches the quality of vanilla GPT-5 on three long-context tasks. Code is available at https://github.com/alexzhang13/rlm.", "score": 0.0, "mention_count": 7, "user_count": 4, "months": ["2026-01", "2026-02", "2026-03", "2026-04", "2026-06", "2026-07"], "last_mentioned": "2026-07-30T19:36:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2603.24755", "slug": "d7c589cec57b79dd", "title": "SlopCodeBench: Benchmarking How Coding Agents Degrade Over Long-Horizon Iterative Tasks", "authors": ["Gabriel Orlanski", "Devjeet Roy", "Alexander Yun", "Changho Shin", "Alex Gu", "Albert Ge", "Dyah Adila", "Nicholas Roberts", "Frederic Sala", "Aws Albarghouthi"], "year": 2026, "venue": "arXiv", "url": "https://arxiv.org/abs/2603.24755", "doi": "10.5281/zenodo.18405900,", "topics": ["Algorithms"], "description": "Software development is iterative, yet agentic coding benchmarks hide design issues through their single-shot setup.", "description_method": "abstract-excerpt", "abstract": "Software development is iterative, yet agentic coding benchmarks hide design issues through their single-shot setup. Recent iterative benchmarks attempt to remedy this but heavily constrain an agent's design decision space, making it impossible to faithfully measure how their decisions shape future extensions. We introduce SlopCodeBench, a benchmark of 36 problems and 196 checkpoints where agents repeatedly extend their own solutions. Unlike prior iterative benchmarks, our evolving specifications demand architectural decisions but leave internal structure to the agent. We measure two forms of degradation: structural erosion (concentrated complexity) and verbosity (redundant code). Evaluating 15 coding agents across open and closed models, we find that no agent fully solves any problem end-to-end, and the best agent passes 14.8% of checkpoints. Quality degrades across checkpoints, with structural erosion rising in 77% of trajectories and verbosity in 75.5%. Compared to 473 open-source Python repositories, agent code is 2.3x more verbose and 2.0x more eroded, and the human repositories degrade less often and by smaller margins across their git histories. Explicit quality guidance reduces initial verbosity and erosion by up to a third, without affecting degradation rates. SlopCodeBench provides the first measurement of code degradation under iterative extension, revealing that agents pass checkpoints while producing code that erodes and bloats with each turn.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2026-03", "2026-05", "2026-07"], "last_mentioned": "2026-07-28T07:15:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:quant-ph/0112176", "slug": "0f8d4380fd8cc6e2", "title": "Experimental realization of Shor's quantum factoring algorithm using nuclear magnetic resonance", "authors": ["Lieven M. K. Vandersypen", "Matthias Steffen", "Gregory Breyta", "Costantino S. Yannoni", "Mark H. Sherwood", "Isaac L. Chuang"], "year": 2001, "venue": "arXiv", "url": "https://arxiv.org/abs/quant-ph/0112176", "doi": "10.1038/414883a", "topics": ["Physics", "Algorithms", "Security & cryptography"], "description": "The number of steps any classical computer requires in order to find the prime factors of an $l$-digit integer $N$ increases exponentially with $l$, at least using algorithms known at present.", "description_method": "abstract-excerpt", "abstract": "The number of steps any classical computer requires in order to find the prime factors of an $l$-digit integer $N$ increases exponentially with $l$, at least using algorithms known at present. Factoring large integers is therefore conjectured to be intractable classically, an observation underlying the security of widely used cryptographic codes. Quantum computers, however, could factor integers in only polynomial time, using Shor's quantum factoring algorithm. Although important for the study of quantum computers, experimental demonstration of this algorithm has proved elusive. Here we report an implementation of the simplest instance of Shor's algorithm: factorization of ${N=15}$ (whose prime factors are 3 and 5). We use seven spin-1/2 nuclei in a molecule as quantum bits, which can be manipulated with room temperature liquid state nuclear magnetic resonance techniques. This method of using nuclei to store quantum information is in principle scalable to many quantum bit systems, but such scalability is not implied by the present work. The significance of our work lies in the demonstration of experimental and theoretical techniques for precise control and modelling of complex quantum computers. In particular, we present a simple, parameter-free but predictive model of decoherence effects in our system.", "score": 0.0, "mention_count": 7, "user_count": 4, "months": ["2013-09", "2021-07", "2024-03", "2024-11", "2026-03", "2026-04"], "last_mentioned": "2026-04-07T00:25:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature09268", "slug": "6a6ee1af91d7f00c", "title": "Global phytoplankton decline over the past century", "authors": ["Daniel G. Boyce", "Marlon R. Lewis", "Boris Worm"], "year": 2010, "venue": "Nature", "url": "https://doi.org/10.1038/nature09268", "doi": "10.1038/nature09268", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 7, "user_count": 5, "months": ["2018-05", "2018-11", "2018-12", "2019-02", "2019-08", "2019-09", "2020-03"], "last_mentioned": "2020-03-12T13:40:07Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/ncomms15112", "slug": "7d2cc17fac2fe287", "title": "An extra-uterine system to physiologically support the extreme premature lamb", "authors": ["Emily A. Partridge", "Marcus G. Davey", "Matthew A. Hornick", "Patrick E. McGovern", "Ali Y. Mejaddam", "Jesse D. Vrecenak", "Carmen Mesas-Burgos", "Aliza Olive", "Robert C. Caskey", "Theodore R. Weiland", "Jiancheng Han", "Alexander J. Schupper", "James T. Connelly", "Kevin C. Dysart", "Jack Rychik", "Holly L. Hedrick", "William H. Peranteau", "Alan W. Flake"], "year": 2017, "venue": "Nature Communications", "url": "https://doi.org/10.1038/ncomms15112", "doi": "10.1038/ncomms15112", "topics": ["Neuroscience"], "description": "Abstract In the developed world, extreme prematurity is the leading cause of neonatal mortality and morbidity due to a combination of organ immaturity and iatrogenic injury.", "description_method": "abstract-excerpt", "abstract": "Abstract In the developed world, extreme prematurity is the leading cause of neonatal mortality and morbidity due to a combination of organ immaturity and iatrogenic injury. Until now, efforts to extend gestation using extracorporeal systems have achieved limited success. Here we report the development of a system that incorporates a pumpless oxygenator circuit connected to the fetus of a lamb via an umbilical cord interface that is maintained within a closed ‘amniotic fluid’ circuit that closely reproduces the environment of the womb. We show that fetal lambs that are developmentally equivalent to the extreme premature human infant can be physiologically supported in this extra-uterine device for up to 4 weeks. Lambs on support maintain stable haemodynamics, have normal blood gas and oxygenation parameters and maintain patency of the fetal circulation. With appropriate nutritional support, lambs on the system demonstrate normal somatic growth, lung maturation and brain growth and myelination.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2017-04", "2021-03", "2022-05", "2022-12", "2023-08", "2026-05"], "last_mentioned": "2026-05-25T17:11:37Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-019-12808-z", "slug": "67994b6e988c1cfa", "title": "New elevation data triple estimates of global vulnerability to sea-level rise and coastal flooding", "authors": ["Scott A. Kulp", "Benjamin H. Strauss"], "year": 2019, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-019-12808-z", "doi": "10.1038/s41467-019-12808-z", "topics": ["Security & cryptography", "Psychology", "Machine learning"], "description": "Abstract Most estimates of global mean sea-level rise this century fall below 2 m.", "description_method": "abstract-excerpt", "abstract": "Abstract Most estimates of global mean sea-level rise this century fall below 2 m. This quantity is comparable to the positive vertical bias of the principle digital elevation model (DEM) used to assess global and national population exposures to extreme coastal water levels, NASA’s SRTM. CoastalDEM is a new DEM utilizing neural networks to reduce SRTM error. Here we show – employing CoastalDEM—that 190 M people (150–250 M, 90% CI) currently occupy global land below projected high tide lines for 2100 under low carbon emissions, up from 110 M today, for a median increase of 80 M. These figures triple SRTM-based values. Under high emissions, CoastalDEM indicates up to 630 M people live on land below projected annual flood levels for 2100, and up to 340 M for mid-century, versus roughly 250 M at present. We estimate one billion people now occupy land less than 10 m above current high tide lines, including 230 M below 1 m.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2019-10", "2019-11", "2019-12", "2021-04", "2021-10", "2025-07"], "last_mentioned": "2025-07-15T08:39:43Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41558-018-0321-8", "slug": "7c215c5cde3ef416", "title": "Bitcoin emissions alone could push global warming above 2°C", "authors": ["Camilo Mora", "Randi L. Rollins", "Katie Taladay", "Michael B. Kantar", "Mason K. Chock", "Mio Shimada", "Erik C. Franklin"], "year": 2018, "venue": "Nature Climate Change", "url": "https://doi.org/10.1038/s41558-018-0321-8", "doi": "10.1038/s41558-018-0321-8", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2018-10", "2018-11", "2018-12", "2019-04", "2020-11", "2021-06", "2022-03"], "last_mentioned": "2022-03-23T09:39:00Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2313-x", "slug": "4da840e33d4c970d", "title": "Isolation of SARS-CoV-2-related coronavirus from Malayan pangolins", "authors": ["Kangpeng Xiao", "Junqiong Zhai", "Yaoyu Feng", "Niu Zhou", "Xu Zhang", "Jie-Jian Zou", "Na Li", "Yaqiong Guo", "Xiaobing Li", "Xuejuan Shen", "Zhipeng Zhang", "Fanfan Shu", "Wanyi Huang", "Yu Li", "Ziding Zhang", "Rui-Ai Chen", "Ya-Jiang Wu", "Shi-Ming Peng", "Mian Huang", "Wei-Jun Xie", "Qin-Hui Cai", "Fang-Hui Hou", "Wu Chen", "Lihua Xiao", "Yongyi Shen"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2313-x", "doi": "10.1038/s41586-020-2313-x", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 7, "user_count": 3, "months": ["2020-11", "2020-12", "2021-02", "2021-03", "2022-06", "2022-07", "2024-11"], "last_mentioned": "2024-11-20T04:13:09Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2818-3", "slug": "a7af00ab4aeb3486", "title": "The major genetic risk factor for severe COVID-19 is inherited from Neanderthals", "authors": ["Hugo Zeberg", "Svante Pääbo"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2818-3", "doi": "10.1038/s41586-020-2818-3", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2020-11", "2020-12", "2021-02", "2021-08", "2021-11", "2021-12", "2023-10"], "last_mentioned": "2023-10-10T03:55:03Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2975-4", "slug": "8e94cde38012b288", "title": "Reprogramming to recover youthful epigenetic information and restore vision", "authors": ["Yuancheng Lu", "Benedikt Brommer", "Xiao Tian", "Anitha Krishnan", "Margarita Meer", "Chen Wang", "Daniel L. Vera", "Qiurui Zeng", "Doudou Yu", "Michael S. Bonkowski", "Jae-Hyun Yang", "Songlin Zhou", "Emma M. Hoffmann", "Margarete M. Karg", "Michael B. Schultz", "Alice E. Kane", "Noah Davidsohn", "Ekaterina Korobkina", "Karolina Chwalek", "Luis A. Rajman", "George M. Church", "Konrad Hochedlinger", "Vadim N. Gladyshev", "Steve Horvath", "Morgan E. Levine", "Meredith S. Gregory-Ksander", "Bruce R. Ksander", "Zhigang He", "David A. Sinclair"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2975-4", "doi": "10.1038/s41586-020-2975-4", "topics": ["Programming languages"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 7, "user_count": 4, "months": ["2020-12", "2021-01", "2021-04", "2022-01", "2022-02", "2023-01"], "last_mentioned": "2023-01-18T16:06:29Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-022-05172-4", "slug": "c175bfe094dbcd53", "title": "Discovering faster matrix multiplication algorithms with reinforcement learning", "authors": ["Alhussein Fawzi", "Matej Balog", "Aja Huang", "Thomas Hubert", "Bernardino Romera-Paredes", "Mohammadamin Barekatain", "Alexander Novikov", "Francisco J. R. Ruiz", "Julian Schrittwieser", "Grzegorz Swirszcz", "David Silver", "Demis Hassabis", "Pushmeet Kohli"], "year": 2022, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-022-05172-4", "doi": "10.1038/s41586-022-05172-4", "topics": ["Algorithms", "Machine learning", "Science & research"], "description": "Abstract Improving the efficiency of algorithms for fundamental computations can have a widespread impact, as it can affect the overall speed of a large amount of computations.", "description_method": "abstract-excerpt", "abstract": "Abstract Improving the efficiency of algorithms for fundamental computations can have a widespread impact, as it can affect the overall speed of a large amount of computations. Matrix multiplication is one such primitive task, occurring in many systems—from neural networks to scientific computing routines. The automatic discovery of algorithms using machine learning offers the prospect of reaching beyond human intuition and outperforming the current best human-designed algorithms. However, automating the algorithm discovery procedure is intricate, as the space of possible algorithms is enormous. Here we report a deep reinforcement learning approach based on AlphaZero 1 for discovering efficient and provably correct algorithms for the multiplication of arbitrary matrices. Our agent, AlphaTensor, is trained to play a single-player game where the objective is finding tensor decompositions within a finite factor space. AlphaTensor discovered algorithms that outperform the state-of-the-art complexity for many matrix sizes. Particularly relevant is the case of 4 × 4 matrices in a finite field, where AlphaTensor’s algorithm improves on Strassen’s two-level algorithm for the first time, to our knowledge, since its discovery 50 years ago 2 . We further showcase the flexibility of AlphaTensor through different use-cases: algorithms with state-of-the-art complexity for structured matrix multiplication and improved practical efficiency by optimizing matrix multiplication for runtime on specific hardware. Our results highlight AlphaTensor’s ability to accelerate the process of algorithmic discovery on a range of problems, and to optimize for different criteria.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2022-11", "2023-12", "2024-03", "2025-03", "2025-05"], "last_mentioned": "2025-05-31T11:18:44Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-020-0822-7", "slug": "783a15b3583a99e8", "title": "Estimating clinical severity of COVID-19 from the transmission dynamics in Wuhan, China", "authors": ["Joseph T. Wu", "Kathy Leung", "Mary Bushman", "Nishant Kishore", "Rene Niehus", "Pablo M. de Salazar", "Benjamin J. Cowling", "Marc Lipsitch", "Gabriel M. Leung"], "year": 2020, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-020-0822-7", "doi": "10.1038/s41591-020-0822-7", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2020-03", "2020-04", "2022-10"], "last_mentioned": "2022-10-14T07:25:41Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-020-62877-0", "slug": "37ca7d7f0c5d8618", "title": "Habitual use of GPS negatively impacts spatial memory during self-guided navigation", "authors": ["Louisa Dahmani", "Véronique D. Bohbot"], "year": 2020, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-020-62877-0", "doi": "10.1038/s41598-020-62877-0", "topics": ["Neuroscience"], "description": "Abstract Global Positioning System (GPS) navigation devices and applications have become ubiquitous over the last decade.", "description_method": "abstract-excerpt", "abstract": "Abstract Global Positioning System (GPS) navigation devices and applications have become ubiquitous over the last decade. However, it is unclear whether using GPS affects our own internal navigation system, or spatial memory, which critically relies on the hippocampus. We assessed the lifetime GPS experience of 50 regular drivers as well as various facets of spatial memory, including spatial memory strategy use, cognitive mapping, and landmark encoding using virtual navigation tasks. We first present cross-sectional results that show that people with greater lifetime GPS experience have worse spatial memory during self-guided navigation, i.e. when they are required to navigate without GPS. In a follow-up session, 13 participants were retested three years after initial testing. Although the longitudinal sample was small, we observed an important effect of GPS use over time, whereby greater GPS use since initial testing was associated with a steeper decline in hippocampal-dependent spatial memory. Importantly, we found that those who used GPS more did not do so because they felt they had a poor sense of direction, suggesting that extensive GPS use led to a decline in spatial memory rather than the other way around. These findings are significant in the context of society’s increasing reliance on GPS.", "score": 0.0, "mention_count": 7, "user_count": 7, "months": ["2025-03", "2025-06", "2025-07", "2025-09", "2026-01", "2026-02", "2026-04"], "last_mentioned": "2026-04-15T21:28:21Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-020-72798-7", "slug": "a6ce5b69ef462313", "title": "Efficacy of masks and face coverings in controlling outward aerosol particle emission from expiratory activities", "authors": ["Sima Asadi", "Christopher D. Cappa", "Santiago Barreda", "Anthony S. Wexler", "Nicole M. Bouvier", "William D. Ristenpart"], "year": 2020, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-020-72798-7", "doi": "10.1038/s41598-020-72798-7", "topics": ["Physics", "Biology & medicine"], "description": "Abstract The COVID-19 pandemic triggered a surge in demand for facemasks to protect against disease transmission.", "description_method": "abstract-excerpt", "abstract": "Abstract The COVID-19 pandemic triggered a surge in demand for facemasks to protect against disease transmission. In response to shortages, many public health authorities have recommended homemade masks as acceptable alternatives to surgical masks and N95 respirators. Although mask wearing is intended, in part, to protect others from exhaled, virus-containing particles, few studies have examined particle emission by mask-wearers into the surrounding air. Here, we measured outward emissions of micron-scale aerosol particles by healthy humans performing various expiratory activities while wearing different types of medical-grade or homemade masks. Both surgical masks and unvented KN95 respirators, even without fit-testing, reduce the outward particle emission rates by 90% and 74% on average during speaking and coughing, respectively, compared to wearing no mask, corroborating their effectiveness at reducing outward emission. These masks similarly decreased the outward particle emission of a coughing superemitter, who for unclear reasons emitted up to two orders of magnitude more expiratory particles via coughing than average. In contrast, shedding of non-expiratory micron-scale particulates from friable cellulosic fibers in homemade cotton-fabric masks confounded explicit determination of their efficacy at reducing expiratory particle emission. Audio analysis of the speech and coughing intensity confirmed that people speak more loudly, but do not cough more loudly, when wearing a mask. Further work is needed to establish the efficacy of cloth masks at blocking expiratory particles for speech and coughing at varied intensity and to assess whether virus-contaminated fabrics can generate aerosolized fomites, but the results strongly corroborate the efficacy of medical-grade masks and highlight the importance of regular washing of homemade masks.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2020-11", "2021-08", "2021-09", "2021-12", "2023-10"], "last_mentioned": "2023-10-01T21:48:44Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-020-80897-8", "slug": "fdee2eeebc83b1a4", "title": "Why do inequality and deprivation produce high crime and low trust?", "authors": ["Benoît De Courson", "Daniel Nettle"], "year": 2021, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-020-80897-8", "doi": "10.1038/s41598-020-80897-8", "topics": ["Economics", "Psychology"], "description": "AbstractHumans sometimes cooperate to mutual advantage, and sometimes exploit one another.", "description_method": "abstract-excerpt", "abstract": "AbstractHumans sometimes cooperate to mutual advantage, and sometimes exploit one another. In industrialised societies, the prevalence of exploitation, in the form of crime, is related to the distribution of economic resources: more unequal societies tend to have higher crime, as well as lower social trust. We created a model of cooperation and exploitation to explore why this should be. Distinctively, our model features a desperation threshold, a level of resources below which it is extremely damaging to fall. Agents do not belong to fixed types, but condition their behaviour on their current resource level and the behaviour in the population around them. We show that the optimal action for individuals who are close to the desperation threshold is to exploit others. This remains true even in the presence of severe and probable punishment for exploitation, since successful exploitation is the quickest route out of desperation, whereas being punished does not make already desperate states much worse. Simulated populations with a sufficiently unequal distribution of resources rapidly evolve an equilibrium of low trust and zero cooperation: desperate individuals try to exploit, and non-desperate individuals avoid interaction altogether. Making the distribution of resources more equal or increasing social mobility is generally effective in producing a high cooperation, high trust equilibrium; increasing punishment severity is not.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2023-04", "2023-08", "2024-05", "2025-08"], "last_mentioned": "2025-08-15T23:56:18Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-022-10928-z", "slug": "b87380d0743bfe0b", "title": "Increased emergency cardiovascular events among under-40 population in Israel during vaccine rollout and third COVID-19 wave", "authors": ["Christopher L. F. Sun", "Eli Jaffe", "Retsef Levi"], "year": 2022, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-022-10928-z", "doi": "10.1038/s41598-022-10928-z", "topics": ["Biology & medicine", "Statistics & probability"], "description": "Abstract Cardiovascular adverse conditions are caused by coronavirus disease 2019 (COVID-19) infections and reported as side-effects of the COVID-19 vaccines.", "description_method": "abstract-excerpt", "abstract": "Abstract Cardiovascular adverse conditions are caused by coronavirus disease 2019 (COVID-19) infections and reported as side-effects of the COVID-19 vaccines. Enriching current vaccine safety surveillance systems with additional data sources may improve the understanding of COVID-19 vaccine safety. Using a unique dataset from Israel National Emergency Medical Services (EMS) from 2019 to 2021, the study aims to evaluate the association between the volume of cardiac arrest and acute coronary syndrome EMS calls in the 16–39-year-old population with potential factors including COVID-19 infection and vaccination rates. An increase of 25% was detected in both call types during January–May 2021, compared with the years 2019–2020. Using Negative Binomial regression models, the weekly emergency call counts were significantly associated with the rates of 1st and 2nd vaccine doses administered to this age group but were not with COVID-19 infection rates. While not establishing causal relationships, the findings raise concerns regarding vaccine-induced undetected severe cardiovascular side-effects and underscore the already established causal relationship between vaccines and myocarditis, a frequent cause of unexpected cardiac arrest in young individuals. Surveillance of potential vaccine side-effects and COVID-19 outcomes should incorporate EMS and other health data to identify public health trends (e.g., increased in EMS calls), and promptly investigate potential underlying causes.", "score": 0.0, "mention_count": 7, "user_count": 5, "months": ["2022-04", "2022-05", "2022-06", "2022-10", "2022-12"], "last_mentioned": "2022-12-27T00:12:23Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-024-54249-9", "slug": "b98970f49020dcc4", "title": "Single dose creatine improves cognitive performance and induces changes in cerebral high energy phosphates during sleep deprivation", "authors": ["Ali Gordji-Nejad", "Andreas Matusch", "Sophie Kleedörfer", "Harshal Jayeshkumar Patel", "Alexander Drzezga", "David Elmenhorst", "Ferdinand Binkofski", "Andreas Bauer"], "year": 2024, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-024-54249-9", "doi": "10.1038/s41598-024-54249-9", "topics": ["Neuroscience", "Biology & medicine", "Statistics & probability"], "description": "AbstractThe inverse effects of creatine supplementation and sleep deprivation on high energy phosphates, neural creatine, and cognitive performances suggest that creatine is a suitable candidate for reducing the negative effects of sleep deprivation.", "description_method": "abstract-excerpt", "abstract": "AbstractThe inverse effects of creatine supplementation and sleep deprivation on high energy phosphates, neural creatine, and cognitive performances suggest that creatine is a suitable candidate for reducing the negative effects of sleep deprivation. With this, the main obstacle is the limited exogenous uptake by the central nervous system (CNS), making creatine only effective over a long-term diet of weeks. Thus far, only repeated dosing of creatine over weeks has been studied, yielding detectable changes in CNS levels. Based on the hypothesis that a high extracellular creatine availability and increased intracellular energy consumption will temporarily increase the central creatine uptake, subjects were orally administered a high single dose of creatinemonohydrate (0.35 g/kg) while performing cognitive tests during sleep deprivation. Two consecutive 31P-MRS scans, 1H-MRS, and cognitive tests were performed each at evening baseline, 3, 5.5, and 7.5 h after single dose creatine (0.35 g/kg) or placebo during sub-total 21 h sleep deprivation (SD). Our results show that creatine induces changes in PCr/Pi, ATP, tCr/tNAA, prevents a drop in pH level, and improves cognitive performance and processing speed. These outcomes suggest that a high single dose of creatine can partially reverse metabolic alterations and fatigue-related cognitive deterioration.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2024-08", "2026-02", "2026-05", "2026-06", "2026-07"], "last_mentioned": "2026-07-22T20:42:49Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-024-54271-x", "slug": "929f2105167a47f2", "title": "The carbon emissions of writing and illustrating are lower for AI than for humans", "authors": ["Bill Tomlinson", "Rebecca W. Black", "Donald J. Patterson", "Andrew W. Torrance"], "year": 2024, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-024-54271-x", "doi": "10.1038/s41598-024-54271-x", "topics": ["Other research"], "description": "Abstract As AI systems proliferate, their greenhouse gas emissions are an increasingly important concern for human societies.", "description_method": "abstract-excerpt", "abstract": "Abstract As AI systems proliferate, their greenhouse gas emissions are an increasingly important concern for human societies. In this article, we present a comparative analysis of the carbon emissions associated with AI systems (ChatGPT, BLOOM, DALL-E2, Midjourney) and human individuals performing equivalent writing and illustrating tasks. Our findings reveal that AI systems emit between 130 and 1500 times less CO2e per page of text generated compared to human writers, while AI illustration systems emit between 310 and 2900 times less CO2e per image than their human counterparts. Emissions analyses do not account for social impacts such as professional displacement, legality, and rebound effects. In addition, AI is not a substitute for all human tasks. Nevertheless, at present, the use of AI holds the potential to carry out several major activities at much lower emission levels than can humans.", "score": 0.0, "mention_count": 7, "user_count": 6, "months": ["2024-04", "2024-08", "2025-01", "2025-07", "2025-12", "2026-01"], "last_mentioned": "2026-01-15T14:23:18Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/srep14914", "slug": "610ca35e9c11b4e7", "title": "Polarization: A Key Difference between Man-made and Natural Electromagnetic Fields, in regard to Biological Activity", "authors": ["Dimitris J. Panagopoulos", "Olle Johansson", "George L. Carlo"], "year": 2015, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/srep14914", "doi": "10.1038/srep14914", "topics": ["Biology & medicine", "Statistics & probability"], "description": "Abstract In the present study we analyze the role of polarization in the biological activity of Electromagnetic Fields (EMFs)/Electromagnetic Radiation (EMR).", "description_method": "abstract-excerpt", "abstract": "Abstract In the present study we analyze the role of polarization in the biological activity of Electromagnetic Fields (EMFs)/Electromagnetic Radiation (EMR). All types of man-made EMFs/EMR - in contrast to natural EMFs/EMR - are polarized. Polarized EMFs/EMR can have increased biological activity, due to: 1) Ability to produce constructive interference effects and amplify their intensities at many locations. 2) Ability to force all charged/polar molecules and especially free ions within and around all living cells to oscillate on parallel planes and in phase with the applied polarized field. Such ionic forced-oscillations exert additive electrostatic forces on the sensors of cell membrane electro-sensitive ion channels, resulting in their irregular gating and consequent disruption of the cell’s electrochemical balance. These features render man-made EMFs/EMR more bioactive than natural non-ionizing EMFs/EMR. This explains the increasing number of biological effects discovered during the past few decades to be induced by man-made EMFs, in contrast to natural EMFs in the terrestrial environment which have always been present throughout evolution, although human exposure to the latter ones is normally of significantly higher intensities/energy and longer durations. Thus, polarization seems to be a trigger that significantly increases the probability for the initiation of biological/health effects.", "score": 0.0, "mention_count": 7, "user_count": 4, "months": ["2020-06", "2021-09", "2024-01", "2024-12"], "last_mentioned": "2024-12-12T15:44:58Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.abm7047", "slug": "db9bf48af14dcfee", "title": "Human generation times across the past 250,000 years", "authors": ["Richard J. Wang", "Samer I. Al-Saffar", "Jeffrey Rogers", "Matthew W. Hahn"], "year": 2023, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.abm7047", "doi": "10.1126/sciadv.abm7047", "topics": ["Biology & medicine"], "description": "The generation times of our recent ancestors can tell us about both the biology and social organization of prehistoric humans, placing human evolution on an absolute time scale.", "description_method": "abstract-excerpt", "abstract": "The generation times of our recent ancestors can tell us about both the biology and social organization of prehistoric humans, placing human evolution on an absolute time scale. We present a method for predicting historical male and female generation times based on changes in the mutation spectrum. Our analyses of whole-genome data reveal an average generation time of 26.9 years across the past 250,000 years, with fathers consistently older (30.7 years) than mothers (23.2 years). Shifts in sex-averaged generation times have been driven primarily by changes to the age of paternity, although we report a substantial increase in female generation times in the recent past. We also find a large difference in generation times among populations, reaching back to a time when all humans occupied Africa.", "score": 0.0, "mention_count": 7, "user_count": 5, "months": ["2023-01", "2023-09", "2023-11", "2024-05", "2025-06"], "last_mentioned": "2025-06-01T02:40:12Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciimmunol.ade2798", "slug": "5ff483dde3dd2156", "title": "Class switch toward noninflammatory, spike-specific IgG4 antibodies after repeated SARS-CoV-2 mRNA vaccination", "authors": ["Pascal Irrgang", "Juliane Gerling", "Katharina Kocher", "Dennis Lapuente", "Philipp Steininger", "Katharina Habenicht", "Monika Wytopil", "Stephanie Beileke", "Simon Schäfer", "Jahn Zhong", "George Ssebyatika", "Thomas Krey", "Valeria Falcone", "Christine Schülein", "Antonia Sophia Peter", "Krystelle Nganou-Makamdop", "Hartmut Hengel", "Jürgen Held", "Christian Bogdan", "Klaus Überla", "Kilian Schober", "Thomas H. Winkler", "Matthias Tenbusch"], "year": 2023, "venue": "Science Immunology", "url": "https://doi.org/10.1126/sciimmunol.ade2798", "doi": "10.1126/sciimmunol.ade2798", "topics": ["Biology & medicine"], "description": "RNA vaccines are efficient preventive measures to combat the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic.", "description_method": "abstract-excerpt", "abstract": "RNA vaccines are efficient preventive measures to combat the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic. High levels of neutralizing SARS-CoV-2 antibodies are an important component of vaccine-induced immunity. Shortly after the initial two mRNA vaccine doses, the immunoglobulin G (IgG) response mainly consists of the proinflammatory subclasses IgG1 and IgG3. Here, we report that several months after the second vaccination, SARS-CoV-2–specific antibodies were increasingly composed of noninflammatory IgG4, which were further boosted by a third mRNA vaccination and/or SARS-CoV-2 variant breakthrough infections. IgG4 antibodies among all spike-specific IgG antibodies rose, on average, from 0.04% shortly after the second vaccination to 19.27% late after the third vaccination. This induction of IgG4 antibodies was not observed after homologous or heterologous SARS-CoV-2 vaccination with adenoviral vectors. Single-cell sequencing and flow cytometry revealed substantial frequencies of IgG4-switched B cells within the spike-binding memory B cell population [median of 14.4%; interquartile range (IQR) of 6.7 to 18.1%] compared with the overall memory B cell repertoire (median of 1.3%; IQR of 0.9 to 2.2%) after three immunizations. This class switch was associated with a reduced capacity of the spike-specific antibodies to mediate antibody-dependent cellular phagocytosis and complement deposition. Because Fc-mediated effector functions are critical for antiviral immunity, these findings may have consequences for the choice and timing of vaccination regimens using mRNA vaccines, including future booster immunizations against SARS-CoV-2.", "score": 0.0, "mention_count": 7, "user_count": 5, "months": ["2022-12", "2023-01", "2023-06", "2023-09", "2024-04"], "last_mentioned": "2024-04-17T22:25:28Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0901.4016", "slug": "456022c6fe49434b", "title": "A Proposal for Proquints: Identifiers that are Readable, Spellable, and Pronounceable", "authors": ["Daniel Shawcross Wilkerson"], "year": 2009, "venue": "arXiv", "url": "https://arxiv.org/abs/0901.4016", "doi": null, "topics": ["Other research"], "description": "Identifiers (IDs) are pervasive throughout our modern life.", "description_method": "abstract-excerpt", "abstract": "Identifiers (IDs) are pervasive throughout our modern life. We suggest that these IDs would be easier to manage and remember if they were easily readable, spellable, and pronounceable. As a solution to this problem we propose using PRO-nouncable QUINT-uplets of alternating unambiguous consonants and vowels: _proquints_.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2014-12", "2015-05", "2018-09", "2019-11", "2020-02", "2020-09"], "last_mentioned": "2020-09-16T02:15:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0908.1787", "slug": "ca776bccb4edba74", "title": "Quantum Picturalism", "authors": ["Bob Coecke"], "year": 2009, "venue": "arXiv", "url": "https://arxiv.org/abs/0908.1787", "doi": "10.1080/00107510903257624", "topics": ["Physics", "Mathematics", "Programming languages"], "description": "The quantum mechanical formalism doesn't support our intuition, nor does it elucidate the key concepts that govern the behaviour of the entities that are subject to the laws of quantum physics.", "description_method": "abstract-excerpt", "abstract": "The quantum mechanical formalism doesn't support our intuition, nor does it elucidate the key concepts that govern the behaviour of the entities that are subject to the laws of quantum physics. The arrays of complex numbers are kin to the arrays of 0s and 1s of the early days of computer programming practice. In this review we present steps towards a diagrammatic `high-level' alternative for the Hilbert space formalism, one which appeals to our intuition. It allows for intuitive reasoning about interacting quantum systems, and trivialises many otherwise involved and tedious computations. It clearly exposes limitations such as the no-cloning theorem, and phenomena such as quantum teleportation. As a logic, it supports `automation'. It allows for a wider variety of underlying theories, and can be easily modified, having the potential to provide the required step-stone towards a deeper conceptual understanding of quantum theory, as well as its unification with other physical theories. Specific applications discussed here are purely diagrammatic proofs of several quantum computational schemes, as well as an analysis of the structural origin of quantum non-locality. The underlying mathematical foundation of this high-level diagrammatic formalism relies on so-called monoidal categories, a product of a fairly recent development in mathematics. These monoidal categories do not only provide a natural foundation for physical theories, but also for proof theory, logic, programming languages, biology, cooking, ... The challenge is to discover the necessary additional pieces of structure that allow us to predict genuine quantum phenomena.", "score": 0.0, "mention_count": 6, "user_count": 4, "months": ["2015-06", "2018-06", "2020-05", "2020-08", "2024-12"], "last_mentioned": "2024-12-05T00:43:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1207.7214", "slug": "b72a714942664d26", "title": "Observation of a new particle in the search for the Standard Model Higgs boson with the ATLAS detector at the LHC", "authors": [" The ATLAS Collaboration"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1207.7214", "doi": "10.1016/j.physletb.2012.08.020", "topics": ["Physics", "Statistics & probability"], "description": "A search for the Standard Model Higgs boson in proton-proton collisions with the ATLAS detector at the LHC is presented.", "description_method": "abstract-excerpt", "abstract": "A search for the Standard Model Higgs boson in proton-proton collisions with the ATLAS detector at the LHC is presented. The datasets used correspond to integrated luminosities of approximately 4.8 fb^-1 collected at sqrt(s) = 7 TeV in 2011 and 5.8 fb^-1 at sqrt(s) = 8 TeV in 2012. Individual searches in the channels H->ZZ^(*)->llll, H->gamma gamma and H->WW->e nu mu nu in the 8 TeV data are combined with previously published results of searches for H->ZZ^(*), WW^(*), bbbar and tau^+tau^- in the 7 TeV data and results from improved analyses of the H->ZZ^(*)->llll and H->gamma gamma channels in the 7 TeV data. Clear evidence for the production of a neutral boson with a measured mass of 126.0 +/- 0.4(stat) +/- 0.4(sys) GeV is presented. This observation, which has a significance of 5.9 standard deviations, corresponding to a background fluctuation probability of 1.7x10^-9, is compatible with the production and decay of the Standard Model Higgs boson.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2015-08", "2019-12", "2021-01", "2023-07", "2024-10", "2026-03"], "last_mentioned": "2026-03-11T22:52:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1209.5145", "slug": "286b3a40d86e1b01", "title": "Julia: A Fast Dynamic Language for Technical Computing", "authors": ["Jeff Bezanson", "Stefan Karpinski", "Viral B. Shah", "Alan Edelman"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1209.5145", "doi": null, "topics": ["Programming languages", "Science & research"], "description": "Dynamic languages have become popular for scientific computing.", "description_method": "abstract-excerpt", "abstract": "Dynamic languages have become popular for scientific computing. They are generally considered highly productive, but lacking in performance. This paper presents Julia, a new dynamic language for technical computing, designed for performance from the beginning by adapting and extending modern programming language techniques. A design based on generic functions and a rich type system simultaneously enables an expressive programming model and successful type inference, leading to good performance for a wide range of programs. This makes it possible for much of the Julia library to be written in Julia itself, while also incorporating best-of-breed C and Fortran libraries.", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2013-04", "2014-11", "2015-11", "2018-08"], "last_mentioned": "2018-08-10T01:00:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1301.1069", "slug": "d5e57ae03726746d", "title": "A Snapshot of Foundational Attitudes Toward Quantum Mechanics", "authors": ["Maximilian Schlosshauer", "Johannes Kofler", "Anton Zeilinger"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1301.1069", "doi": "10.1016/j.shpsb.2013.04.004", "topics": ["Physics", "Science & research", "Mathematics"], "description": "Foundational investigations in quantum mechanics, both experimental and theoretical, gave birth to the field of quantum information science.", "description_method": "abstract-excerpt", "abstract": "Foundational investigations in quantum mechanics, both experimental and theoretical, gave birth to the field of quantum information science. Nevertheless, the foundations of quantum mechanics themselves remain hotly debated in the scientific community, and no consensus on essential questions has been reached. Here, we present the results of a poll carried out among 33 participants of a conference on the foundations of quantum mechanics. The participants completed a questionnaire containing 16 multiple-choice questions probing opinions on quantum-foundational issues. Participants included physicists, philosophers, and mathematicians. We describe our findings, identify commonly held views, and determine strong, medium, and weak correlations between the answers. Our study provides a unique snapshot of current views in the field of quantum foundations, as well as an analysis of the relationships between these views.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2014-11", "2015-09", "2015-10", "2017-10", "2018-09", "2020-06"], "last_mentioned": "2020-06-26T00:17:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1310.4546", "slug": "4c8d178070a5a40a", "title": "Distributed Representations of Words and Phrases and their Compositionality", "authors": ["Tomas Mikolov", "Ilya Sutskever", "Kai Chen", "Greg Corrado", "Jeffrey Dean"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1310.4546", "doi": null, "topics": ["Distributed systems"], "description": "The recently introduced continuous Skip-gram model is an efficient method for learning high-quality distributed vector representations that capture a large number of precise syntactic and semantic word relationships.", "description_method": "abstract-excerpt", "abstract": "The recently introduced continuous Skip-gram model is an efficient method for learning high-quality distributed vector representations that capture a large number of precise syntactic and semantic word relationships. In this paper we present several extensions that improve both the quality of the vectors and the training speed. By subsampling of the frequent words we obtain significant speedup and also learn more regular word representations. We also describe a simple alternative to the hierarchical softmax called negative sampling. An inherent limitation of word representations is their indifference to word order and their inability to represent idiomatic phrases. For example, the meanings of \"Canada\" and \"Air\" cannot be easily combined to obtain \"Air Canada\". Motivated by this example, we present a simple method for finding phrases in text, and show that learning good vector representations for millions of phrases is possible.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2015-08", "2016-08", "2019-06"], "last_mentioned": "2019-06-03T23:30:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1402.3722", "slug": "18b9cfa89effc041", "title": "word2vec Explained: deriving Mikolov et al.'s negative-sampling word-embedding method", "authors": ["Yoav Goldberg", "Omer Levy"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1402.3722", "doi": null, "topics": ["Language models", "Distributed systems"], "description": "The word2vec software of Tomas Mikolov and colleagues (https://code.google.com/p/word2vec/ ) has gained a lot of traction lately, and provides state-of-the-art word embeddings.", "description_method": "abstract-excerpt", "abstract": "The word2vec software of Tomas Mikolov and colleagues (https://code.google.com/p/word2vec/ ) has gained a lot of traction lately, and provides state-of-the-art word embeddings. The learning models behind the software are described in two research papers. We found the description of the models in these papers to be somewhat cryptic and hard to follow. While the motivations and presentation may be obvious to the neural-networks language-modeling crowd, we had to struggle quite a bit to figure out the rationale behind the equations. This note is an attempt to explain equation (4) (negative sampling) in \"Distributed Representations of Words and Phrases and their Compositionality\" by Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg Corrado and Jeffrey Dean.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2015-06", "2015-10", "2017-03", "2017-09", "2017-11"], "last_mentioned": "2017-11-14T19:07:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1405.4053", "slug": "766a9dafe7671b23", "title": "Distributed Representations of Sentences and Documents", "authors": ["Quoc V. Le", "Tomas Mikolov"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1405.4053", "doi": null, "topics": ["Algorithms", "Distributed systems", "Machine learning"], "description": "Many machine learning algorithms require the input to be represented as a fixed-length feature vector.", "description_method": "abstract-excerpt", "abstract": "Many machine learning algorithms require the input to be represented as a fixed-length feature vector. When it comes to texts, one of the most common fixed-length features is bag-of-words. Despite their popularity, bag-of-words features have two major weaknesses: they lose the ordering of the words and they also ignore semantics of the words. For example, \"powerful,\" \"strong\" and \"Paris\" are equally distant. In this paper, we propose Paragraph Vector, an unsupervised algorithm that learns fixed-length feature representations from variable-length pieces of texts, such as sentences, paragraphs, and documents. Our algorithm represents each document by a dense vector which is trained to predict words in the document. Its construction gives our algorithm the potential to overcome the weaknesses of bag-of-words models. Empirical results show that Paragraph Vectors outperform bag-of-words models as well as other techniques for text representations. Finally, we achieve new state-of-the-art results on several text classification and sentiment analysis tasks.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2014-12", "2015-08", "2015-11", "2018-03", "2023-12"], "last_mentioned": "2023-12-18T23:47:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1409.1556", "slug": "cb4537a4d670ccc3", "title": "Very Deep Convolutional Networks for Large-Scale Image Recognition", "authors": ["Karen Simonyan", "Andrew Zisserman"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1409.1556", "doi": null, "topics": ["Computer vision", "Psychology"], "description": "In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting.", "description_method": "abstract-excerpt", "abstract": "In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2014-09", "2015-03", "2015-04", "2015-07", "2016-03", "2021-06"], "last_mentioned": "2021-06-02T20:05:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1411.0352", "slug": "0b1077b8ca61747d", "title": "Simple and Effective Type Check Removal through Lazy Basic Block Versioning", "authors": ["Maxime Chevalier-Boisvert", "Marc Feeley"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1411.0352", "doi": null, "topics": ["Programming languages", "Algorithms"], "description": "Dynamically typed programming languages such as JavaScript and Python defer type checking to run time.", "description_method": "abstract-excerpt", "abstract": "Dynamically typed programming languages such as JavaScript and Python defer type checking to run time. In order to maximize performance, dynamic language VM implementations must attempt to eliminate redundant dynamic type checks. However, type inference analyses are often costly and involve tradeoffs between compilation time and resulting precision. This has lead to the creation of increasingly complex multi-tiered VM architectures. This paper introduces lazy basic block versioning, a simple JIT compilation technique which effectively removes redundant type checks from critical code paths. This novel approach lazily generates type-specialized versions of basic blocks on-the-fly while propagating context-dependent type information. This does not require the use of costly program analyses, is not restricted by the precision limitations of traditional type analyses and avoids the implementation complexity of speculative optimization techniques. We have implemented intraprocedural lazy basic block versioning in a JavaScript JIT compiler. This approach is compared with a classical flow-based type analysis. Lazy basic block versioning performs as well or better on all benchmarks. On average, 71% of type tests are eliminated, yielding speedups of up to 50%. We also show that our implementation generates more efficient machine code than TraceMonkey, a tracing JIT compiler for JavaScript, on several benchmarks. The combination of implementation simplicity, low algorithmic complexity and good run time performance makes basic block versioning attractive for baseline JIT compilers.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2015-02", "2015-12", "2016-03", "2022-01", "2024-02", "2025-02"], "last_mentioned": "2025-02-13T03:23:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1412.6564", "slug": "b5e6cb1ffcc3768f", "title": "Move Evaluation in Go Using Deep Convolutional Neural Networks", "authors": ["Chris J. Maddison", "Aja Huang", "Ilya Sutskever", "David Silver"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1412.6564", "doi": null, "topics": ["Computer vision", "Machine learning", "Databases"], "description": "The game of Go is more challenging than other board games, due to the difficulty of constructing a position or move evaluation function.", "description_method": "abstract-excerpt", "abstract": "The game of Go is more challenging than other board games, due to the difficulty of constructing a position or move evaluation function. In this paper we investigate whether deep convolutional networks can be used to directly represent and learn this knowledge. We train a large 12-layer convolutional neural network by supervised learning from a database of human professional games. The network correctly predicts the expert move in 55% of positions, equalling the accuracy of a 6 dan human player. When the trained convolutional network was used directly to play games of Go, without any search, it beat the traditional search program GnuGo in 97% of games, and matched the performance of a state-of-the-art Monte-Carlo tree search that simulates a million positions per move.", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2015-05", "2015-07", "2015-11", "2016-01"], "last_mentioned": "2016-01-01T06:48:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.03044", "slug": "966418d4b002a121", "title": "Show, Attend and Tell: Neural Image Caption Generation with Visual Attention", "authors": ["Kelvin Xu", "Jimmy Ba", "Ryan Kiros", "Kyunghyun Cho", "Aaron Courville", "Ruslan Salakhutdinov", "Richard Zemel", "Yoshua Bengio"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.03044", "doi": null, "topics": ["Statistics & probability", "Machine learning", "Human–computer interaction"], "description": "Inspired by recent work in machine translation and object detection, we introduce an attention based model that automatically learns to describe the content of images.", "description_method": "abstract-excerpt", "abstract": "Inspired by recent work in machine translation and object detection, we introduce an attention based model that automatically learns to describe the content of images. We describe how we can train this model in a deterministic manner using standard backpropagation techniques and stochastically by maximizing a variational lower bound. We also show through visualization how the model is able to automatically learn to fix its gaze on salient objects while generating the corresponding words in the output sequence. We validate the use of attention with state-of-the-art performance on three benchmark datasets: Flickr8k, Flickr30k and MS COCO.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2015-08", "2015-10", "2016-03", "2016-10", "2017-02", "2018-02"], "last_mentioned": "2018-02-03T08:52:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.03808", "slug": "34215f118c24deb2", "title": "Gravitational Lensing by Spinning Black Holes in Astrophysics, and in the Movie Interstellar", "authors": ["Oliver James", "Eugenie von Tunzelmann", "Paul Franklin", "Kip S. Thorne"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.03808", "doi": "10.1088/0264-9381/32/6/065001", "topics": ["Physics"], "description": "Interstellar is the first Hollywood movie to attempt depicting a black hole as it would actually be seen by somebody nearby.", "description_method": "abstract-excerpt", "abstract": "Interstellar is the first Hollywood movie to attempt depicting a black hole as it would actually be seen by somebody nearby. For this we developed a code called DNGR (Double Negative Gravitational Renderer) to solve the equations for ray-bundle (light-beam) propagation through the curved spacetime of a spinning (Kerr) black hole, and to render IMAX-quality, rapidly changing images. Our ray-bundle techniques were crucial for achieving IMAX-quality smoothness without flickering. This paper has four purposes: (i) To describe DNGR for physicists and CGI practitioners . (ii) To present the equations we use, when the camera is in arbitrary motion at an arbitrary location near a Kerr black hole, for mapping light sources to camera images via elliptical ray bundles. (iii) To describe new insights, from DNGR, into gravitational lensing when the camera is near the spinning black hole, rather than far away as in almost all prior studies. (iv) To describe how the images of the black hole Gargantua and its accretion disk, in the movie \\emph{Interstellar}, were generated with DNGR. There are no new astrophysical insights in this accretion-disk section of the paper, but disk novices may find it pedagogically interesting, and movie buffs may find its discussions of Interstellar interesting.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2018-09", "2019-03", "2019-04", "2024-08", "2024-11", "2026-05"], "last_mentioned": "2026-05-21T08:26:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1503.03832", "slug": "3542d04db6a22f4b", "title": "FaceNet: A Unified Embedding for Face Recognition and Clustering", "authors": ["Florian Schroff", "Dmitry Kalenichenko", "James Philbin"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1503.03832", "doi": "10.1109/CVPR.2015.7298682", "topics": ["Psychology", "Machine learning", "Computer vision"], "description": "Despite significant recent advances in the field of face recognition, implementing face verification and recognition efficiently at scale presents serious challenges to current approaches.", "description_method": "abstract-excerpt", "abstract": "Despite significant recent advances in the field of face recognition, implementing face verification and recognition efficiently at scale presents serious challenges to current approaches. In this paper we present a system, called FaceNet, that directly learns a mapping from face images to a compact Euclidean space where distances directly correspond to a measure of face similarity. Once this space has been produced, tasks such as face recognition, verification and clustering can be easily implemented using standard techniques with FaceNet embeddings as feature vectors. Our method uses a deep convolutional network trained to directly optimize the embedding itself, rather than an intermediate bottleneck layer as in previous deep learning approaches. To train, we use triplets of roughly aligned matching / non-matching face patches generated using a novel online triplet mining method. The benefit of our approach is much greater representational efficiency: we achieve state-of-the-art face recognition performance using only 128-bytes per face. On the widely used Labeled Faces in the Wild (LFW) dataset, our system achieves a new record accuracy of 99.63%. On YouTube Faces DB it achieves 95.12%. Our system cuts the error rate in comparison to the best published result by 30% on both datasets. We also introduce the concept of harmonic embeddings, and a harmonic triplet loss, which describe different versions of face embeddings (produced by different networks) that are compatible to each other and allow for direct comparison between each other.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2016-04", "2016-05", "2017-05", "2017-06", "2018-08", "2019-07"], "last_mentioned": "2019-07-07T22:33:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1506.02142", "slug": "e8d528bac8cbe996", "title": "Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning", "authors": ["Yarin Gal", "Zoubin Ghahramani"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1506.02142", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Mathematics"], "description": "Deep learning tools have gained tremendous attention in applied machine learning.", "description_method": "abstract-excerpt", "abstract": "Deep learning tools have gained tremendous attention in applied machine learning. However such tools for regression and classification do not capture model uncertainty. In comparison, Bayesian models offer a mathematically grounded framework to reason about model uncertainty, but usually come with a prohibitive computational cost. In this paper we develop a new theoretical framework casting dropout training in deep neural networks (NNs) as approximate Bayesian inference in deep Gaussian processes. A direct result of this theory gives us tools to model uncertainty with dropout NNs -- extracting information from existing models that has been thrown away so far. This mitigates the problem of representing uncertainty in deep learning without sacrificing either computational complexity or test accuracy. We perform an extensive study of the properties of dropout's uncertainty. Various network architectures and non-linearities are assessed on tasks of regression and classification, using MNIST as an example. We show a considerable improvement in predictive log-likelihood and RMSE compared to existing state-of-the-art methods, and finish by using dropout's uncertainty in deep reinforcement learning.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2015-09", "2015-12", "2017-01", "2017-11", "2018-07", "2023-01"], "last_mentioned": "2023-01-15T21:05:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1601.06759", "slug": "f72d9a2af92f6133", "title": "Pixel Recurrent Neural Networks", "authors": ["Aaron van den Oord", "Nal Kalchbrenner", "Koray Kavukcuoglu"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1601.06759", "doi": null, "topics": ["Machine learning", "Statistics & probability"], "description": "Modeling the distribution of natural images is a landmark problem in unsupervised learning.", "description_method": "abstract-excerpt", "abstract": "Modeling the distribution of natural images is a landmark problem in unsupervised learning. This task requires an image model that is at once expressive, tractable and scalable. We present a deep neural network that sequentially predicts the pixels in an image along the two spatial dimensions. Our method models the discrete probability of the raw pixel values and encodes the complete set of dependencies in the image. Architectural novelties include fast two-dimensional recurrent layers and an effective use of residual connections in deep recurrent networks. We achieve log-likelihood scores on natural images that are considerably better than the previous state of the art. Our main results also provide benchmarks on the diverse ImageNet dataset. Samples generated from the model appear crisp, varied and globally coherent.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2016-06", "2018-10", "2020-02", "2020-07", "2024-10", "2025-10"], "last_mentioned": "2025-10-21T19:40:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1602.02697", "slug": "3a2568e2a5bdf07e", "title": "Practical Black-Box Attacks against Machine Learning", "authors": ["Nicolas Papernot", "Patrick McDaniel", "Ian Goodfellow", "Somesh Jha", "Z. Berkay Celik", "Ananthram Swami"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1602.02697", "doi": null, "topics": ["Security & cryptography", "Machine learning", "Statistics & probability"], "description": "Machine learning (ML) models, e.g., deep neural networks (DNNs), are vulnerable to adversarial examples: malicious inputs modified to yield erroneous model outputs, while appearing unmodified to human observers.", "description_method": "abstract-excerpt", "abstract": "Machine learning (ML) models, e.g., deep neural networks (DNNs), are vulnerable to adversarial examples: malicious inputs modified to yield erroneous model outputs, while appearing unmodified to human observers. Potential attacks include having malicious content like malware identified as legitimate or controlling vehicle behavior. Yet, all existing adversarial example attacks require knowledge of either the model internals or its training data. We introduce the first practical demonstration of an attacker controlling a remotely hosted DNN with no such knowledge. Indeed, the only capability of our black-box adversary is to observe labels given by the DNN to chosen inputs. Our attack strategy consists in training a local model to substitute for the target DNN, using inputs synthetically generated by an adversary and labeled by the target DNN. We use the local substitute to craft adversarial examples, and find that they are misclassified by the targeted DNN. To perform a real-world and properly-blinded evaluation, we attack a DNN hosted by MetaMind, an online deep learning API. We find that their DNN misclassifies 84.24% of the adversarial examples crafted with our substitute. We demonstrate the general applicability of our strategy to many ML techniques by conducting the same attack against models hosted by Amazon and Google, using logistic regression substitutes. They yield adversarial examples misclassified by Amazon and Google at rates of 96.19% and 88.94%. We also find that this black-box attack strategy is capable of evading defense strategies previously found to make adversarial example crafting harder.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2016-04", "2017-08", "2017-11", "2018-01", "2018-03", "2019-10"], "last_mentioned": "2019-10-28T21:40:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1606.03657", "slug": "e928718b75c8137e", "title": "InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets", "authors": ["Xi Chen", "Yan Duan", "Rein Houthooft", "John Schulman", "Ilya Sutskever", "Pieter Abbeel"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1606.03657", "doi": null, "topics": ["Algorithms"], "description": "This paper describes InfoGAN, an information-theoretic extension to the Generative Adversarial Network that is able to learn disentangled representations in a completely unsupervised manner.", "description_method": "abstract-excerpt", "abstract": "This paper describes InfoGAN, an information-theoretic extension to the Generative Adversarial Network that is able to learn disentangled representations in a completely unsupervised manner. InfoGAN is a generative adversarial network that also maximizes the mutual information between a small subset of the latent variables and the observation. We derive a lower bound to the mutual information objective that can be optimized efficiently, and show that our training procedure can be interpreted as a variation of the Wake-Sleep algorithm. Specifically, InfoGAN successfully disentangles writing styles from digit shapes on the MNIST dataset, pose from lighting of 3D rendered images, and background digits from the central digit on the SVHN dataset. It also discovers visual concepts that include hair styles, presence/absence of eyeglasses, and emotions on the CelebA face dataset. Experiments show that InfoGAN learns interpretable representations that are competitive with representations learned by existing fully supervised methods.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2016-06", "2016-08", "2017-01", "2017-02", "2017-04", "2017-09"], "last_mentioned": "2017-09-19T16:03:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1607.01341", "slug": "f526d18e369d0c69", "title": "Algorand", "authors": ["Jing Chen", "Silvio Micali"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1607.01341", "doi": null, "topics": ["Databases", "Statistics & probability", "Distributed systems"], "description": "A public ledger is a tamperproof sequence of data that can be read and augmented by everyone.", "description_method": "abstract-excerpt", "abstract": "A public ledger is a tamperproof sequence of data that can be read and augmented by everyone. Public ledgers have innumerable and compelling uses. They can secure, in plain sight, all kinds of transactions ---such as titles, sales, and payments--- in the exact order in which they occur. Public ledgers not only curb corruption, but also enable very sophisticated applications ---such as cryptocurrencies and smart contracts. They stand to revolutionize the way a democratic society operates. As currently implemented, however, they scale poorly and cannot achieve their potential. Algorand is a truly democratic and efficient way to implement a public ledger. Unlike prior implementations based on proof of work, it requires a negligible amount of computation, and generates a transaction history that will not \"fork\" with overwhelmingly high probability. Algorand is based on (a novel and super fast) message-passing Byzantine agreement. For concreteness, we shall describe Algorand only as a money platform.", "score": 0.0, "mention_count": 6, "user_count": 4, "months": ["2017-04", "2017-07", "2018-01", "2020-10"], "last_mentioned": "2020-10-18T16:30:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.04836", "slug": "064eda6fe313c36f", "title": "On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima", "authors": ["Nitish Shirish Keskar", "Dheevatsa Mudigere", "Jorge Nocedal", "Mikhail Smelyanskiy", "Ping Tak Peter Tang"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.04836", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Algorithms"], "description": "The stochastic gradient descent (SGD) method and its variants are algorithms of choice for many Deep Learning tasks.", "description_method": "abstract-excerpt", "abstract": "The stochastic gradient descent (SGD) method and its variants are algorithms of choice for many Deep Learning tasks. These methods operate in a small-batch regime wherein a fraction of the training data, say $32$-$512$ data points, is sampled to compute an approximation to the gradient. It has been observed in practice that when using a larger batch there is a degradation in the quality of the model, as measured by its ability to generalize. We investigate the cause for this generalization drop in the large-batch regime and present numerical evidence that supports the view that large-batch methods tend to converge to sharp minimizers of the training and testing functions - and as is well known, sharp minima lead to poorer generalization. In contrast, small-batch methods consistently converge to flat minimizers, and our experiments support a commonly held view that this is due to the inherent noise in the gradient estimation. We discuss several strategies to attempt to help large-batch methods eliminate this generalization gap.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2016-11", "2017-03", "2017-06", "2017-12", "2025-11"], "last_mentioned": "2025-11-02T07:51:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1610.02415", "slug": "14f0cd856cf103fd", "title": "Automatic chemical design using a data-driven continuous representation of molecules", "authors": ["Rafael Gómez-Bombarelli", "Jennifer N. Wei", "David Duvenaud", "José Miguel Hernández-Lobato", "Benjamín Sánchez-Lengeling", "Dennis Sheberla", "Jorge Aguilera-Iparraguirre", "Timothy D. Hirzel", "Ryan P. Adams", "Alán Aspuru-Guzik"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1610.02415", "doi": "10.1021/acscentsci.7b00572", "topics": ["Machine learning", "Algorithms", "Biology & medicine"], "description": "We report a method to convert discrete representations of molecules to and from a multidimensional continuous representation.", "description_method": "abstract-excerpt", "abstract": "We report a method to convert discrete representations of molecules to and from a multidimensional continuous representation. This model allows us to generate new molecules for efficient exploration and optimization through open-ended spaces of chemical compounds. A deep neural network was trained on hundreds of thousands of existing chemical structures to construct three coupled functions: an encoder, a decoder and a predictor. The encoder converts the discrete representation of a molecule into a real-valued continuous vector, and the decoder converts these continuous vectors back to discrete molecular representations. The predictor estimates chemical properties from the latent continuous vector representation of the molecule. Continuous representations allow us to automatically generate novel chemical structures by performing simple operations in the latent space, such as decoding random vectors, perturbing known chemical structures, or interpolating between molecules. Continuous representations also allow the use of powerful gradient-based optimization to efficiently guide the search for optimized functional compounds. We demonstrate our method in the domain of drug-like molecules and also in the set of molecules with fewer that nine heavy atoms.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2016-10", "2017-01", "2021-06", "2024-07", "2025-11"], "last_mentioned": "2025-11-05T23:31:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1612.08242", "slug": "73689bc075659c9e", "title": "YOLO9000: Better, Faster, Stronger", "authors": ["Joseph Redmon", "Ali Farhadi"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1612.08242", "doi": null, "topics": ["Computer vision"], "description": "We introduce YOLO9000, a state-of-the-art, real-time object detection system that can detect over 9000 object categories.", "description_method": "abstract-excerpt", "abstract": "We introduce YOLO9000, a state-of-the-art, real-time object detection system that can detect over 9000 object categories. First we propose various improvements to the YOLO detection method, both novel and drawn from prior work. The improved model, YOLOv2, is state-of-the-art on standard detection tasks like PASCAL VOC and COCO. At 67 FPS, YOLOv2 gets 76.8 mAP on VOC 2007. At 40 FPS, YOLOv2 gets 78.6 mAP, outperforming state-of-the-art methods like Faster RCNN with ResNet and SSD while still running significantly faster. Finally we propose a method to jointly train on object detection and classification. Using this method we train YOLO9000 simultaneously on the COCO detection dataset and the ImageNet classification dataset. Our joint training allows YOLO9000 to predict detections for object classes that don't have labelled detection data. We validate our approach on the ImageNet detection task. YOLO9000 gets 19.7 mAP on the ImageNet detection validation set despite only having detection data for 44 of the 200 classes. On the 156 classes not in COCO, YOLO9000 gets 16.0 mAP. But YOLO can detect more than just 200 classes; it predicts detections for more than 9000 different object categories. And it still runs in real-time.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2017-06", "2017-07", "2017-08", "2017-11", "2020-03"], "last_mentioned": "2020-03-20T12:56:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1703.02874", "slug": "d9118c3a6b42d6c0", "title": "A Study of MAC Address Randomization in Mobile Devices and When it Fails", "authors": ["Jeremy Martin", "Travis Mayberry", "Collin Donahue", "Lucas Foppe", "Lamont Brown", "Chadwick Riggins", "Erik C. Rye", "Dane Brown"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1703.02874", "doi": null, "topics": ["Security & cryptography", "Computer systems"], "description": "MAC address randomization is a privacy technique whereby mobile devices rotate through random hardware addresses in order to prevent observers from singling out their traffic or physical location from other nearby devices.", "description_method": "abstract-excerpt", "abstract": "MAC address randomization is a privacy technique whereby mobile devices rotate through random hardware addresses in order to prevent observers from singling out their traffic or physical location from other nearby devices. Adoption of this technology, however, has been sporadic and varied across device manufacturers. In this paper, we present the first wide-scale study of MAC address randomization in the wild, including a detailed breakdown of different randomization techniques by operating system, manufacturer, and model of device. We then identify multiple flaws in these implementations which can be exploited to defeat randomization as performed by existing devices. First, we show that devices commonly make improper use of randomization by sending wireless frames with the true, global address when they should be using a randomized address. We move on to extend the passive identification techniques of Vanhoef et al. to effectively defeat randomization in ~96% of Android phones. Finally, we show a method that can be used to track 100% of devices using randomization, regardless of manufacturer, by exploiting a previously unknown flaw in the way existing wireless chipsets handle low-level control frames.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2017-08", "2017-09", "2017-10", "2017-12", "2018-11", "2019-01"], "last_mentioned": "2019-01-17T19:17:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1711.04422", "slug": "b745ffadf2fe6bd7", "title": "Souper: A Synthesizing Superoptimizer", "authors": ["Raimondas Sasnauskas", "Yang Chen", "Peter Collingbourne", "Jeroen Ketema", "Gratian Lup", "Jubi Taneja", "John Regehr"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1711.04422", "doi": null, "topics": ["Programming languages", "Algorithms"], "description": "If we can automatically derive compiler optimizations, we might be able to sidestep some of the substantial engineering challenges involved in creating and maintaining a high-quality compiler.", "description_method": "abstract-excerpt", "abstract": "If we can automatically derive compiler optimizations, we might be able to sidestep some of the substantial engineering challenges involved in creating and maintaining a high-quality compiler. We developed Souper, a synthesizing superoptimizer, to see how far these ideas might be pushed in the context of LLVM. Along the way, we discovered that Souper's intermediate representation was sufficiently similar to the one in Microsoft Visual C++ that we applied Souper to that compiler as well. Shipping, or about-to-ship, versions of both compilers contain optimizations suggested by Souper but implemented by hand. Alternately, when Souper is used as a fully automated optimization pass it compiles a Clang compiler binary that is about 3 MB (4.4%) smaller than the one compiled by LLVM.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2017-12", "2019-02", "2021-02", "2025-01", "2025-07"], "last_mentioned": "2025-07-21T13:22:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1801.00862", "slug": "bad43a2713a36f91", "title": "Quantum Computing in the NISQ era and beyond", "authors": ["John Preskill"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1801.00862", "doi": "10.22331/q-2018-08-06-79", "topics": ["Physics", "Distributed systems"], "description": "Noisy Intermediate-Scale Quantum (NISQ) technology will be available in the near future.", "description_method": "abstract-excerpt", "abstract": "Noisy Intermediate-Scale Quantum (NISQ) technology will be available in the near future. Quantum computers with 50-100 qubits may be able to perform tasks which surpass the capabilities of today's classical digital computers, but noise in quantum gates will limit the size of quantum circuits that can be executed reliably. NISQ devices will be useful tools for exploring many-body quantum physics, and may have other useful applications, but the 100-qubit quantum computer will not change the world right away --- we should regard it as a significant step toward the more powerful quantum technologies of the future. Quantum technologists should continue to strive for more accurate quantum gates and, eventually, fully fault-tolerant quantum computing.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2018-07", "2018-09", "2019-05", "2019-09", "2020-12"], "last_mentioned": "2020-12-21T15:46:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.10122", "slug": "c3ef378f3c6bf874", "title": "World Models", "authors": ["David Ha", "Jürgen Schmidhuber"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.10122", "doi": "10.5281/zenodo.1207631", "topics": ["Machine learning"], "description": "We explore building generative neural network models of popular reinforcement learning environments.", "description_method": "abstract-excerpt", "abstract": "We explore building generative neural network models of popular reinforcement learning environments. Our world model can be trained quickly in an unsupervised manner to learn a compressed spatial and temporal representation of the environment. By using features extracted from the world model as inputs to an agent, we can train a very compact and simple policy that can solve the required task. We can even train our agent entirely inside of its own hallucinated dream generated by its world model, and transfer this policy back into the actual environment. An interactive version of this paper is available at https://worldmodels.github.io/", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2018-07", "2021-06", "2021-07", "2023-02", "2025-09", "2026-02"], "last_mentioned": "2026-02-06T18:33:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1809.05923", "slug": "f5b2a552aea79f84", "title": "What is Applied Category Theory?", "authors": ["Tai-Danae Bradley"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1809.05923", "doi": null, "topics": ["Mathematics"], "description": "This is a collection of introductory, expository notes on applied category theory, inspired by the 2018 Applied Category Theory Workshop, and in these notes we take a leisurely stroll through two themes (functorial semantics and compositionality), two constructions (monoidal categories and decorated cospans) and two examples (chemical rea", "description_method": "abstract-excerpt", "abstract": "This is a collection of introductory, expository notes on applied category theory, inspired by the 2018 Applied Category Theory Workshop, and in these notes we take a leisurely stroll through two themes (functorial semantics and compositionality), two constructions (monoidal categories and decorated cospans) and two examples (chemical reaction networks and natural language processing) within the field.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2019-04", "2019-09", "2019-12", "2020-04", "2020-05", "2021-05"], "last_mentioned": "2021-05-22T23:57:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1901.02860", "slug": "6bfd4c6b533577a6", "title": "Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context", "authors": ["Zihang Dai", "Zhilin Yang", "Yiming Yang", "Jaime Carbonell", "Quoc V. Le", "Ruslan Salakhutdinov"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1901.02860", "doi": null, "topics": ["Language models"], "description": "Transformers have a potential of learning longer-term dependency, but are limited by a fixed-length context in the setting of language modeling.", "description_method": "abstract-excerpt", "abstract": "Transformers have a potential of learning longer-term dependency, but are limited by a fixed-length context in the setting of language modeling. We propose a novel neural architecture Transformer-XL that enables learning dependency beyond a fixed length without disrupting temporal coherence. It consists of a segment-level recurrence mechanism and a novel positional encoding scheme. Our method not only enables capturing longer-term dependency, but also resolves the context fragmentation problem. As a result, Transformer-XL learns dependency that is 80% longer than RNNs and 450% longer than vanilla Transformers, achieves better performance on both short and long sequences, and is up to 1,800+ times faster than vanilla Transformers during evaluation. Notably, we improve the state-of-the-art results of bpc/perplexity to 0.99 on enwiki8, 1.08 on text8, 18.3 on WikiText-103, 21.8 on One Billion Word, and 54.5 on Penn Treebank (without finetuning). When trained only on WikiText-103, Transformer-XL manages to generate reasonably coherent, novel text articles with thousands of tokens. Our code, pretrained models, and hyperparameters are available in both Tensorflow and PyTorch.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2019-09", "2019-11", "2019-12", "2020-08", "2023-01", "2023-02"], "last_mentioned": "2023-02-08T16:19:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1904.05234", "slug": "ee48f6c3c71669f0", "title": "Flash Boys 2.0: Frontrunning, Transaction Reordering, and Consensus Instability in Decentralized Exchanges", "authors": ["Philip Daian", "Steven Goldfeder", "Tyler Kell", "Yunqi Li", "Xueyuan Zhao", "Iddo Bentov", "Lorenz Breidenbach", "Ari Juels"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1904.05234", "doi": null, "topics": ["Databases", "Distributed systems", "Economics"], "description": "Blockchains, and specifically smart contracts, have promised to create fair and transparent trading ecosystems.", "description_method": "abstract-excerpt", "abstract": "Blockchains, and specifically smart contracts, have promised to create fair and transparent trading ecosystems. Unfortunately, we show that this promise has not been met. We document and quantify the widespread and rising deployment of arbitrage bots in blockchain systems, specifically in decentralized exchanges (or \"DEXes\"). Like high-frequency traders on Wall Street, these bots exploit inefficiencies in DEXes, paying high transaction fees and optimizing network latency to frontrun, i.e., anticipate and exploit, ordinary users' DEX trades. We study the breadth of DEX arbitrage bots in a subset of transactions that yield quantifiable revenue to these bots. We also study bots' profit-making strategies, with a focus on blockchain-specific elements. We observe bots engage in what we call priority gas auctions (PGAs), competitively bidding up transaction fees in order to obtain priority ordering, i.e., early block position and execution, for their transactions. PGAs present an interesting and complex new continuous-time, partial-information, game-theoretic model that we formalize and study. We release an interactive web portal, http://frontrun.me/, to provide the community with real-time data on PGAs. We additionally show that high fees paid for priority transaction ordering poses a systemic risk to consensus-layer security. We explain that such fees are just one form of a general phenomenon in DEXes and beyond---what we call miner extractable value (MEV)---that poses concrete, measurable, consensus-layer security risks. We show empirically that MEV poses a realistic threat to Ethereum today. Our work highlights the large, complex risks created by transaction-ordering dependencies in smart contracts and the ways in which traditional forms of financial-market exploitation are adapting to and penetrating blockchain economies.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2019-04", "2019-11", "2021-01", "2021-02", "2022-09"], "last_mentioned": "2022-09-17T01:29:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1909.08593", "slug": "5bab129b08b82900", "title": "Fine-Tuning Language Models from Human Preferences", "authors": ["Daniel M. Ziegler", "Nisan Stiennon", "Jeffrey Wu", "Tom B. Brown", "Alec Radford", "Dario Amodei", "Paul Christiano", "Geoffrey Irving"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1909.08593", "doi": null, "topics": ["Language models", "Machine learning"], "description": "Reward learning enables the application of reinforcement learning (RL) to tasks where reward is defined by human judgment, building a model of reward by asking humans questions.", "description_method": "abstract-excerpt", "abstract": "Reward learning enables the application of reinforcement learning (RL) to tasks where reward is defined by human judgment, building a model of reward by asking humans questions. Most work on reward learning has used simulated environments, but complex information about values is often expressed in natural language, and we believe reward learning for language is a key to making RL practical and safe for real-world tasks. In this paper, we build on advances in generative pretraining of language models to apply reward learning to four natural language tasks: continuing text with positive sentiment or physically descriptive language, and summarization tasks on the TL;DR and CNN/Daily Mail datasets. For stylistic continuation we achieve good results with only 5,000 comparisons evaluated by humans. For summarization, models trained with 60,000 comparisons copy whole sentences from the input but skip irrelevant preamble; this leads to reasonable ROUGE scores and very good performance according to our human labelers, but may be exploiting the fact that labelers rely on simple heuristics.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2019-12", "2023-03", "2023-09", "2024-04", "2025-08"], "last_mentioned": "2025-08-16T10:53:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1912.02292", "slug": "eea710cdf677bea3", "title": "Deep Double Descent: Where Bigger Models and More Data Hurt", "authors": ["Preetum Nakkiran", "Gal Kaplun", "Yamini Bansal", "Tristan Yang", "Boaz Barak", "Ilya Sutskever"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1912.02292", "doi": null, "topics": ["Algorithms", "Machine learning"], "description": "We show that a variety of modern deep learning tasks exhibit a \"double-descent\" phenomenon where, as we increase model size, performance first gets worse and then gets better.", "description_method": "abstract-excerpt", "abstract": "We show that a variety of modern deep learning tasks exhibit a \"double-descent\" phenomenon where, as we increase model size, performance first gets worse and then gets better. Moreover, we show that double descent occurs not just as a function of model size, but also as a function of the number of training epochs. We unify the above phenomena by defining a new complexity measure we call the effective model complexity and conjecture a generalized double descent with respect to this measure. Furthermore, our notion of model complexity allows us to identify certain regimes where increasing (even quadrupling) the number of train samples actually hurts test performance.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2019-12", "2020-07", "2021-10", "2023-04", "2026-02", "2026-05"], "last_mentioned": "2026-05-23T12:24:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1912.08258", "slug": "de8daa55415140cb", "title": "Xor Filters: Faster and Smaller Than Bloom and Cuckoo Filters", "authors": ["Thomas Mueller Graf", "Daniel Lemire"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1912.08258", "doi": "10.1145/3376122", "topics": ["Machine learning"], "description": "The Bloom filter provides fast approximate set membership while using little memory.", "description_method": "abstract-excerpt", "abstract": "The Bloom filter provides fast approximate set membership while using little memory. Engineers often use these filters to avoid slow operations such as disk or network accesses. As an alternative, a cuckoo filter may need less space than a Bloom filter and it is faster. Chazelle et al. proposed a generalization of the Bloom filter called the Bloomier filter. Dietzfelbinger and Pagh described a variation on the Bloomier filter that can be used effectively for approximate membership queries. It has never been tested empirically, to our knowledge. We review an efficient implementation of their approach, which we call the xor filter. We find that xor filters can be faster than Bloom and cuckoo filters while using less memory. We further show that a more compact version of xor filters (xor+) can use even less space than highly compact alternatives (e.g., Golomb-compressed sequences) while providing speeds competitive with Bloom filters.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2019-12", "2020-07", "2021-08", "2025-05", "2025-07"], "last_mentioned": "2025-07-02T18:38:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2001.00888", "slug": "27ad3e45004f34f2", "title": "Towards Scalable Dataframe Systems", "authors": ["Devin Petersohn", "Stephen Macke", "Doris Xin", "William Ma", "Doris Lee", "Xiangxi Mo", "Joseph E. Gonzalez", "Joseph M. Hellerstein", "Anthony D. Joseph", "Aditya Parameswaran"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2001.00888", "doi": null, "topics": ["Mathematics", "Databases"], "description": "Dataframes are a popular abstraction to represent, prepare, and analyze data.", "description_method": "abstract-excerpt", "abstract": "Dataframes are a popular abstraction to represent, prepare, and analyze data. Despite the remarkable success of dataframe libraries in Rand Python, dataframes face performance issues even on moderately large datasets. Moreover, there is significant ambiguity regarding dataframe semantics. In this paper we lay out a vision and roadmap for scalable dataframe systems. To demonstrate the potential in this area, we report on our experience building MODIN, a scaled-up implementation of the most widely-used and complex dataframe API today, Python's pandas. With pandas as a reference, we propose a simple data model and algebra for dataframes to ground discussion in the field. Given this foundation, we lay out an agenda of open research opportunities where the distinct features of dataframes will require extending the state of the art in many dimensions of data management. We discuss the implications of signature data-frame features including flexible schemas, ordering, row/column equivalence, and data/metadata fluidity, as well as the piecemeal, trial-and-error-based approach to interacting with dataframes.", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2021-03", "2021-12", "2022-02", "2022-03", "2022-06", "2025-09"], "last_mentioned": "2025-09-11T13:13:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2003.03296", "slug": "35319e77620e160d", "title": "Memory-Safety Challenge Considered Solved? An In-Depth Study with All Rust CVEs", "authors": ["Hui Xu", "Zhuangbin Chen", "Mingshen Sun", "Yangfan Zhou", "Michael Lyu"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2003.03296", "doi": null, "topics": ["Security & cryptography"], "description": "Rust is an emerging programing language that aims at preventing memory-safety bugs without sacrificing much efficiency.", "description_method": "abstract-excerpt", "abstract": "Rust is an emerging programing language that aims at preventing memory-safety bugs without sacrificing much efficiency. The claimed property is very attractive to developers, and many projects start using the language. However, can Rust achieve the memory-safety promise? This paper studies the question by surveying 186 real-world bug reports collected from several origins which contain all existing Rust CVEs (common vulnerability and exposures) of memory-safety issues by 2020-12-31. We manually analyze each bug and extract their culprit patterns. Our analysis result shows that Rust can keep its promise that all memory-safety bugs require unsafe code, and many memory-safety bugs in our dataset are mild soundness issues that only leave a possibility to write memory-safety bugs without unsafe code. Furthermore, we summarize three typical categories of memory-safety bugs, including automatic memory reclaim, unsound function, and unsound generic or trait. While automatic memory claim bugs are related to the side effect of Rust newly-adopted ownership-based resource management scheme, unsound function reveals the essential challenge of Rust development for avoiding unsound code, and unsound generic or trait intensifies the risk of introducing unsoundness. Based on these findings, we propose two promising directions towards improving the security of Rust development, including several best practices of using specific APIs and methods to detect particular bugs involving unsafe code. Our work intends to raise more discussions regarding the memory-safety issues of Rust and facilitate the maturity of the language.", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2020-06", "2020-07", "2020-08", "2021-02", "2023-01", "2024-07"], "last_mentioned": "2024-07-16T23:13:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2003.03808", "slug": "0d43166b5bffbd63", "title": "PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models", "authors": ["Sachit Menon", "Alexandru Damian", "Shijia Hu", "Nikhil Ravi", "Cynthia Rudin"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2003.03808", "doi": null, "topics": ["Machine learning", "Psychology", "Databases"], "description": "The primary aim of single-image super-resolution is to construct high-resolution (HR) images from corresponding low-resolution (LR) inputs.", "description_method": "abstract-excerpt", "abstract": "The primary aim of single-image super-resolution is to construct high-resolution (HR) images from corresponding low-resolution (LR) inputs. In previous approaches, which have generally been supervised, the training objective typically measures a pixel-wise average distance between the super-resolved (SR) and HR images. Optimizing such metrics often leads to blurring, especially in high variance (detailed) regions. We propose an alternative formulation of the super-resolution problem based on creating realistic SR images that downscale correctly. We present an algorithm addressing this problem, PULSE (Photo Upsampling via Latent Space Exploration), which generates high-resolution, realistic images at resolutions previously unseen in the literature. It accomplishes this in an entirely self-supervised fashion and is not confined to a specific degradation operator used during training, unlike previous methods (which require supervised training on databases of LR-HR image pairs). Instead of starting with the LR image and slowly adding detail, PULSE traverses the high-resolution natural image manifold, searching for images that downscale to the original LR image. This is formalized through the \"downscaling loss,\" which guides exploration through the latent space of a generative model. By leveraging properties of high-dimensional Gaussians, we restrict the search space to guarantee realistic outputs. PULSE thereby generates super-resolved images that both are realistic and downscale correctly. We show proof of concept of our approach in the domain of face super-resolution (i.e., face hallucination). We also present a discussion of the limitations and biases of the method as currently implemented with an accompanying model card with relevant metrics. Our method outperforms state-of-the-art methods in perceptual quality at higher resolutions and scale factors than previously possible.", "score": 0.0, "mention_count": 6, "user_count": 4, "months": ["2020-06", "2020-07", "2022-02", "2022-05", "2022-12"], "last_mentioned": "2022-12-22T04:10:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2009.03393", "slug": "09da6d06fde77581", "title": "Generative Language Modeling for Automated Theorem Proving", "authors": ["Stanislas Polu", "Ilya Sutskever"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2009.03393", "doi": null, "topics": ["Mathematics", "Language models"], "description": "We explore the application of transformer-based language models to automated theorem proving.", "description_method": "abstract-excerpt", "abstract": "We explore the application of transformer-based language models to automated theorem proving. This work is motivated by the possibility that a major limitation of automated theorem provers compared to humans -- the generation of original mathematical terms -- might be addressable via generation from language models. We present an automated prover and proof assistant, GPT-f, for the Metamath formalization language, and analyze its performance. GPT-f found new short proofs that were accepted into the main Metamath library, which is to our knowledge, the first time a deep-learning based system has contributed proofs that were adopted by a formal mathematics community.", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2020-09", "2020-11", "2021-01", "2021-06", "2022-06", "2023-10"], "last_mentioned": "2023-10-27T18:52:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2101.03961", "slug": "580e3810298dba5e", "title": "Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity", "authors": ["William Fedus", "Barret Zoph", "Noam Shazeer"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2101.03961", "doi": null, "topics": ["Language models", "Algorithms", "Networks"], "description": "In deep learning, models typically reuse the same parameters for all inputs.", "description_method": "abstract-excerpt", "abstract": "In deep learning, models typically reuse the same parameters for all inputs. Mixture of Experts (MoE) defies this and instead selects different parameters for each incoming example. The result is a sparsely-activated model -- with outrageous numbers of parameters -- but a constant computational cost. However, despite several notable successes of MoE, widespread adoption has been hindered by complexity, communication costs and training instability -- we address these with the Switch Transformer. We simplify the MoE routing algorithm and design intuitive improved models with reduced communication and computational costs. Our proposed training techniques help wrangle the instabilities and we show large sparse models may be trained, for the first time, with lower precision (bfloat16) formats. We design models based off T5-Base and T5-Large to obtain up to 7x increases in pre-training speed with the same computational resources. These improvements extend into multilingual settings where we measure gains over the mT5-Base version across all 101 languages. Finally, we advance the current scale of language models by pre-training up to trillion parameter models on the \"Colossal Clean Crawled Corpus\" and achieve a 4x speedup over the T5-XXL model.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2021-08", "2022-01", "2023-03", "2023-06", "2023-07", "2026-05"], "last_mentioned": "2026-05-01T10:21:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2102.11245", "slug": "846b2287b5d6634f", "title": "Silent Data Corruptions at Scale", "authors": ["Harish Dattatraya Dixit", "Sneha Pendharkar", "Matt Beadon", "Chris Mason", "Tejasvi Chakravarthy", "Bharath Muthiah", "Sriram Sankar"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2102.11245", "doi": null, "topics": ["Distributed systems"], "description": "Silent Data Corruption (SDC) can have negative impact on large-scale infrastructure services.", "description_method": "abstract-excerpt", "abstract": "Silent Data Corruption (SDC) can have negative impact on large-scale infrastructure services. SDCs are not captured by error reporting mechanisms within a Central Processing Unit (CPU) and hence are not traceable at the hardware level. However, the data corruptions propagate across the stack and manifest as application-level problems. These types of errors can result in data loss and can require months of debug engineering time. In this paper, we describe common defect types observed in silicon manufacturing that leads to SDCs. We discuss a real-world example of silent data corruption within a datacenter application. We provide the debug flow followed to root-cause and triage faulty instructions within a CPU using a case study, as an illustration on how to debug this class of errors. We provide a high-level overview of the mitigations to reduce the risk of silent data corruptions within a large production fleet. In our large-scale infrastructure, we have run a vast library of silent error test scenarios across hundreds of thousands of machines in our fleet. This has resulted in hundreds of CPUs detected for these errors, showing that SDCs are a systemic issue across generations. We have monitored SDCs for a period longer than 18 months. Based on this experience, we determine that reducing silent data corruptions requires not only hardware resiliency and production detection mechanisms, but also robust fault-tolerant software architectures.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2021-06", "2021-07", "2022-05", "2023-01", "2024-02", "2024-08"], "last_mentioned": "2024-08-05T09:57:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2105.13626", "slug": "7f2389a82e0530b4", "title": "ByT5: Towards a token-free future with pre-trained byte-to-byte models", "authors": ["Linting Xue", "Aditya Barua", "Noah Constant", "Rami Al-Rfou", "Sharan Narang", "Mihir Kale", "Adam Roberts", "Colin Raffel"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2105.13626", "doi": null, "topics": ["Language models"], "description": "Most widely-used pre-trained language models operate on sequences of tokens corresponding to word or subword units.", "description_method": "abstract-excerpt", "abstract": "Most widely-used pre-trained language models operate on sequences of tokens corresponding to word or subword units. By comparison, token-free models that operate directly on raw text (bytes or characters) have many benefits: they can process text in any language out of the box, they are more robust to noise, and they minimize technical debt by removing complex and error-prone text preprocessing pipelines. Since byte or character sequences are longer than token sequences, past work on token-free models has often introduced new model architectures designed to amortize the cost of operating directly on raw text. In this paper, we show that a standard Transformer architecture can be used with minimal modifications to process byte sequences. We characterize the trade-offs in terms of parameter count, training FLOPs, and inference speed, and show that byte-level models are competitive with their token-level counterparts. We also demonstrate that byte-level models are significantly more robust to noise and perform better on tasks that are sensitive to spelling and pronunciation. As part of our contribution, we release a new set of pre-trained byte-level Transformer models based on the T5 architecture, as well as all code and data used in our experiments.", "score": 0.0, "mention_count": 6, "user_count": 4, "months": ["2022-05", "2022-12", "2023-03", "2023-04", "2024-08", "2026-06"], "last_mentioned": "2026-06-20T08:47:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2106.01345", "slug": "e4e87e32bb1d9527", "title": "Decision Transformer: Reinforcement Learning via Sequence Modeling", "authors": ["Lili Chen", "Kevin Lu", "Aravind Rajeswaran", "Kimin Lee", "Aditya Grover", "Michael Laskin", "Pieter Abbeel", "Aravind Srinivas", "Igor Mordatch"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2106.01345", "doi": null, "topics": ["Language models", "Machine learning"], "description": "We introduce a framework that abstracts Reinforcement Learning (RL) as a sequence modeling problem.", "description_method": "abstract-excerpt", "abstract": "We introduce a framework that abstracts Reinforcement Learning (RL) as a sequence modeling problem. This allows us to draw upon the simplicity and scalability of the Transformer architecture, and associated advances in language modeling such as GPT-x and BERT. In particular, we present Decision Transformer, an architecture that casts the problem of RL as conditional sequence modeling. Unlike prior approaches to RL that fit value functions or compute policy gradients, Decision Transformer simply outputs the optimal actions by leveraging a causally masked Transformer. By conditioning an autoregressive model on the desired return (reward), past states, and actions, our Decision Transformer model can generate future actions that achieve the desired return. Despite its simplicity, Decision Transformer matches or exceeds the performance of state-of-the-art model-free offline RL baselines on Atari, OpenAI Gym, and Key-to-Door tasks.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2021-07", "2023-10", "2024-08", "2025-06", "2025-10"], "last_mentioned": "2025-10-05T21:46:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2107.06499", "slug": "29c77aac85d00533", "title": "Deduplicating Training Data Makes Language Models Better", "authors": ["Katherine Lee", "Daphne Ippolito", "Andrew Nystrom", "Chiyuan Zhang", "Douglas Eck", "Chris Callison-Burch", "Nicholas Carlini"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2107.06499", "doi": null, "topics": ["Language models"], "description": "We find that existing language modeling datasets contain many near-duplicate examples and long repetitive substrings.", "description_method": "abstract-excerpt", "abstract": "We find that existing language modeling datasets contain many near-duplicate examples and long repetitive substrings. As a result, over 1% of the unprompted output of language models trained on these datasets is copied verbatim from the training data. We develop two tools that allow us to deduplicate training datasets -- for example removing from C4 a single 61 word English sentence that is repeated over 60,000 times. Deduplication allows us to train models that emit memorized text ten times less frequently and require fewer train steps to achieve the same or better accuracy. We can also reduce train-test overlap, which affects over 4% of the validation set of standard datasets, thus allowing for more accurate evaluation. We release code for reproducing our work and performing dataset deduplication at https://github.com/google-research/deduplicate-text-datasets.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2022-05", "2022-08", "2022-11", "2024-06", "2025-01"], "last_mentioned": "2025-01-02T01:45:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2201.03545", "slug": "24d9a8f72099b856", "title": "A ConvNet for the 2020s", "authors": ["Zhuang Liu", "Hanzi Mao", "Chao-Yuan Wu", "Christoph Feichtenhofer", "Trevor Darrell", "Saining Xie"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2201.03545", "doi": null, "topics": ["Language models", "Computer vision", "Psychology"], "description": "The \"Roaring 20s\" of visual recognition began with the introduction of Vision Transformers (ViTs), which quickly superseded ConvNets as the state-of-the-art image classification model.", "description_method": "abstract-excerpt", "abstract": "The \"Roaring 20s\" of visual recognition began with the introduction of Vision Transformers (ViTs), which quickly superseded ConvNets as the state-of-the-art image classification model. A vanilla ViT, on the other hand, faces difficulties when applied to general computer vision tasks such as object detection and semantic segmentation. It is the hierarchical Transformers (e.g., Swin Transformers) that reintroduced several ConvNet priors, making Transformers practically viable as a generic vision backbone and demonstrating remarkable performance on a wide variety of vision tasks. However, the effectiveness of such hybrid approaches is still largely credited to the intrinsic superiority of Transformers, rather than the inherent inductive biases of convolutions. In this work, we reexamine the design spaces and test the limits of what a pure ConvNet can achieve. We gradually \"modernize\" a standard ResNet toward the design of a vision Transformer, and discover several key components that contribute to the performance difference along the way. The outcome of this exploration is a family of pure ConvNet models dubbed ConvNeXt. Constructed entirely from standard ConvNet modules, ConvNeXts compete favorably with Transformers in terms of accuracy and scalability, achieving 87.8% ImageNet top-1 accuracy and outperforming Swin Transformers on COCO detection and ADE20K segmentation, while maintaining the simplicity and efficiency of standard ConvNets.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2022-04", "2023-04", "2024-01", "2024-08"], "last_mentioned": "2024-08-24T14:58:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2202.07646", "slug": "a6d8884c3d434098", "title": "Quantifying Memorization Across Neural Language Models", "authors": ["Nicholas Carlini", "Daphne Ippolito", "Matthew Jagielski", "Katherine Lee", "Florian Tramer", "Chiyuan Zhang"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2202.07646", "doi": null, "topics": ["Language models", "Security & cryptography"], "description": "Large language models (LMs) have been shown to memorize parts of their training data, and when prompted appropriately, they will emit the memorized training data verbatim.", "description_method": "abstract-excerpt", "abstract": "Large language models (LMs) have been shown to memorize parts of their training data, and when prompted appropriately, they will emit the memorized training data verbatim. This is undesirable because memorization violates privacy (exposing user data), degrades utility (repeated easy-to-memorize text is often low quality), and hurts fairness (some texts are memorized over others). We describe three log-linear relationships that quantify the degree to which LMs emit memorized training data. Memorization significantly grows as we increase (1) the capacity of a model, (2) the number of times an example has been duplicated, and (3) the number of tokens of context used to prompt the model. Surprisingly, we find the situation becomes more complicated when generalizing these results across model families. On the whole, we find that memorization in LMs is more prevalent than previously believed and will likely get worse as models continues to scale, at least without active mitigations.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2022-08", "2022-11", "2023-03", "2024-08", "2025-04", "2026-07"], "last_mentioned": "2026-07-21T10:46:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2203.13131", "slug": "2a4785163afdaabd", "title": "Make-A-Scene: Scene-Based Text-to-Image Generation with Human Priors", "authors": ["Oran Gafni", "Adam Polyak", "Oron Ashual", "Shelly Sheynin", "Devi Parikh", "Yaniv Taigman"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2203.13131", "doi": null, "topics": ["Computer vision", "Machine learning", "Language models"], "description": "Recent text-to-image generation methods provide a simple yet exciting conversion capability between text and image domains.", "description_method": "abstract-excerpt", "abstract": "Recent text-to-image generation methods provide a simple yet exciting conversion capability between text and image domains. While these methods have incrementally improved the generated image fidelity and text relevancy, several pivotal gaps remain unanswered, limiting applicability and quality. We propose a novel text-to-image method that addresses these gaps by (i) enabling a simple control mechanism complementary to text in the form of a scene, (ii) introducing elements that substantially improve the tokenization process by employing domain-specific knowledge over key image regions (faces and salient objects), and (iii) adapting classifier-free guidance for the transformer use case. Our model achieves state-of-the-art FID and human evaluation results, unlocking the ability to generate high fidelity images in a resolution of 512x512 pixels, significantly improving visual quality. Through scene controllability, we introduce several new capabilities: (i) Scene editing, (ii) text editing with anchor scenes, (iii) overcoming out-of-distribution text prompts, and (iv) story illustration generation, as demonstrated in the story we wrote.", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2022-04", "2022-05", "2022-07"], "last_mentioned": "2022-07-14T15:17:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2205.06114", "slug": "8b1d31e3d2bbc259", "title": "Evil Never Sleeps: When Wireless Malware Stays On After Turning Off iPhones", "authors": ["Jiska Classen", "Alexander Heinrich", "Robert Reith", "Matthias Hollick"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2205.06114", "doi": "10.1145/3507657.3528547", "topics": ["Security & cryptography"], "description": "When an iPhone is turned off, most wireless chips stay on.", "description_method": "abstract-excerpt", "abstract": "When an iPhone is turned off, most wireless chips stay on. For instance, upon user-initiated shutdown, the iPhone remains locatable via the Find My network. If the battery runs low, the iPhone shuts down automatically and enters a power reserve mode. Yet, users can still access credit cards, student passes, and other items in their Wallet. We analyze how Apple implements these standalone wireless features, working while iOS is not running, and determine their security boundaries. On recent iPhones, Bluetooth, Near Field Communication (NFC), and Ultra-wideband (UWB) keep running after power off, and all three wireless chips have direct access to the secure element. As a practical example what this means to security, we demonstrate the possibility to load malware onto a Bluetooth chip that is executed while the iPhone is off.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2022-05", "2022-06", "2022-08", "2022-12", "2025-08"], "last_mentioned": "2025-08-25T11:38:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2207.00729", "slug": "c33029322af1bb7f", "title": "The Parallelism Tradeoff: Limitations of Log-Precision Transformers", "authors": ["William Merrill", "Ashish Sabharwal"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2207.00729", "doi": null, "topics": ["Language models", "Algorithms"], "description": "Despite their omnipresence in modern NLP, characterizing the computational power of transformer neural nets remains an interesting open question.", "description_method": "abstract-excerpt", "abstract": "Despite their omnipresence in modern NLP, characterizing the computational power of transformer neural nets remains an interesting open question. We prove that transformers whose arithmetic precision is logarithmic in the number of input tokens (and whose feedforward nets are computable using space linear in their input) can be simulated by constant-depth logspace-uniform threshold circuits. This provides insight on the power of transformers using known results in complexity theory. For example, if $\\mathsf L \\neq \\mathsf P$ (i.e., not all poly-time problems can be solved using logarithmic space), then transformers cannot even accurately solve linear equalities or check membership in an arbitrary context-free grammar with empty productions. Our result intuitively emerges from the transformer architecture's high parallelizability. We thus speculatively introduce the idea of a fundamental parallelism tradeoff: any model architecture as parallelizable as the transformer will obey limitations similar to it. Since parallelism is key to training models at massive scale, this suggests a potential inherent weakness of the scaling paradigm.", "score": 0.0, "mention_count": 6, "user_count": 4, "months": ["2023-05", "2024-03", "2024-06", "2024-09"], "last_mentioned": "2024-09-17T13:22:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2208.07339", "slug": "842dd0204d45dd71", "title": "LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale", "authors": ["Tim Dettmers", "Mike Lewis", "Younes Belkada", "Luke Zettlemoyer"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2208.07339", "doi": null, "topics": ["Language models"], "description": "Large language models have been widely adopted but require significant GPU memory for inference.", "description_method": "abstract-excerpt", "abstract": "Large language models have been widely adopted but require significant GPU memory for inference. We develop a procedure for Int8 matrix multiplication for feed-forward and attention projection layers in transformers, which cut the memory needed for inference by half while retaining full precision performance. With our method, a 175B parameter 16/32-bit checkpoint can be loaded, converted to Int8, and used immediately without performance degradation. This is made possible by understanding and working around properties of highly systematic emergent features in transformer language models that dominate attention and transformer predictive performance. To cope with these features, we develop a two-part quantization procedure, LLM.int8(). We first use vector-wise quantization with separate normalization constants for each inner product in the matrix multiplication, to quantize most of the features. However, for the emergent outliers, we also include a new mixed-precision decomposition scheme, which isolates the outlier feature dimensions into a 16-bit matrix multiplication while still more than 99.9% of values are multiplied in 8-bit. Using LLM.int8(), we show empirically it is possible to perform inference in LLMs with up to 175B parameters without any performance degradation. This result makes such models much more accessible, for example making it possible to use OPT-175B/BLOOM on a single server with consumer GPUs. We open-source our software.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2022-09", "2023-02", "2023-03", "2023-05", "2024-06"], "last_mentioned": "2024-06-20T18:17:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2209.07663", "slug": "767479f5b9bbd976", "title": "Monolith: Real Time Recommendation System With Collisionless Embedding Table", "authors": ["Zhuoran Liu", "Leqi Zou", "Xuan Zou", "Caihua Wang", "Biao Zhang", "Da Tang", "Bolin Zhu", "Yijie Zhu", "Peng Wu", "Ke Wang", "Youlong Cheng"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2209.07663", "doi": null, "topics": ["Machine learning", "Distributed systems", "Algorithms"], "description": "Building a scalable and real-time recommendation system is vital for many businesses driven by time-sensitive customer feedback, such as short-videos ranking or online ads.", "description_method": "abstract-excerpt", "abstract": "Building a scalable and real-time recommendation system is vital for many businesses driven by time-sensitive customer feedback, such as short-videos ranking or online ads. Despite the ubiquitous adoption of production-scale deep learning frameworks like TensorFlow or PyTorch, these general-purpose frameworks fall short of business demands in recommendation scenarios for various reasons: on one hand, tweaking systems based on static parameters and dense computations for recommendation with dynamic and sparse features is detrimental to model quality; on the other hand, such frameworks are designed with batch-training stage and serving stage completely separated, preventing the model from interacting with customer feedback in real-time. These issues led us to reexamine traditional approaches and explore radically different design choices. In this paper, we present Monolith, a system tailored for online training. Our design has been driven by observations of our application workloads and production environment that reflects a marked departure from other recommendations systems. Our contributions are manifold: first, we crafted a collisionless embedding table with optimizations such as expirable embeddings and frequency filtering to reduce its memory footprint; second, we provide an production-ready online training architecture with high fault-tolerance; finally, we proved that system reliability could be traded-off for real-time learning. Monolith has successfully landed in the BytePlus Recommend product.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2022-12", "2023-11", "2024-12", "2025-01", "2025-05", "2026-02"], "last_mentioned": "2026-02-06T13:27:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2210.07128", "slug": "d53be7cd5be35dd8", "title": "Language Models of Code are Few-Shot Commonsense Learners", "authors": ["Aman Madaan", "Shuyan Zhou", "Uri Alon", "Yiming Yang", "Graham Neubig"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2210.07128", "doi": null, "topics": ["Language models"], "description": "We address the general task of structured commonsense reasoning: given a natural language input, the goal is to generate a graph such as an event -- or a reasoning-graph.", "description_method": "abstract-excerpt", "abstract": "We address the general task of structured commonsense reasoning: given a natural language input, the goal is to generate a graph such as an event -- or a reasoning-graph. To employ large language models (LMs) for this task, existing approaches ``serialize'' the output graph as a flat list of nodes and edges. Although feasible, these serialized graphs strongly deviate from the natural language corpora that LMs were pre-trained on, hindering LMs from generating them correctly. In this paper, we show that when we instead frame structured commonsense reasoning tasks as code generation tasks, pre-trained LMs of code are better structured commonsense reasoners than LMs of natural language, even when the downstream task does not involve source code at all. We demonstrate our approach across three diverse structured commonsense reasoning tasks. In all these natural language tasks, we show that using our approach, a code generation LM (CODEX) outperforms natural-LMs that are fine-tuned on the target task (e.g., T5) and other strong LMs such as GPT-3 in the few-shot setting.", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2023-05", "2023-06", "2023-07", "2023-09"], "last_mentioned": "2023-09-01T15:59:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2210.15723", "slug": "381eb9cdf1fe507f", "title": "Birdwatch: Crowd Wisdom and Bridging Algorithms can Inform Understanding and Reduce the Spread of Misinformation", "authors": ["Stefan Wojcik", "Sophie Hilgard", "Nick Judd", "Delia Mocanu", "Stephen Ragain", "M. B. Fallin Hunzaker", "Keith Coleman", "Jay Baxter"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2210.15723", "doi": null, "topics": ["Algorithms"], "description": "We present an approach for selecting objectively informative and subjectively helpful annotations to social media posts.", "description_method": "abstract-excerpt", "abstract": "We present an approach for selecting objectively informative and subjectively helpful annotations to social media posts. We draw on data from on an online environment where contributors annotate misinformation and simultaneously rate the contributions of others. Our algorithm uses a matrix-factorization (MF) based approach to identify annotations that appeal broadly across heterogeneous user groups - sometimes referred to as \"bridging-based ranking.\" We pair these data with a survey experiment in which individuals are randomly assigned to see annotations to posts. We find that annotations selected by the algorithm improve key indicators compared with overall average and crowd-generated baselines. Further, when deployed on Twitter, people who saw annotations selected through this bridging-based approach were significantly less likely to reshare social media posts than those who did not see the annotations.", "score": 0.0, "mention_count": 6, "user_count": 4, "months": ["2022-11", "2023-02", "2024-02", "2025-01"], "last_mentioned": "2025-01-07T23:33:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2211.17192", "slug": "d37e62062c2a8803", "title": "Fast Inference from Transformers via Speculative Decoding", "authors": ["Yaniv Leviathan", "Matan Kalman", "Yossi Matias"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2211.17192", "doi": null, "topics": ["Language models", "Algorithms"], "description": "Inference from large autoregressive models like Transformers is slow - decoding K tokens takes K serial runs of the model.", "description_method": "abstract-excerpt", "abstract": "Inference from large autoregressive models like Transformers is slow - decoding K tokens takes K serial runs of the model. In this work we introduce speculative decoding - an algorithm to sample from autoregressive models faster without any changes to the outputs, by computing several tokens in parallel. At the heart of our approach lie the observations that (1) hard language-modeling tasks often include easier subtasks that can be approximated well by more efficient models, and (2) using speculative execution and a novel sampling method, we can make exact decoding from the large models faster, by running them in parallel on the outputs of the approximation models, potentially generating several tokens concurrently, and without changing the distribution. Our method can accelerate existing off-the-shelf models without retraining or architecture changes. We demonstrate it on T5-XXL and show a 2X-3X acceleration compared to the standard T5X implementation, with identical outputs.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2025-07", "2026-02", "2026-04", "2026-05", "2026-06"], "last_mentioned": "2026-06-27T15:34:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.07677", "slug": "1fd467ae7d76e8ae", "title": "Transformers learn in-context by gradient descent", "authors": ["Johannes von Oswald", "Eyvind Niklasson", "Ettore Randazzo", "João Sacramento", "Alexander Mordvintsev", "Andrey Zhmoginov", "Max Vladymyrov"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.07677", "doi": null, "topics": ["Language models", "Machine learning", "Statistics & probability"], "description": "At present, the mechanisms of in-context learning in Transformers are not well understood and remain mostly an intuition.", "description_method": "abstract-excerpt", "abstract": "At present, the mechanisms of in-context learning in Transformers are not well understood and remain mostly an intuition. In this paper, we suggest that training Transformers on auto-regressive objectives is closely related to gradient-based meta-learning formulations. We start by providing a simple weight construction that shows the equivalence of data transformations induced by 1) a single linear self-attention layer and by 2) gradient-descent (GD) on a regression loss. Motivated by that construction, we show empirically that when training self-attention-only Transformers on simple regression tasks either the models learned by GD and Transformers show great similarity or, remarkably, the weights found by optimization match the construction. Thus we show how trained Transformers become mesa-optimizers i.e. learn models by gradient descent in their forward pass. This allows us, at least in the domain of regression problems, to mechanistically understand the inner workings of in-context learning in optimized Transformers. Building on this insight, we furthermore identify how Transformers surpass the performance of plain gradient descent by learning an iterative curvature correction and learn linear models on deep data representations to solve non-linear regression tasks. Finally, we discuss intriguing parallels to a mechanism identified to be crucial for in-context learning termed induction-head (Olsson et al., 2022) and show how it could be understood as a specific case of in-context learning by gradient descent learning within Transformers. Code to reproduce the experiments can be found at https://github.com/google-research/self-organising-systems/tree/master/transformers_learn_icl_by_gd .", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2023-01", "2023-02", "2023-04", "2025-03"], "last_mentioned": "2025-03-04T21:39:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.10560", "slug": "0e442cfef7c56fd7", "title": "Self-Instruct: Aligning Language Models with Self-Generated Instructions", "authors": ["Yizhong Wang", "Yeganeh Kordi", "Swaroop Mishra", "Alisa Liu", "Noah A. Smith", "Daniel Khashabi", "Hannaneh Hajishirzi"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.10560", "doi": null, "topics": ["Language models"], "description": "Large \"instruction-tuned\" language models (i.e., finetuned to respond to instructions) have demonstrated a remarkable ability to generalize zero-shot to new tasks.", "description_method": "abstract-excerpt", "abstract": "Large \"instruction-tuned\" language models (i.e., finetuned to respond to instructions) have demonstrated a remarkable ability to generalize zero-shot to new tasks. Nevertheless, they depend heavily on human-written instruction data that is often limited in quantity, diversity, and creativity, therefore hindering the generality of the tuned model. We introduce Self-Instruct, a framework for improving the instruction-following capabilities of pretrained language models by bootstrapping off their own generations. Our pipeline generates instructions, input, and output samples from a language model, then filters invalid or similar ones before using them to finetune the original model. Applying our method to the vanilla GPT3, we demonstrate a 33% absolute improvement over the original model on Super-NaturalInstructions, on par with the performance of InstructGPT-001, which was trained with private user data and human annotations. For further evaluation, we curate a set of expert-written instructions for novel tasks, and show through human evaluation that tuning GPT3 with Self-Instruct outperforms using existing public instruction datasets by a large margin, leaving only a 5% absolute gap behind InstructGPT-001. Self-Instruct provides an almost annotation-free method for aligning pre-trained language models with instructions, and we release our large synthetic dataset to facilitate future studies on instruction tuning. Our code and data are available at https://github.com/yizhongw/self-instruct.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2023-03", "2023-05", "2023-12", "2024-12", "2025-01"], "last_mentioned": "2025-01-14T00:44:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.13345", "slug": "71b9d3fd68475889", "title": "The Forward-Forward Algorithm: Some Preliminary Investigations", "authors": ["Geoffrey Hinton"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.13345", "doi": null, "topics": ["Algorithms", "Machine learning"], "description": "The aim of this paper is to introduce a new learning procedure for neural networks and to demonstrate that it works well enough on a few small problems to be worth further investigation.", "description_method": "abstract-excerpt", "abstract": "The aim of this paper is to introduce a new learning procedure for neural networks and to demonstrate that it works well enough on a few small problems to be worth further investigation. The Forward-Forward algorithm replaces the forward and backward passes of backpropagation by two forward passes, one with positive (i.e. real) data and the other with negative data which could be generated by the network itself. Each layer has its own objective function which is simply to have high goodness for positive data and low goodness for negative data. The sum of the squared activities in a layer can be used as the goodness but there are many other possibilities, including minus the sum of the squared activities. If the positive and negative passes could be separated in time, the negative passes could be done offline, which would make the learning much simpler in the positive pass and allow video to be pipelined through the network without ever storing activities or stopping to propagate derivatives.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2023-01", "2023-06", "2024-04", "2024-07", "2025-01"], "last_mentioned": "2025-01-30T21:31:04Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.00774", "slug": "43e7a8ef05317e78", "title": "SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shot", "authors": ["Elias Frantar", "Dan Alistarh"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.00774", "doi": null, "topics": ["Language models"], "description": "We show for the first time that large-scale generative pretrained transformer (GPT) family models can be pruned to at least 50% sparsity in one-shot, without any retraining, at minimal loss of accuracy.", "description_method": "abstract-excerpt", "abstract": "We show for the first time that large-scale generative pretrained transformer (GPT) family models can be pruned to at least 50% sparsity in one-shot, without any retraining, at minimal loss of accuracy. This is achieved via a new pruning method called SparseGPT, specifically designed to work efficiently and accurately on massive GPT-family models. We can execute SparseGPT on the largest available open-source models, OPT-175B and BLOOM-176B, in under 4.5 hours, and can reach 60% unstructured sparsity with negligible increase in perplexity: remarkably, more than 100 billion weights from these models can be ignored at inference time. SparseGPT generalizes to semi-structured (2:4 and 4:8) patterns, and is compatible with weight quantization approaches. The code is available at: https://github.com/IST-DASLab/sparsegpt.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2023-01", "2023-04", "2023-05", "2024-04"], "last_mentioned": "2024-04-18T04:50:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.06627", "slug": "3791d7ed420be517", "title": "Dissociating language and thought in large language models", "authors": ["Kyle Mahowald", "Anna A. Ivanova", "Idan A. Blank", "Nancy Kanwisher", "Joshua B. Tenenbaum", "Evelina Fedorenko"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.06627", "doi": null, "topics": ["Language models", "Neuroscience"], "description": "Large Language Models (LLMs) have come closest among all models to date to mastering human language, yet opinions about their linguistic and cognitive capabilities remain split.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) have come closest among all models to date to mastering human language, yet opinions about their linguistic and cognitive capabilities remain split. Here, we evaluate LLMs using a distinction between formal linguistic competence -- knowledge of linguistic rules and patterns -- and functional linguistic competence -- understanding and using language in the world. We ground this distinction in human neuroscience, which has shown that formal and functional competence rely on different neural mechanisms. Although LLMs are surprisingly good at formal competence, their performance on functional competence tasks remains spotty and often requires specialized fine-tuning and/or coupling with external modules. We posit that models that use language in human-like ways would need to master both of these competence types, which, in turn, could require the emergence of mechanisms specialized for formal linguistic competence, distinct from functional competence.", "score": 0.0, "mention_count": 6, "user_count": 4, "months": ["2023-02", "2023-04", "2023-05", "2023-07", "2023-10"], "last_mentioned": "2023-10-19T14:42:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.10191", "slug": "8f523b1930b001a4", "title": "Distinct Elements in Streams: An Algorithm for the (Text) Book", "authors": ["Sourav Chakraborty", "N. V. Vinodchandran", "Kuldeep S. Meel"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.10191", "doi": "10.4230/LIPIcs.ESA.2022.34", "topics": ["Algorithms", "Statistics & probability"], "description": "Given a data stream $\\mathcal{A} = \\langle a_1, a_2, \\ldots, a_m \\rangle$ of $m$ elements where each $a_i \\in [n]$, the Distinct Elements problem is to estimate the number of distinct elements in $\\mathcal{A}$.Distinct Elements has been a subject of theoretical and empirical investigations over the past four decades resulting in space opt", "description_method": "abstract-excerpt", "abstract": "Given a data stream $\\mathcal{A} = \\langle a_1, a_2, \\ldots, a_m \\rangle$ of $m$ elements where each $a_i \\in [n]$, the Distinct Elements problem is to estimate the number of distinct elements in $\\mathcal{A}$.Distinct Elements has been a subject of theoretical and empirical investigations over the past four decades resulting in space optimal algorithms for it.All the current state-of-the-art algorithms are, however, beyond the reach of an undergraduate textbook owing to their reliance on the usage of notions such as pairwise independence and universal hash functions. We present a simple, intuitive, sampling-based space-efficient algorithm whose description and the proof are accessible to undergraduates with the knowledge of basic probability theory.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2023-11", "2024-05", "2025-02"], "last_mentioned": "2025-02-11T06:51:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.13867", "slug": "f2051b2a5d2e036d", "title": "Mathematical Capabilities of ChatGPT", "authors": ["Simon Frieder", "Luca Pinchetti", "Alexis Chevalier", "Ryan-Rhys Griffiths", "Tommaso Salvatori", "Thomas Lukasiewicz", "Philipp Christian Petersen", "Julius Berner"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.13867", "doi": null, "topics": ["Mathematics", "Language models", "Databases"], "description": "We investigate the mathematical capabilities of two iterations of ChatGPT (released 9-January-2023 and 30-January-2023) and of GPT-4 by testing them on publicly available datasets, as well as hand-crafted ones, using a novel methodology.", "description_method": "abstract-excerpt", "abstract": "We investigate the mathematical capabilities of two iterations of ChatGPT (released 9-January-2023 and 30-January-2023) and of GPT-4 by testing them on publicly available datasets, as well as hand-crafted ones, using a novel methodology. In contrast to formal mathematics, where large databases of formal proofs are available (e.g., the Lean Mathematical Library), current datasets of natural-language mathematics, used to benchmark language models, either cover only elementary mathematics or are very small. We address this by publicly releasing two new datasets: GHOSTS and miniGHOSTS. These are the first natural-language datasets curated by working researchers in mathematics that (1) aim to cover graduate-level mathematics, (2) provide a holistic overview of the mathematical capabilities of language models, and (3) distinguish multiple dimensions of mathematical reasoning. These datasets also test whether ChatGPT and GPT-4 can be helpful assistants to professional mathematicians by emulating use cases that arise in the daily professional activities of mathematicians. We benchmark the models on a range of fine-grained performance metrics. For advanced mathematics, this is the most detailed evaluation effort to date. We find that ChatGPT can be used most successfully as a mathematical assistant for querying facts, acting as a mathematical search engine and knowledge base interface. GPT-4 can additionally be used for undergraduate-level mathematics but fails on graduate-level difficulty. Contrary to many positive reports in the media about GPT-4 and ChatGPT's exam-solving abilities (a potential case of selection bias), their overall mathematical performance is well below the level of a graduate student. Hence, if your goal is to use ChatGPT to pass a graduate-level math exam, you would be better off copying from your average peer!", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2023-03", "2023-04", "2023-07", "2023-10"], "last_mentioned": "2023-10-16T16:30:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.02301", "slug": "b322351c021299bd", "title": "Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizes", "authors": ["Cheng-Yu Hsieh", "Chun-Liang Li", "Chih-Kuan Yeh", "Hootan Nakhost", "Yasuhisa Fujii", "Alexander Ratner", "Ranjay Krishna", "Chen-Yu Lee", "Tomas Pfister"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.02301", "doi": null, "topics": ["Language models"], "description": "Deploying large language models (LLMs) is challenging because they are memory inefficient and compute-intensive for practical applications.", "description_method": "abstract-excerpt", "abstract": "Deploying large language models (LLMs) is challenging because they are memory inefficient and compute-intensive for practical applications. In reaction, researchers train smaller task-specific models by either finetuning with human labels or distilling using LLM-generated labels. However, finetuning and distillation require large amounts of training data to achieve comparable performance to LLMs. We introduce Distilling step-by-step, a new mechanism that (a) trains smaller models that outperform LLMs, and (b) achieves so by leveraging less training data needed by finetuning or distillation. Our method extracts LLM rationales as additional supervision for training small models within a multi-task framework. We present three findings across 4 NLP benchmarks: First, compared to both finetuning and distillation, our mechanism achieves better performance with much fewer labeled/unlabeled training examples. Second, compared to few-shot prompted LLMs, we achieve better performance using substantially smaller model sizes. Third, we reduce both the model size and the amount of data required to outperform LLMs; our finetuned 770M T5 model outperforms the few-shot prompted 540B PaLM model using only 80% of available data on a benchmark, whereas standard finetuning the same T5 model struggles to match even by using 100% of the dataset. We release the code at: https://github.com/google-research/distilling-step-by-step .", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2023-05", "2023-07", "2023-12", "2025-01"], "last_mentioned": "2025-01-28T00:01:39Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.16291", "slug": "b71b1297191d5227", "title": "Voyager: An Open-Ended Embodied Agent with Large Language Models", "authors": ["Guanzhi Wang", "Yuqi Xie", "Yunfan Jiang", "Ajay Mandlekar", "Chaowei Xiao", "Yuke Zhu", "Linxi Fan", "Anima Anandkumar"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.16291", "doi": null, "topics": ["Language models", "Robotics", "Psychology"], "description": "We introduce Voyager, the first LLM-powered embodied lifelong learning agent in Minecraft that continuously explores the world, acquires diverse skills, and makes novel discoveries without human intervention.", "description_method": "abstract-excerpt", "abstract": "We introduce Voyager, the first LLM-powered embodied lifelong learning agent in Minecraft that continuously explores the world, acquires diverse skills, and makes novel discoveries without human intervention. Voyager consists of three key components: 1) an automatic curriculum that maximizes exploration, 2) an ever-growing skill library of executable code for storing and retrieving complex behaviors, and 3) a new iterative prompting mechanism that incorporates environment feedback, execution errors, and self-verification for program improvement. Voyager interacts with GPT-4 via blackbox queries, which bypasses the need for model parameter fine-tuning. The skills developed by Voyager are temporally extended, interpretable, and compositional, which compounds the agent's abilities rapidly and alleviates catastrophic forgetting. Empirically, Voyager shows strong in-context lifelong learning capability and exhibits exceptional proficiency in playing Minecraft. It obtains 3.3x more unique items, travels 2.3x longer distances, and unlocks key tech tree milestones up to 15.3x faster than prior SOTA. Voyager is able to utilize the learned skill library in a new Minecraft world to solve novel tasks from scratch, while other techniques struggle to generalize. We open-source our full codebase and prompts at https://voyager.minedojo.org/.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2023-08", "2023-11", "2024-03", "2025-05", "2025-09", "2026-01"], "last_mentioned": "2026-01-19T06:50:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.10763", "slug": "c76b5987766d4113", "title": "Guiding Language Models of Code with Global Context using Monitors", "authors": ["Lakshya A Agrawal", "Aditya Kanade", "Navin Goyal", "Shuvendu K. Lahiri", "Sriram K. Rajamani"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.10763", "doi": null, "topics": ["Language models", "Programming languages"], "description": "Language models of code (LMs) work well when the surrounding code provides sufficient context.", "description_method": "abstract-excerpt", "abstract": "Language models of code (LMs) work well when the surrounding code provides sufficient context. This is not true when it becomes necessary to use types, functionality or APIs defined elsewhere in the repository or a linked library, especially those not seen during training. LMs suffer from limited awareness of such global context and end up hallucinating. Integrated development environments (IDEs) assist developers in understanding repository context using static analysis. We extend this assistance, enjoyed by developers, to LMs. We propose monitor-guided decoding (MGD) where a monitor uses static analysis to guide the decoding. We construct a repository-level dataset PragmaticCode for method-completion in Java and evaluate MGD on it. On models of varying parameter scale, by monitoring for type-consistent object dereferences, MGD consistently improves compilation rates and agreement with ground truth. Further, LMs with fewer parameters, when augmented with MGD, can outperform larger LMs. With MGD, SantaCoder-1.1B achieves better compilation rate and next-identifier match than the much larger text-davinci-003 model. We also conduct a generalizability study to evaluate the ability of MGD to generalize to multiple programming languages (Java, C# and Rust), coding scenarios (e.g., correct number of arguments to method calls), and to enforce richer semantic constraints (e.g., stateful API protocols). Our data and implementation are available at https://github.com/microsoft/monitors4codegen .", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2023-08", "2023-11", "2024-03", "2025-05"], "last_mentioned": "2025-05-14T01:57:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2308.08708", "slug": "e7d3e0c10e0ae2d2", "title": "Consciousness in Artificial Intelligence: Insights from the Science of Consciousness", "authors": ["Patrick Butlin", "Robert Long", "Eric Elmoznino", "Yoshua Bengio", "Jonathan Birch", "Axel Constant", "George Deane", "Stephen M. Fleming", "Chris Frith", "Xu Ji", "Ryota Kanai", "Colin Klein", "Grace Lindsay", "Matthias Michel", "Liad Mudrik", "Megan A. K. Peters", "Eric Schwitzgebel", "Jonathan Simon", "Rufin VanRullen"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2308.08708", "doi": null, "topics": ["Neuroscience", "Science & research", "Psychology"], "description": "Whether current or near-term AI systems could be conscious is a topic of scientific interest and increasing public concern.", "description_method": "abstract-excerpt", "abstract": "Whether current or near-term AI systems could be conscious is a topic of scientific interest and increasing public concern. This report argues for, and exemplifies, a rigorous and empirically grounded approach to AI consciousness: assessing existing AI systems in detail, in light of our best-supported neuroscientific theories of consciousness. We survey several prominent scientific theories of consciousness, including recurrent processing theory, global workspace theory, higher-order theories, predictive processing, and attention schema theory. From these theories we derive \"indicator properties\" of consciousness, elucidated in computational terms that allow us to assess AI systems for these properties. We use these indicator properties to assess several recent AI systems, and we discuss how future systems might implement them. Our analysis suggests that no current AI systems are conscious, but also suggests that there are no obvious technical barriers to building AI systems which satisfy these indicators.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2023-08", "2023-11", "2023-12", "2024-03", "2025-12"], "last_mentioned": "2025-12-01T13:02:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2309.17453", "slug": "953c9fc74091f77c", "title": "Efficient Streaming Language Models with Attention Sinks", "authors": ["Guangxuan Xiao", "Yuandong Tian", "Beidi Chen", "Song Han", "Mike Lewis"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2309.17453", "doi": null, "topics": ["Language models"], "description": "Deploying Large Language Models (LLMs) in streaming applications such as multi-round dialogue, where long interactions are expected, is urgently needed but poses two major challenges.", "description_method": "abstract-excerpt", "abstract": "Deploying Large Language Models (LLMs) in streaming applications such as multi-round dialogue, where long interactions are expected, is urgently needed but poses two major challenges. Firstly, during the decoding stage, caching previous tokens' Key and Value states (KV) consumes extensive memory. Secondly, popular LLMs cannot generalize to longer texts than the training sequence length. Window attention, where only the most recent KVs are cached, is a natural approach -- but we show that it fails when the text length surpasses the cache size. We observe an interesting phenomenon, namely attention sink, that keeping the KV of initial tokens will largely recover the performance of window attention. In this paper, we first demonstrate that the emergence of attention sink is due to the strong attention scores towards initial tokens as a \"sink\" even if they are not semantically important. Based on the above analysis, we introduce StreamingLLM, an efficient framework that enables LLMs trained with a finite length attention window to generalize to infinite sequence lengths without any fine-tuning. We show that StreamingLLM can enable Llama-2, MPT, Falcon, and Pythia to perform stable and efficient language modeling with up to 4 million tokens and more. In addition, we discover that adding a placeholder token as a dedicated attention sink during pre-training can further improve streaming deployment. In streaming settings, StreamingLLM outperforms the sliding window recomputation baseline by up to 22.2x speedup. Code and datasets are provided at https://github.com/mit-han-lab/streaming-llm.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2023-10", "2024-12", "2025-10"], "last_mentioned": "2025-10-24T19:48:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2311.09247", "slug": "2e59bd87ed88c20e", "title": "Comparing Humans, GPT-4, and GPT-4V On Abstraction and Reasoning Tasks", "authors": ["Melanie Mitchell", "Alessandro B. Palmarini", "Arseny Moskvichev"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2311.09247", "doi": null, "topics": ["Language models"], "description": "We explore the abstract reasoning abilities of text-only and multimodal versions of GPT-4, using the ConceptARC benchmark [10], which is designed to evaluate robust understanding and reasoning with core-knowledge concepts.", "description_method": "abstract-excerpt", "abstract": "We explore the abstract reasoning abilities of text-only and multimodal versions of GPT-4, using the ConceptARC benchmark [10], which is designed to evaluate robust understanding and reasoning with core-knowledge concepts. We extend the work of Moskvichev et al. [10] by evaluating GPT-4 on more detailed, one-shot prompting (rather than simple, zero-shot prompts) with text versions of ConceptARC tasks, and by evaluating GPT-4V, the multimodal version of GPT-4, on zero- and one-shot prompts using image versions of the simplest tasks. Our experimental results support the conclusion that neither version of GPT-4 has developed robust abstraction abilities at humanlike levels.", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2023-11", "2023-12", "2024-08", "2025-06"], "last_mentioned": "2025-06-15T14:32:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2312.11805", "slug": "f54ec754c1d7009a", "title": "Gemini: A Family of Highly Capable Multimodal Models", "authors": [" Gemini Team", "Rohan Anil", "Sebastian Borgeaud", "Jean-Baptiste Alayrac", "Jiahui Yu", "Radu Soricut", "Johan Schalkwyk", "Andrew M. Dai", "Anja Hauth", "Katie Millican", "David Silver", "Melvin Johnson", "Ioannis Antonoglou", "Julian Schrittwieser", "Amelia Glaese", "Jilin Chen", "Emily Pitler", "Timothy Lillicrap", "Angeliki Lazaridou", "Orhan Firat", "James Molloy", "Michael Isard", "Paul R. Barham", "Tom Hennigan", "Benjamin Lee", "Fabio Viola", "Malcolm Reynolds", "Yuanzhong Xu", "Ryan Doherty", "Eli Collins", "Clemens Meyer", "Eliza Rutherford", "Erica Moreira", "Kareem Ayoub", "Megha Goel", "Jack Krawczyk", "Cosmo Du", "Ed Chi", "Heng-Tze Cheng", "Eric Ni", "Purvi Shah", "Patrick Kane", "Betty Chan", "Manaal Faruqui", "Aliaksei Severyn", "Hanzhao Lin", "YaGuang Li", "Yong Cheng", "Abe Ittycheriah", "Mahdis Mahdieh", "Mia Chen", "Pei Sun", "Dustin Tran", "Sumit Bagri", "Balaji Lakshminarayanan", "Jeremiah Liu", "Andras Orban", "Fabian Güra", "Hao Zhou", "Xinying Song", "Aurelien Boffy", "Harish Ganapathy", "Steven Zheng", "HyunJeong Choe", "Ágoston Weisz", "Tao Zhu", "Yifeng Lu", "Siddharth Gopal", "Jarrod Kahn", "Maciej Kula", "Jeff Pitman", "Rushin Shah", "Emanuel Taropa", "Majd Al Merey", "Martin Baeuml", "Zhifeng Chen", "Laurent El Shafey", "Yujing Zhang", "Olcan Sercinoglu", "George Tucker", "Enrique Piqueras", "Maxim Krikun", "Iain Barr", "Nikolay Savinov", "Ivo Danihelka", "Becca Roelofs", "Anaïs White", "Anders Andreassen", "Tamara von Glehn", "Lakshman Yagati", "Mehran Kazemi", "Lucas Gonzalez", "Misha Khalman", "Jakub Sygnowski", "Alexandre Frechette", "Charlotte Smith", "Laura Culp", "Lev Proleev", "Yi Luan", "Xi Chen", "James Lottes", "Nathan Schucher", "Federico Lebron", "Alban Rrustemi", "Natalie Clay", "Phil Crone", "Tomas Kocisky", "Jeffrey Zhao", "Bartek Perz", "Dian Yu", "Heidi Howard", "Adam Bloniarz", "Jack W. Rae", "Han Lu", "Laurent Sifre", "Marcello Maggioni", "Fred Alcober", "Dan Garrette", "Megan Barnes", "Shantanu Thakoor", "Jacob Austin", "Gabriel Barth-Maron", "William Wong", "Rishabh Joshi", "Rahma Chaabouni", "Deeni Fatiha", "Arun Ahuja", "Gaurav Singh Tomar", "Evan Senter", "Martin Chadwick", "Ilya Kornakov", "Nithya Attaluri", "Iñaki Iturrate", "Ruibo Liu", "Yunxuan Li", "Sarah Cogan", "Jeremy Chen", "Chao Jia", "Chenjie Gu", "Qiao Zhang", "Jordan Grimstad", "Ale Jakse Hartman", "Xavier Garcia", "Thanumalayan Sankaranarayana Pillai", "Jacob Devlin", "Michael Laskin", "Diego de Las Casas", "Dasha Valter", "Connie Tao", "Lorenzo Blanco", "Adrià Puigdomènech Badia", "David Reitter", "Mianna Chen", "Jenny Brennan", "Clara Rivera", "Sergey Brin", "Shariq Iqbal", "Gabriela Surita", "Jane Labanowski", "Abhi Rao", "Stephanie Winkler", "Emilio Parisotto", "Yiming Gu", "Kate Olszewska", "Ravi Addanki", "Antoine Miech", "Annie Louis", "Denis Teplyashin", "Geoff Brown", "Elliot Catt", "Jan Balaguer", "Jackie Xiang", "Pidong Wang", "Zoe Ashwood", "Anton Briukhov", "Albert Webson", "Sanjay Ganapathy", "Smit Sanghavi", "Ajay Kannan", "Ming-Wei Chang", "Axel Stjerngren", "Josip Djolonga", "Yuting Sun", "Ankur Bapna", "Matthew Aitchison", "Pedram Pejman", "Henryk Michalewski", "Tianhe Yu", "Cindy Wang", "Juliette Love", "Junwhan Ahn", "Dawn Bloxwich", "Kehang Han", "Peter Humphreys", "Thibault Sellam", "James Bradbury", "Varun Godbole", "Sina Samangooei", "Bogdan Damoc", "Alex Kaskasoli", "Sébastien M. R. Arnold", "Vijay Vasudevan", "Shubham Agrawal", "Jason Riesa", "Dmitry Lepikhin", "Richard Tanburn", "Srivatsan Srinivasan", "Hyeontaek Lim", "Sarah Hodkinson", "Pranav Shyam", "Johan Ferret", "Steven Hand", "Ankush Garg", "Tom Le Paine", "Jian Li", "Yujia Li", "Minh Giang", "Alexander Neitz", "Zaheer Abbas", "Sarah York", "Machel Reid", "Elizabeth Cole", "Aakanksha Chowdhery", "Dipanjan Das", "Dominika Rogozińska", "Vitaliy Nikolaev", "Pablo Sprechmann", "Zachary Nado", "Lukas Zilka", "Flavien Prost", "Luheng He", "Marianne Monteiro", "Gaurav Mishra", "Chris Welty", "Josh Newlan", "Dawei Jia", "Miltiadis Allamanis", "Clara Huiyi Hu", "Raoul de Liedekerke", "Justin Gilmer", "Carl Saroufim", "Shruti Rijhwani", "Shaobo Hou", "Disha Shrivastava", "Anirudh Baddepudi", "Alex Goldin", "Adnan Ozturel", "Albin Cassirer", "Yunhan Xu", "Daniel Sohn", "Devendra Sachan", "Reinald Kim Amplayo", "Craig Swanson", "Dessie Petrova", "Shashi Narayan", "Arthur Guez", "Siddhartha Brahma", "Jessica Landon", "Miteyan Patel", "Ruizhe Zhao", "Kevin Villela", "Luyu Wang", "Wenhao Jia", "Matthew Rahtz", "Mai Giménez", "Legg Yeung", "James Keeling", "Petko Georgiev", "Diana Mincu", "Boxi Wu", "Salem Haykal", "Rachel Saputro", "Kiran Vodrahalli", "James Qin", "Zeynep Cankara", "Abhanshu Sharma", "Nick Fernando", "Will Hawkins", "Behnam Neyshabur", "Solomon Kim", "Adrian Hutter", "Priyanka Agrawal", "Alex Castro-Ros", "George van den Driessche", "Tao Wang", "Fan Yang", "Shuo-yiin Chang", "Paul Komarek", "Ross McIlroy", "Mario Lučić", "Guodong Zhang", "Wael Farhan", "Michael Sharman", "Paul Natsev", "Paul Michel", "Yamini Bansal", "Siyuan Qiao", "Kris Cao", "Siamak Shakeri", "Christina Butterfield", "Justin Chung", "Paul Kishan Rubenstein", "Shivani Agrawal", "Arthur Mensch", "Kedar Soparkar", "Karel Lenc", "Timothy Chung", "Aedan Pope", "Loren Maggiore", "Jackie Kay", "Priya Jhakra", "Shibo Wang", "Joshua Maynez", "Mary Phuong", "Taylor Tobin", "Andrea Tacchetti", "Maja Trebacz", "Kevin Robinson", "Yash Katariya", "Sebastian Riedel", "Paige Bailey", "Kefan Xiao", "Nimesh Ghelani", "Lora Aroyo", "Ambrose Slone", "Neil Houlsby", "Xuehan Xiong", "Zhen Yang", "Elena Gribovskaya", "Jonas Adler", "Mateo Wirth", "Lisa Lee", "Music Li", "Thais Kagohara", "Jay Pavagadhi", "Sophie Bridgers", "Anna Bortsova", "Sanjay Ghemawat", "Zafarali Ahmed", "Tianqi Liu", "Richard Powell", "Vijay Bolina", "Mariko Iinuma", "Polina Zablotskaia", "James Besley", "Da-Woon Chung", "Timothy Dozat", "Ramona Comanescu", "Xiance Si", "Jeremy Greer", "Guolong Su", "Martin Polacek", "Raphaël Lopez Kaufman", "Simon Tokumine", "Hexiang Hu", "Elena Buchatskaya", "Yingjie Miao", "Mohamed Elhawaty", "Aditya Siddhant", "Nenad Tomasev", "Jinwei Xing", "Christina Greer", "Helen Miller", "Shereen Ashraf", "Aurko Roy", "Zizhao Zhang", "Ada Ma", "Angelos Filos", "Milos Besta", "Rory Blevins", "Ted Klimenko", "Chih-Kuan Yeh", "Soravit Changpinyo", "Jiaqi Mu", "Oscar Chang", "Mantas Pajarskas", "Carrie Muir", "Vered Cohen", "Charline Le Lan", "Krishna Haridasan", "Amit Marathe", "Steven Hansen", "Sholto Douglas", "Rajkumar Samuel", "Mingqiu Wang", "Sophia Austin", "Chang Lan", "Jiepu Jiang", "Justin Chiu", "Jaime Alonso Lorenzo", "Lars Lowe Sjösund", "Sébastien Cevey", "Zach Gleicher", "Thi Avrahami", "Anudhyan Boral", "Hansa Srinivasan", "Vittorio Selo", "Rhys May", "Konstantinos Aisopos", "Léonard Hussenot", "Livio Baldini Soares", "Kate Baumli", "Michael B. Chang", "Adrià Recasens", "Ben Caine", "Alexander Pritzel", "Filip Pavetic", "Fabio Pardo", "Anita Gergely", "Justin Frye", "Vinay Ramasesh", "Dan Horgan", "Kartikeya Badola", "Nora Kassner", "Subhrajit Roy", "Ethan Dyer", "Víctor Campos Campos", "Alex Tomala", "Yunhao Tang", "Dalia El Badawy", "Elspeth White", "Basil Mustafa", "Oran Lang", "Abhishek Jindal", "Sharad Vikram", "Zhitao Gong", "Sergi Caelles", "Ross Hemsley", "Gregory Thornton", "Fangxiaoyu Feng", "Wojciech Stokowiec", "Ce Zheng", "Phoebe Thacker", "Çağlar Ünlü", "Zhishuai Zhang", "Mohammad Saleh", "James Svensson", "Max Bileschi", "Piyush Patil", "Ankesh Anand", "Roman Ring", "Katerina Tsihlas", "Arpi Vezer", "Marco Selvi", "Toby Shevlane", "Mikel Rodriguez", "Tom Kwiatkowski", "Samira Daruki", "Keran Rong", "Allan Dafoe", "Nicholas FitzGerald", "Keren Gu-Lemberg", "Mina Khan", "Lisa Anne Hendricks", "Marie Pellat", "Vladimir Feinberg", "James Cobon-Kerr", "Tara Sainath", "Maribeth Rauh", "Sayed Hadi Hashemi", "Richard Ives", "Yana Hasson", "Eric Noland", "Yuan Cao", "Nathan Byrd", "Le Hou", "Qingze Wang", "Thibault Sottiaux", "Michela Paganini", "Jean-Baptiste Lespiau", "Alexandre Moufarek", "Samer Hassan", "Kaushik Shivakumar", "Joost van Amersfoort", "Amol Mandhane", "Pratik Joshi", "Anirudh Goyal", "Matthew Tung", "Andrew Brock", "Hannah Sheahan", "Vedant Misra", "Cheng Li", "Nemanja Rakićević", "Mostafa Dehghani", "Fangyu Liu", "Sid Mittal", "Junhyuk Oh", "Seb Noury", "Eren Sezener", "Fantine Huot", "Matthew Lamm", "Nicola De Cao", "Charlie Chen", "Sidharth Mudgal", "Romina Stella", "Kevin Brooks", "Gautam Vasudevan", "Chenxi Liu", "Mainak Chain", "Nivedita Melinkeri", "Aaron Cohen", "Venus Wang", "Kristie Seymore", "Sergey Zubkov", "Rahul Goel", "Summer Yue", "Sai Krishnakumaran", "Brian Albert", "Nate Hurley", "Motoki Sano", "Anhad Mohananey", "Jonah Joughin", "Egor Filonov", "Tomasz Kępa", "Yomna Eldawy", "Jiawern Lim", "Rahul Rishi", "Shirin Badiezadegan", "Taylor Bos", "Jerry Chang", "Sanil Jain", "Sri Gayatri Sundara Padmanabhan", "Subha Puttagunta", "Kalpesh Krishna", "Leslie Baker", "Norbert Kalb", "Vamsi Bedapudi", "Adam Kurzrok", "Shuntong Lei", "Anthony Yu", "Oren Litvin", "Xiang Zhou", "Zhichun Wu", "Sam Sobell", "Andrea Siciliano", "Alan Papir", "Robby Neale", "Jonas Bragagnolo", "Tej Toor", "Tina Chen", "Valentin Anklin", "Feiran Wang", "Richie Feng", "Milad Gholami", "Kevin Ling", "Lijuan Liu", "Jules Walter", "Hamid Moghaddam", "Arun Kishore", "Jakub Adamek", "Tyler Mercado", "Jonathan Mallinson", "Siddhinita Wandekar", "Stephen Cagle", "Eran Ofek", "Guillermo Garrido", "Clemens Lombriser", "Maksim Mukha", "Botu Sun", "Hafeezul Rahman Mohammad", "Josip Matak", "Yadi Qian", "Vikas Peswani", "Pawel Janus", "Quan Yuan", "Leif Schelin", "Oana David", "Ankur Garg", "Yifan He", "Oleksii Duzhyi", "Anton Älgmyr", "Timothée Lottaz", "Qi Li", "Vikas Yadav", "Luyao Xu", "Alex Chinien", "Rakesh Shivanna", "Aleksandr Chuklin", "Josie Li", "Carrie Spadine", "Travis Wolfe", "Kareem Mohamed", "Subhabrata Das", "Zihang Dai", "Kyle He", "Daniel von Dincklage", "Shyam Upadhyay", "Akanksha Maurya", "Luyan Chi", "Sebastian Krause", "Khalid Salama", "Pam G Rabinovitch", "Pavan Kumar Reddy M", "Aarush Selvan", "Mikhail Dektiarev", "Golnaz Ghiasi", "Erdem Guven", "Himanshu Gupta", "Boyi Liu", "Deepak Sharma", "Idan Heimlich Shtacher", "Shachi Paul", "Oscar Akerlund", "François-Xavier Aubet", "Terry Huang", "Chen Zhu", "Eric Zhu", "Elico Teixeira", "Matthew Fritze", "Francesco Bertolini", "Liana-Eleonora Marinescu", "Martin Bölle", "Dominik Paulus", "Khyatti Gupta", "Tejasi Latkar", "Max Chang", "Jason Sanders", "Roopa Wilson", "Xuewei Wu", "Yi-Xuan Tan", "Lam Nguyen Thiet", "Tulsee Doshi", "Sid Lall", "Swaroop Mishra", "Wanming Chen", "Thang Luong", "Seth Benjamin", "Jasmine Lee", "Ewa Andrejczuk", "Dominik Rabiej", "Vipul Ranjan", "Krzysztof Styrc", "Pengcheng Yin", "Jon Simon", "Malcolm Rose Harriott", "Mudit Bansal", "Alexei Robsky", "Geoff Bacon", "David Greene", "Daniil Mirylenka", "Chen Zhou", "Obaid Sarvana", "Abhimanyu Goyal", "Samuel Andermatt", "Patrick Siegler", "Ben Horn", "Assaf Israel", "Francesco Pongetti", "Chih-Wei \"Louis\" Chen", "Marco Selvatici", "Pedro Silva", "Kathie Wang", "Jackson Tolins", "Kelvin Guu", "Roey Yogev", "Xiaochen Cai", "Alessandro Agostini", "Maulik Shah", "Hung Nguyen", "Noah Ó Donnaile", "Sébastien Pereira", "Linda Friso", "Adam Stambler", "Adam Kurzrok", "Chenkai Kuang", "Yan Romanikhin", "Mark Geller", "ZJ Yan", "Kane Jang", "Cheng-Chun Lee", "Wojciech Fica", "Eric Malmi", "Qijun Tan", "Dan Banica", "Daniel Balle", "Ryan Pham", "Yanping Huang", "Diana Avram", "Hongzhi Shi", "Jasjot Singh", "Chris Hidey", "Niharika Ahuja", "Pranab Saxena", "Dan Dooley", "Srividya Pranavi Potharaju", "Eileen O'Neill", "Anand Gokulchandran", "Ryan Foley", "Kai Zhao", "Mike Dusenberry", "Yuan Liu", "Pulkit Mehta", "Ragha Kotikalapudi", "Chalence Safranek-Shrader", "Andrew Goodman", "Joshua Kessinger", "Eran Globen", "Prateek Kolhar", "Chris Gorgolewski", "Ali Ibrahim", "Yang Song", "Ali Eichenbaum", "Thomas Brovelli", "Sahitya Potluri", "Preethi Lahoti", "Cip Baetu", "Ali Ghorbani", "Charles Chen", "Andy Crawford", "Shalini Pal", "Mukund Sridhar", "Petru Gurita", "Asier Mujika", "Igor Petrovski", "Pierre-Louis Cedoz", "Chenmei Li", "Shiyuan Chen", "Niccolò Dal Santo", "Siddharth Goyal", "Jitesh Punjabi", "Karthik Kappaganthu", "Chester Kwak", "Pallavi LV", "Sarmishta Velury", "Himadri Choudhury", "Jamie Hall", "Premal Shah", "Ricardo Figueira", "Matt Thomas", "Minjie Lu", "Ting Zhou", "Chintu Kumar", "Thomas Jurdi", "Sharat Chikkerur", "Yenai Ma", "Adams Yu", "Soo Kwak", "Victor Ähdel", "Sujeevan Rajayogam", "Travis Choma", "Fei Liu", "Aditya Barua", "Colin Ji", "Ji Ho Park", "Vincent Hellendoorn", "Alex Bailey", "Taylan Bilal", "Huanjie Zhou", "Mehrdad Khatir", "Charles Sutton", "Wojciech Rzadkowski", "Fiona Macintosh", "Roopali Vij", "Konstantin Shagin", "Paul Medina", "Chen Liang", "Jinjing Zhou", "Pararth Shah", "Yingying Bi", "Attila Dankovics", "Shipra Banga", "Sabine Lehmann", "Marissa Bredesen", "Zifan Lin", "John Eric Hoffmann", "Jonathan Lai", "Raynald Chung", "Kai Yang", "Nihal Balani", "Arthur Bražinskas", "Andrei Sozanschi", "Matthew Hayes", "Héctor Fernández Alcalde", "Peter Makarov", "Will Chen", "Antonio Stella", "Liselotte Snijders", "Michael Mandl", "Ante Kärrman", "Paweł Nowak", "Xinyi Wu", "Alex Dyck", "Krishnan Vaidyanathan", "Raghavender R", "Jessica Mallet", "Mitch Rudominer", "Eric Johnston", "Sushil Mittal", "Akhil Udathu", "Janara Christensen", "Vishal Verma", "Zach Irving", "Andreas Santucci", "Gamaleldin Elsayed", "Elnaz Davoodi", "Marin Georgiev", "Ian Tenney", "Nan Hua", "Geoffrey Cideron", "Edouard Leurent", "Mahmoud Alnahlawi", "Ionut Georgescu", "Nan Wei", "Ivy Zheng", "Dylan Scandinaro", "Heinrich Jiang", "Jasper Snoek", "Mukund Sundararajan", "Xuezhi Wang", "Zack Ontiveros", "Itay Karo", "Jeremy Cole", "Vinu Rajashekhar", "Lara Tumeh", "Eyal Ben-David", "Rishub Jain", "Jonathan Uesato", "Romina Datta", "Oskar Bunyan", "Shimu Wu", "John Zhang", "Piotr Stanczyk", "Ye Zhang", "David Steiner", "Subhajit Naskar", "Michael Azzam", "Matthew Johnson", "Adam Paszke", "Chung-Cheng Chiu", "Jaume Sanchez Elias", "Afroz Mohiuddin", "Faizan Muhammad", "Jin Miao", "Andrew Lee", "Nino Vieillard", "Jane Park", "Jiageng Zhang", "Jeff Stanway", "Drew Garmon", "Abhijit Karmarkar", "Zhe Dong", "Jong Lee", "Aviral Kumar", "Luowei Zhou", "Jonathan Evens", "William Isaac", "Geoffrey Irving", "Edward Loper", "Michael Fink", "Isha Arkatkar", "Nanxin Chen", "Izhak Shafran", "Ivan Petrychenko", "Zhe Chen", "Johnson Jia", "Anselm Levskaya", "Zhenkai Zhu", "Peter Grabowski", "Yu Mao", "Alberto Magni", "Kaisheng Yao", "Javier Snaider", "Norman Casagrande", "Evan Palmer", "Paul Suganthan", "Alfonso Castaño", "Irene Giannoumis", "Wooyeol Kim", "Mikołaj Rybiński", "Ashwin Sreevatsa", "Jennifer Prendki", "David Soergel", "Adrian Goedeckemeyer", "Willi Gierke", "Mohsen Jafari", "Meenu Gaba", "Jeremy Wiesner", "Diana Gage Wright", "Yawen Wei", "Harsha Vashisht", "Yana Kulizhskaya", "Jay Hoover", "Maigo Le", "Lu Li", "Chimezie Iwuanyanwu", "Lu Liu", "Kevin Ramirez", "Andrey Khorlin", "Albert Cui", "Tian LIN", "Marcus Wu", "Ricardo Aguilar", "Keith Pallo", "Abhishek Chakladar", "Ginger Perng", "Elena Allica Abellan", "Mingyang Zhang", "Ishita Dasgupta", "Nate Kushman", "Ivo Penchev", "Alena Repina", "Xihui Wu", "Tom van der Weide", "Priya Ponnapalli", "Caroline Kaplan", "Jiri Simsa", "Shuangfeng Li", "Olivier Dousse", "Fan Yang", "Jeff Piper", "Nathan Ie", "Rama Pasumarthi", "Nathan Lintz", "Anitha Vijayakumar", "Daniel Andor", "Pedro Valenzuela", "Minnie Lui", "Cosmin Paduraru", "Daiyi Peng", "Katherine Lee", "Shuyuan Zhang", "Somer Greene", "Duc Dung Nguyen", "Paula Kurylowicz", "Cassidy Hardin", "Lucas Dixon", "Lili Janzer", "Kiam Choo", "Ziqiang Feng", "Biao Zhang", "Achintya Singhal", "Dayou Du", "Dan McKinnon", "Natasha Antropova", "Tolga Bolukbasi", "Orgad Keller", "David Reid", "Daniel Finchelstein", "Maria Abi Raad", "Remi Crocker", "Peter Hawkins", "Robert Dadashi", "Colin Gaffney", "Ken Franko", "Anna Bulanova", "Rémi Leblond", "Shirley Chung", "Harry Askham", "Luis C. Cobo", "Kelvin Xu", "Felix Fischer", "Jun Xu", "Christina Sorokin", "Chris Alberti", "Chu-Cheng Lin", "Colin Evans", "Alek Dimitriev", "Hannah Forbes", "Dylan Banarse", "Zora Tung", "Mark Omernick", "Colton Bishop", "Rachel Sterneck", "Rohan Jain", "Jiawei Xia", "Ehsan Amid", "Francesco Piccinno", "Xingyu Wang", "Praseem Banzal", "Daniel J. Mankowitz", "Alex Polozov", "Victoria Krakovna", "Sasha Brown", "MohammadHossein Bateni", "Dennis Duan", "Vlad Firoiu", "Meghana Thotakuri", "Tom Natan", "Matthieu Geist", "Ser tan Girgin", "Hui Li", "Jiayu Ye", "Ofir Roval", "Reiko Tojo", "Michael Kwong", "James Lee-Thorp", "Christopher Yew", "Danila Sinopalnikov", "Sabela Ramos", "John Mellor", "Abhishek Sharma", "Kathy Wu", "David Miller", "Nicolas Sonnerat", "Denis Vnukov", "Rory Greig", "Jennifer Beattie", "Emily Caveness", "Libin Bai", "Julian Eisenschlos", "Alex Korchemniy", "Tomy Tsai", "Mimi Jasarevic", "Weize Kong", "Phuong Dao", "Zeyu Zheng", "Frederick Liu", "Fan Yang", "Rui Zhu", "Tian Huey Teh", "Jason Sanmiya", "Evgeny Gladchenko", "Nejc Trdin", "Daniel Toyama", "Evan Rosen", "Sasan Tavakkol", "Linting Xue", "Chen Elkind", "Oliver Woodman", "John Carpenter", "George Papamakarios", "Rupert Kemp", "Sushant Kafle", "Tanya Grunina", "Rishika Sinha", "Alice Talbert", "Diane Wu", "Denese Owusu-Afriyie", "Cosmo Du", "Chloe Thornton", "Jordi Pont-Tuset", "Pradyumna Narayana", "Jing Li", "Saaber Fatehi", "John Wieting", "Omar Ajmeri", "Benigno Uria", "Yeongil Ko", "Laura Knight", "Amélie Héliou", "Ning Niu", "Shane Gu", "Chenxi Pang", "Yeqing Li", "Nir Levine", "Ariel Stolovich", "Rebeca Santamaria-Fernandez", "Sonam Goenka", "Wenny Yustalim", "Robin Strudel", "Ali Elqursh", "Charlie Deck", "Hyo Lee", "Zonglin Li", "Kyle Levin", "Raphael Hoffmann", "Dan Holtmann-Rice", "Olivier Bachem", "Sho Arora", "Christy Koh", "Soheil Hassas Yeganeh", "Siim Põder", "Mukarram Tariq", "Yanhua Sun", "Lucian Ionita", "Mojtaba Seyedhosseini", "Pouya Tafti", "Zhiyu Liu", "Anmol Gulati", "Jasmine Liu", "Xinyu Ye", "Bart Chrzaszcz", "Lily Wang", "Nikhil Sethi", "Tianrun Li", "Ben Brown", "Shreya Singh", "Wei Fan", "Aaron Parisi", "Joe Stanton", "Vinod Koverkathu", "Christopher A. Choquette-Choo", "Yunjie Li", "TJ Lu", "Abe Ittycheriah", "Prakash Shroff", "Mani Varadarajan", "Sanaz Bahargam", "Rob Willoughby", "David Gaddy", "Guillaume Desjardins", "Marco Cornero", "Brona Robenek", "Bhavishya Mittal", "Ben Albrecht", "Ashish Shenoy", "Fedor Moiseev", "Henrik Jacobsson", "Alireza Ghaffarkhah", "Morgane Rivière", "Alanna Walton", "Clément Crepy", "Alicia Parrish", "Zongwei Zhou", "Clement Farabet", "Carey Radebaugh", "Praveen Srinivasan", "Claudia van der Salm", "Andreas Fidjeland", "Salvatore Scellato", "Eri Latorre-Chimoto", "Hanna Klimczak-Plucińska", "David Bridson", "Dario de Cesare", "Tom Hudson", "Piermaria Mendolicchio", "Lexi Walker", "Alex Morris", "Matthew Mauger", "Alexey Guseynov", "Alison Reid", "Seth Odoom", "Lucia Loher", "Victor Cotruta", "Madhavi Yenugula", "Dominik Grewe", "Anastasia Petrushkina", "Tom Duerig", "Antonio Sanchez", "Steve Yadlowsky", "Amy Shen", "Amir Globerson", "Lynette Webb", "Sahil Dua", "Dong Li", "Surya Bhupatiraju", "Dan Hurt", "Haroon Qureshi", "Ananth Agarwal", "Tomer Shani", "Matan Eyal", "Anuj Khare", "Shreyas Rammohan Belle", "Lei Wang", "Chetan Tekur", "Mihir Sanjay Kale", "Jinliang Wei", "Ruoxin Sang", "Brennan Saeta", "Tyler Liechty", "Yi Sun", "Yao Zhao", "Stephan Lee", "Pandu Nayak", "Doug Fritz", "Manish Reddy Vuyyuru", "John Aslanides", "Nidhi Vyas", "Martin Wicke", "Xiao Ma", "Evgenii Eltyshev", "Nina Martin", "Hardie Cate", "James Manyika", "Keyvan Amiri", "Yelin Kim", "Xi Xiong", "Kai Kang", "Florian Luisier", "Nilesh Tripuraneni", "David Madras", "Mandy Guo", "Austin Waters", "Oliver Wang", "Joshua Ainslie", "Jason Baldridge", "Han Zhang", "Garima Pruthi", "Jakob Bauer", "Feng Yang", "Riham Mansour", "Jason Gelman", "Yang Xu", "George Polovets", "Ji Liu", "Honglong Cai", "Warren Chen", "XiangHai Sheng", "Emily Xue", "Sherjil Ozair", "Christof Angermueller", "Xiaowei Li", "Anoop Sinha", "Weiren Wang", "Julia Wiesinger", "Emmanouil Koukoumidis", "Yuan Tian", "Anand Iyer", "Madhu Gurumurthy", "Mark Goldenson", "Parashar Shah", "MK Blake", "Hongkun Yu", "Anthony Urbanowicz", "Jennimaria Palomaki", "Chrisantha Fernando", "Ken Durden", "Harsh Mehta", "Nikola Momchev", "Elahe Rahimtoroghi", "Maria Georgaki", "Amit Raul", "Sebastian Ruder", "Morgan Redshaw", "Jinhyuk Lee", "Denny Zhou", "Komal Jalan", "Dinghua Li", "Blake Hechtman", "Parker Schuh", "Milad Nasr", "Kieran Milan", "Vladimir Mikulik", "Juliana Franco", "Tim Green", "Nam Nguyen", "Joe Kelley", "Aroma Mahendru", "Andrea Hu", "Joshua Howland", "Ben Vargas", "Jeffrey Hui", "Kshitij Bansal", "Vikram Rao", "Rakesh Ghiya", "Emma Wang", "Ke Ye", "Jean Michel Sarr", "Melanie Moranski Preston", "Madeleine Elish", "Steve Li", "Aakash Kaku", "Jigar Gupta", "Ice Pasupat", "Da-Cheng Juan", "Milan Someswar", "Tejvi M.", "Xinyun Chen", "Aida Amini", "Alex Fabrikant", "Eric Chu", "Xuanyi Dong", "Amruta Muthal", "Senaka Buthpitiya", "Sarthak Jauhari", "Nan Hua", "Urvashi Khandelwal", "Ayal Hitron", "Jie Ren", "Larissa Rinaldi", "Shahar Drath", "Avigail Dabush", "Nan-Jiang Jiang", "Harshal Godhia", "Uli Sachs", "Anthony Chen", "Yicheng Fan", "Hagai Taitelbaum", "Hila Noga", "Zhuyun Dai", "James Wang", "Chen Liang", "Jenny Hamer", "Chun-Sung Ferng", "Chenel Elkind", "Aviel Atias", "Paulina Lee", "Vít Listík", "Mathias Carlen", "Jan van de Kerkhof", "Marcin Pikus", "Krunoslav Zaher", "Paul Müller", "Sasha Zykova", "Richard Stefanec", "Vitaly Gatsko", "Christoph Hirnschall", "Ashwin Sethi", "Xingyu Federico Xu", "Chetan Ahuja", "Beth Tsai", "Anca Stefanoiu", "Bo Feng", "Keshav Dhandhania", "Manish Katyal", "Akshay Gupta", "Atharva Parulekar", "Divya Pitta", "Jing Zhao", "Vivaan Bhatia", "Yashodha Bhavnani", "Omar Alhadlaq", "Xiaolin Li", "Peter Danenberg", "Dennis Tu", "Alex Pine", "Vera Filippova", "Abhipso Ghosh", "Ben Limonchik", "Bhargava Urala", "Chaitanya Krishna Lanka", "Derik Clive", "Yi Sun", "Edward Li", "Hao Wu", "Kevin Hongtongsak", "Ianna Li", "Kalind Thakkar", "Kuanysh Omarov", "Kushal Majmundar", "Michael Alverson", "Michael Kucharski", "Mohak Patel", "Mudit Jain", "Maksim Zabelin", "Paolo Pelagatti", "Rohan Kohli", "Saurabh Kumar", "Joseph Kim", "Swetha Sankar", "Vineet Shah", "Lakshmi Ramachandruni", "Xiangkai Zeng", "Ben Bariach", "Laura Weidinger", "Tu Vu", "Alek Andreev", "Antoine He", "Kevin Hui", "Sheleem Kashem", "Amar Subramanya", "Sissie Hsiao", "Demis Hassabis", "Koray Kavukcuoglu", "Adam Sadovsky", "Quoc Le", "Trevor Strohman", "Yonghui Wu", "Slav Petrov", "Jeffrey Dean", "Oriol Vinyals"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2312.11805", "doi": null, "topics": ["Other research"], "description": "This report introduces a new family of multimodal models, Gemini, that exhibit remarkable capabilities across image, audio, video, and text understanding.", "description_method": "abstract-excerpt", "abstract": "This report introduces a new family of multimodal models, Gemini, that exhibit remarkable capabilities across image, audio, video, and text understanding. The Gemini family consists of Ultra, Pro, and Nano sizes, suitable for applications ranging from complex reasoning tasks to on-device memory-constrained use-cases. Evaluation on a broad range of benchmarks shows that our most-capable Gemini Ultra model advances the state of the art in 30 of 32 of these benchmarks - notably being the first model to achieve human-expert performance on the well-studied exam benchmark MMLU, and improving the state of the art in every one of the 20 multimodal benchmarks we examined. We believe that the new capabilities of the Gemini family in cross-modal reasoning and language understanding will enable a wide variety of use cases. We discuss our approach toward post-training and deploying Gemini models responsibly to users through services including Gemini, Gemini Advanced, Google AI Studio, and Cloud Vertex AI.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2024-02", "2024-03", "2024-04", "2024-05", "2025-12"], "last_mentioned": "2025-12-06T11:15:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.01032", "slug": "e56b171aa093c56e", "title": "Repeat After Me: Transformers are Better than State Space Models at Copying", "authors": ["Samy Jelassi", "David Brandfonbrener", "Sham M. Kakade", "Eran Malach"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.01032", "doi": null, "topics": ["Language models", "Machine learning"], "description": "Transformers are the dominant architecture for sequence modeling, but there is growing interest in models that use a fixed-size latent state that does not depend on the sequence length, which we refer to as \"generalized state space models\" (GSSMs).", "description_method": "abstract-excerpt", "abstract": "Transformers are the dominant architecture for sequence modeling, but there is growing interest in models that use a fixed-size latent state that does not depend on the sequence length, which we refer to as \"generalized state space models\" (GSSMs). In this paper we show that while GSSMs are promising in terms of inference-time efficiency, they are limited compared to transformer models on tasks that require copying from the input context. We start with a theoretical analysis of the simple task of string copying and prove that a two layer transformer can copy strings of exponential length while GSSMs are fundamentally limited by their fixed-size latent state. Empirically, we find that transformers outperform GSSMs in terms of efficiency and generalization on synthetic tasks that require copying the context. Finally, we evaluate pretrained large language models and find that transformer models dramatically outperform state space models at copying and retrieving information from context. Taken together, these results suggest a fundamental gap between transformers and GSSMs on tasks of practical interest.", "score": 0.0, "mention_count": 6, "user_count": 4, "months": ["2024-02", "2024-04", "2024-06", "2024-10"], "last_mentioned": "2024-10-03T20:36:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.03239", "slug": "d70ff15db535e0e8", "title": "Bluesky and the AT Protocol: Usable Decentralized Social Media", "authors": ["Martin Kleppmann", "Paul Frazee", "Jake Gold", "Jay Graber", "Daniel Holmgren", "Devin Ivy", "Jeromy Johnson", "Bryan Newbold", "Jaz Volpert"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.03239", "doi": "10.1145/3694809.3700740", "topics": ["Human–computer interaction", "Algorithms"], "description": "Bluesky is a new social network built upon the AT Protocol, a decentralized foundation for public social media.", "description_method": "abstract-excerpt", "abstract": "Bluesky is a new social network built upon the AT Protocol, a decentralized foundation for public social media. It was launched in private beta in February 2023, and has grown to over 10 million registered users by October 2024. In this paper we introduce the architecture of Bluesky and the AT Protocol, and explain how the technical design of Bluesky is informed by our goals: to enable decentralization by having multiple interoperable providers for every part of the system; to make it easy for users to switch providers; to give users agency over the content they see; and to provide a simple user experience that does not burden users with complexity arising from the system's decentralized nature. The system's openness allows anybody to contribute to content moderation and community management, and we invite the research community to use Bluesky as a dataset and testing ground for new approaches in social media moderation.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2024-02", "2024-09", "2025-04", "2026-01"], "last_mentioned": "2026-01-27T21:33:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.05120", "slug": "e81041468ceb8bf5", "title": "More Agents Is All You Need", "authors": ["Junyou Li", "Qin Zhang", "Yangbin Yu", "Qiang Fu", "Deheng Ye"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.05120", "doi": null, "topics": ["Language models"], "description": "We find that, simply via a sampling-and-voting method, the performance of large language models (LLMs) scales with the number of agents instantiated.", "description_method": "abstract-excerpt", "abstract": "We find that, simply via a sampling-and-voting method, the performance of large language models (LLMs) scales with the number of agents instantiated. Also, this method, termed as Agent Forest, is orthogonal to existing complicated methods to further enhance LLMs, while the degree of enhancement is correlated to the task difficulty. We conduct comprehensive experiments on a wide range of LLM benchmarks to verify the presence of our finding, and to study the properties that can facilitate its occurrence. Our code is publicly available at: https://github.com/MoreAgentsIsAllYouNeed/AgentForest", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2024-03", "2024-04", "2024-08", "2024-09", "2025-01"], "last_mentioned": "2025-01-09T16:36:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2403.09629", "slug": "385559d74aab7763", "title": "Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking", "authors": ["Eric Zelikman", "Georges Harik", "Yijia Shao", "Varuna Jayasiri", "Nick Haber", "Noah D. Goodman"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2403.09629", "doi": null, "topics": ["Language models", "Networks", "Machine learning"], "description": "When writing and talking, people sometimes pause to think.", "description_method": "abstract-excerpt", "abstract": "When writing and talking, people sometimes pause to think. Although reasoning-focused works have often framed reasoning as a method of answering questions or completing agentic tasks, reasoning is implicit in almost all written text. For example, this applies to the steps not stated between the lines of a proof or to the theory of mind underlying a conversation. In the Self-Taught Reasoner (STaR, Zelikman et al. 2022), useful thinking is learned by inferring rationales from few-shot examples in question-answering and learning from those that lead to a correct answer. This is a highly constrained setting -- ideally, a language model could instead learn to infer unstated rationales in arbitrary text. We present Quiet-STaR, a generalization of STaR in which LMs learn to generate rationales at each token to explain future text, improving their predictions. We address key challenges, including 1) the computational cost of generating continuations, 2) the fact that the LM does not initially know how to generate or use internal thoughts, and 3) the need to predict beyond individual next tokens. To resolve these, we propose a tokenwise parallel sampling algorithm, using learnable tokens indicating a thought's start and end, and an extended teacher-forcing technique. Encouragingly, generated rationales disproportionately help model difficult-to-predict tokens and improve the LM's ability to directly answer difficult questions. In particular, after continued pretraining of an LM on a corpus of internet text with Quiet-STaR, we find zero-shot improvements on GSM8K (5.9%$\\rightarrow$10.9%) and CommonsenseQA (36.3%$\\rightarrow$47.2%) and observe a perplexity improvement of difficult tokens in natural text. Crucially, these improvements require no fine-tuning on these tasks. Quiet-STaR marks a step towards LMs that can learn to reason in a more general and scalable way.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2024-03", "2024-04", "2024-05", "2024-08", "2024-09"], "last_mentioned": "2024-09-16T13:27:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2403.15498", "slug": "0d8bd8322f943ef2", "title": "Emergent World Models and Latent Variable Estimation in Chess-Playing Language Models", "authors": ["Adam Karvonen"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2403.15498", "doi": null, "topics": ["Language models", "Statistics & probability"], "description": "Language models have shown unprecedented capabilities, sparking debate over the source of their performance.", "description_method": "abstract-excerpt", "abstract": "Language models have shown unprecedented capabilities, sparking debate over the source of their performance. Is it merely the outcome of learning syntactic patterns and surface level statistics, or do they extract semantics and a world model from the text? Prior work by Li et al. investigated this by training a GPT model on synthetic, randomly generated Othello games and found that the model learned an internal representation of the board state. We extend this work into the more complex domain of chess, training on real games and investigating our model's internal representations using linear probes and contrastive activations. The model is given no a priori knowledge of the game and is solely trained on next character prediction, yet we find evidence of internal representations of board state. We validate these internal representations by using them to make interventions on the model's activations and edit its internal board state. Unlike Li et al's prior synthetic dataset approach, our analysis finds that the model also learns to estimate latent variables like player skill to better predict the next character. We derive a player skill vector and add it to the model, improving the model's win rate by up to 2.6 times.", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2024-03", "2024-09", "2025-07", "2025-10", "2026-02"], "last_mentioned": "2026-02-19T22:52:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2412.09871", "slug": "beb9b33bee96088b", "title": "Byte Latent Transformer: Patches Scale Better Than Tokens", "authors": ["Artidoro Pagnoni", "Ram Pasunuru", "Pedro Rodriguez", "John Nguyen", "Benjamin Muller", "Margaret Li", "Chunting Zhou", "Lili Yu", "Jason Weston", "Luke Zettlemoyer", "Gargi Ghosh", "Mike Lewis", "Ari Holtzman", "Srinivasan Iyer"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2412.09871", "doi": null, "topics": ["Language models", "Statistics & probability", "Machine learning"], "description": "We introduce the Byte Latent Transformer (BLT), a new byte-level LLM architecture that, for the first time, matches tokenization-based LLM performance at scale with significant improvements in inference efficiency and robustness.", "description_method": "abstract-excerpt", "abstract": "We introduce the Byte Latent Transformer (BLT), a new byte-level LLM architecture that, for the first time, matches tokenization-based LLM performance at scale with significant improvements in inference efficiency and robustness. BLT encodes bytes into dynamically sized patches, which serve as the primary units of computation. Patches are segmented based on the entropy of the next byte, allocating more compute and model capacity where increased data complexity demands it. We present the first FLOP controlled scaling study of byte-level models up to 8B parameters and 4T training bytes. Our results demonstrate the feasibility of scaling models trained on raw bytes without a fixed vocabulary. Both training and inference efficiency improve due to dynamically selecting long patches when data is predictable, along with qualitative improvements on reasoning and long tail generalization. Overall, for fixed inference costs, BLT shows significantly better scaling than tokenization-based models, by simultaneously growing both patch and model size.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2024-12", "2025-01", "2025-02", "2025-05", "2026-06"], "last_mentioned": "2026-06-20T08:47:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2502.05171", "slug": "24a6ab8657772906", "title": "Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach", "authors": ["Jonas Geiping", "Sean McLeish", "Neel Jain", "John Kirchenbauer", "Siddharth Singh", "Brian R. Bartoldson", "Bhavya Kailkhura", "Abhinav Bhatele", "Tom Goldstein"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2502.05171", "doi": null, "topics": ["Language models"], "description": "We study a novel language model architecture that is capable of scaling test-time computation by implicitly reasoning in latent space.", "description_method": "abstract-excerpt", "abstract": "We study a novel language model architecture that is capable of scaling test-time computation by implicitly reasoning in latent space. Our model works by iterating a recurrent block, thereby unrolling to arbitrary depth at test-time. This stands in contrast to mainstream reasoning models that scale up compute by producing more tokens. Unlike approaches based on chain-of-thought, our approach does not require any specialized training data, can work with small context windows, and can capture types of reasoning that are not easily represented in words. We scale a proof-of-concept model to 3.5 billion parameters and 800 billion tokens. We show that the resulting model can improve its performance on reasoning benchmarks, sometimes dramatically, up to a computation load equivalent to 50 billion parameters.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2025-02", "2025-07", "2025-10", "2025-12", "2026-03", "2026-04"], "last_mentioned": "2026-04-07T21:56:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2502.15840", "slug": "3e821cc90ae17481", "title": "Vending-Bench: A Benchmark for Long-Term Coherence of Autonomous Agents", "authors": ["Axel Backlund", "Lukas Petersson"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2502.15840", "doi": null, "topics": ["Language models", "Robotics", "Psychology"], "description": "While Large Language Models (LLMs) can exhibit impressive proficiency in isolated, short-term tasks, they often fail to maintain coherent performance over longer time horizons.", "description_method": "abstract-excerpt", "abstract": "While Large Language Models (LLMs) can exhibit impressive proficiency in isolated, short-term tasks, they often fail to maintain coherent performance over longer time horizons. In this paper, we present Vending-Bench, a simulated environment designed to specifically test an LLM-based agent's ability to manage a straightforward, long-running business scenario: operating a vending machine. Agents must balance inventories, place orders, set prices, and handle daily fees - tasks that are each simple but collectively, over long horizons (>20M tokens per run) stress an LLM's capacity for sustained, coherent decision-making. Our experiments reveal high variance in performance across multiple LLMs: Claude 3.5 Sonnet and o3-mini manage the machine well in most runs and turn a profit, but all models have runs that derail, either through misinterpreting delivery schedules, forgetting orders, or descending into tangential \"meltdown\" loops from which they rarely recover. We find no clear correlation between failures and the point at which the model's context window becomes full, suggesting that these breakdowns do not stem from memory limits. Apart from highlighting the high variance in performance over long time horizons, Vending-Bench also tests models' ability to acquire capital, a necessity in many hypothetical dangerous AI scenarios. We hope the benchmark can help in preparing for the advent of stronger AI systems.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2025-05", "2025-06", "2025-07", "2025-10"], "last_mentioned": "2025-10-28T18:36:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2506.21734", "slug": "3604312101ff09ab", "title": "Hierarchical Reasoning Model", "authors": ["Guan Wang", "Jin Li", "Yuhao Sun", "Xing Chen", "Changling Liu", "Yue Wu", "Meng Lu", "Sen Song", "Yasin Abbasi Yadkori"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2506.21734", "doi": null, "topics": ["Psychology", "Neuroscience", "Language models"], "description": "Reasoning, the process of devising and executing complex goal-oriented action sequences, remains a critical challenge in AI.", "description_method": "abstract-excerpt", "abstract": "Reasoning, the process of devising and executing complex goal-oriented action sequences, remains a critical challenge in AI. Current large language models (LLMs) primarily employ Chain-of-Thought (CoT) techniques, which suffer from brittle task decomposition, extensive data requirements, and high latency. Inspired by the hierarchical and multi-timescale processing in the human brain, we propose the Hierarchical Reasoning Model (HRM), a novel recurrent architecture that attains significant computational depth while maintaining both training stability and efficiency. HRM executes sequential reasoning tasks in a single forward pass without explicit supervision of the intermediate process, through two interdependent recurrent modules: a high-level module responsible for slow, abstract planning, and a low-level module handling rapid, detailed computations. With only 27 million parameters, HRM achieves exceptional performance on complex reasoning tasks using only 1000 training samples. The model operates without pre-training or CoT data, yet achieves nearly perfect performance on challenging tasks including complex Sudoku puzzles and optimal path finding in large mazes. Furthermore, HRM outperforms much larger models with significantly longer context windows on the Abstraction and Reasoning Corpus (ARC), a key benchmark for measuring artificial general intelligence capabilities. These results underscore HRM's potential as a transformative advancement toward universal computation and general-purpose reasoning systems.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2025-07", "2025-08", "2025-10"], "last_mentioned": "2025-10-07T19:36:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2508.09101", "slug": "34536f1c43ada96b", "title": "AutoCodeBench: Large Language Models are Automatic Code Benchmark Generators", "authors": ["Jason Chou", "Ao Liu", "Yuchi Deng", "Zhiying Zeng", "Tao Zhang", "Haotian Zhu", "Jianwei Cai", "Yue Mao", "Chenchen Zhang", "Lingyun Tan", "Ziyan Xu", "Bohui Zhai", "Hengyi Liu", "Speed Zhu", "Wiggin Zhou", "Fengzong Lian"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2508.09101", "doi": null, "topics": ["Language models", "Programming languages", "Distributed systems"], "description": "Large Language Models (LLMs) have demonstrated remarkable capabilities across various domains, with code generation emerging as a key area of focus.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) have demonstrated remarkable capabilities across various domains, with code generation emerging as a key area of focus. While numerous benchmarks have been proposed to evaluate their code generation abilities, these benchmarks face several critical limitations. First, they often rely on manual annotations, which are time-consuming and difficult to scale across different programming languages and problem complexities. Second, most existing benchmarks focus primarily on Python, while the few multilingual benchmarks suffer from limited difficulty and uneven language distribution. To address these challenges, we propose AutoCodeGen, an automated method for generating high-difficulty multilingual code generation datasets without manual annotations. AutoCodeGen ensures the correctness and completeness of test cases by generating test inputs with LLMs and obtaining test outputs through a multilingual sandbox, while achieving high data quality through reverse-order problem generation and multiple filtering steps. Using this novel method, we introduce AutoCodeBench, a large-scale code generation benchmark comprising 3,920 problems evenly distributed across 20 programming languages. It is specifically designed to evaluate LLMs on challenging, diverse, and practical multilingual tasks. We evaluate over 30 leading open-source and proprietary LLMs on AutoCodeBench and its simplified version AutoCodeBench-Lite. The results show that even the most advanced LLMs struggle with the complexity, diversity, and multilingual nature of these tasks. Besides, we introduce AutoCodeBench-Complete, specifically designed for base models to assess their few-shot code generation capabilities. We hope the AutoCodeBench series will serve as a valuable resource and inspire the community to focus on more challenging and practical multilingual code generation scenarios.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2026-03", "2026-04", "2026-05", "2026-07"], "last_mentioned": "2026-07-18T17:25:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2511.19468", "slug": "4f5c4287d686117d", "title": "Towards a future space-based, highly scalable AI infrastructure system design", "authors": ["Blaise Agüera y Arcas", "Travis Beals", "Maria Biggs", "Jessica V. Bloom", "Thomas Fischbacher", "Konstantin Gromov", "Urs Köster", "Rishiraj Pravahan", "James Manyika"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2511.19468", "doi": null, "topics": ["Machine learning"], "description": "If AI is a foundational general-purpose technology, we should anticipate that demand for AI compute -- and energy -- will continue to grow.", "description_method": "abstract-excerpt", "abstract": "If AI is a foundational general-purpose technology, we should anticipate that demand for AI compute -- and energy -- will continue to grow. The Sun is by far the largest energy source in our solar system, and thus it warrants consideration how future AI infrastructure could most efficiently tap into that power. This work explores a scalable compute system for machine learning in space, using fleets of satellites equipped with solar arrays, inter-satellite links using free-space optics, and Google tensor processing unit (TPU) accelerator chips. To facilitate high-bandwidth, low-latency inter-satellite communication, the satellites would be flown in close proximity. We illustrate the basic approach to formation flight via an 81-satellite cluster of 1 km radius, and describe an approach for using high-precision ML-based models to control large-scale constellations. Trillium TPUs are radiation tested. They survive a total ionizing dose equivalent to a 5 year mission life without permanent failures, and are characterized for bit-flip errors. Launch costs are a critical part of overall system cost; a learning curve analysis suggests launch to low-Earth orbit (LEO) may reach $\\lesssim$\\$200/kg by the mid-2030s.", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2026-03", "2026-04", "2026-05", "2026-06", "2026-07"], "last_mentioned": "2026-07-26T09:31:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:math/0305282", "slug": "45e9ccf2b6ca3ab5", "title": "A Universal Approach to Self-Referential Paradoxes, Incompleteness and Fixed Points", "authors": ["Noson S. Yanofsky"], "year": 2003, "venue": "arXiv", "url": "https://arxiv.org/abs/math/0305282", "doi": "10.2178/bsl/1058448677", "topics": ["Mathematics", "Algorithms"], "description": "Following F.", "description_method": "abstract-excerpt", "abstract": "Following F. William Lawvere, we show that many self-referential paradoxes, incompleteness theorems and fixed point theorems fall out of the same simple scheme. We demonstrate these similarities by showing how this simple scheme encompasses the semantic paradoxes, and how they arise as diagonal arguments and fixed point theorems in logic, computability theory, complexity theory and formal language theory.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2015-11", "2020-04", "2020-05", "2021-07", "2023-06", "2023-09"], "last_mentioned": "2023-09-07T02:30:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:quant-ph/9801041", "slug": "8f3a6a5df602ae5e", "title": "Nonlinear quantum mechanics implies polynomial-time solution for NP-complete and #P problems", "authors": ["Daniel S. Abrams", "Seth Lloyd"], "year": 1998, "venue": "arXiv", "url": "https://arxiv.org/abs/quant-ph/9801041", "doi": "10.1103/PhysRevLett.81.3992", "topics": ["Physics", "Algorithms"], "description": "If quantum states exhibit small nonlinearities during time evolution, then quantum computers can be used to solve NP-complete problems in polynomial time.", "description_method": "abstract-excerpt", "abstract": "If quantum states exhibit small nonlinearities during time evolution, then quantum computers can be used to solve NP-complete problems in polynomial time. We provide algorithms that solve NP-complete and #P oracle problems by exploiting nonlinear quantum logic gates. It is argued that virtually any deterministic nonlinear quantum theory will include such gates, and the method is explicitly demonstrated using the Weinberg model of nonlinear quantum mechanics.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2010-05", "2010-08", "2011-02", "2015-09", "2026-06"], "last_mentioned": "2026-06-17T20:35:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/438900a", "slug": "a4207d88bd621aba", "title": "Internet encyclopaedias go head to head", "authors": ["Jim Giles"], "year": 2005, "venue": "Nature", "url": "https://doi.org/10.1038/438900a", "doi": "10.1038/438900a", "topics": ["Networks"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2021-02", "2021-05", "2022-03", "2022-07", "2022-10", "2025-04"], "last_mentioned": "2025-04-13T19:40:59Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nrg.2017.104", "slug": "9ca20d46de3413c7", "title": "The new genetics of intelligence", "authors": ["Robert Plomin", "Sophie von Stumm"], "year": 2018, "venue": "Nature Reviews Genetics", "url": "https://doi.org/10.1038/nrg.2017.104", "doi": "10.1038/nrg.2017.104", "topics": ["Psychology"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2019-07", "2021-10", "2022-06", "2022-08", "2024-01", "2024-04"], "last_mentioned": "2024-04-16T14:41:00Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41370-018-0109-y", "slug": "3e3709c8b360ea53", "title": "Serum concentrations of PFASs and exposure-related behaviors in African American and non-Hispanic white women", "authors": ["Katherine E. Boronow", "Julia Green Brody", "Laurel A. Schaider", "Graham F. Peaslee", "Laurie Havas", "Barbara A. Cohn"], "year": 2019, "venue": "Journal of Exposure Science &amp; Environmental Epidemiology", "url": "https://doi.org/10.1038/s41370-018-0109-y", "doi": "10.1038/s41370-018-0109-y", "topics": ["Psychology"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2019-01", "2022-08", "2023-06", "2023-07", "2023-09", "2024-02"], "last_mentioned": "2024-02-27T16:51:10Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41380-018-0070-0", "slug": "025498e2593fcc0f", "title": "Genetics of attention deficit hyperactivity disorder", "authors": ["Stephen V. Faraone", "Henrik Larsson"], "year": 2018, "venue": "Molecular Psychiatry", "url": "https://doi.org/10.1038/s41380-018-0070-0", "doi": "10.1038/s41380-018-0070-0", "topics": ["Biology & medicine", "Statistics & probability"], "description": "Abstract Decades of research show that genes play an vital role in the etiology of attention deficit hyperactivity disorder (ADHD) and its comorbidity with other disorders.", "description_method": "abstract-excerpt", "abstract": "Abstract Decades of research show that genes play an vital role in the etiology of attention deficit hyperactivity disorder (ADHD) and its comorbidity with other disorders. Family, twin, and adoption studies show that ADHD runs in families. ADHD’s high heritability of 74% motivated the search for ADHD susceptibility genes. Genetic linkage studies show that the effects of DNA risk variants on ADHD must, individually, be very small. Genome-wide association studies (GWAS) have implicated several genetic loci at the genome-wide level of statistical significance. These studies also show that about a third of ADHD’s heritability is due to a polygenic component comprising many common variants each having small effects. From studies of copy number variants we have also learned that the rare insertions or deletions account for part of ADHD’s heritability. These findings have implicated new biological pathways that may eventually have implications for treatment development.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2020-05", "2022-05", "2025-01", "2025-08", "2025-09"], "last_mentioned": "2025-09-22T10:00:54Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-019-10933-3", "slug": "95aedb817203cd17", "title": "Estimating the success of re-identifications in incomplete datasets using generative models", "authors": ["Luc Rocher", "Julien M. Hendrickx", "Yves-Alexandre de Montjoye"], "year": 2019, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-019-10933-3", "doi": "10.1038/s41467-019-10933-3", "topics": ["Machine learning", "Security & cryptography", "Psychology"], "description": "Abstract While rich medical, behavioral, and socio-demographic data are key to modern data-driven research, their collection and use raise legitimate privacy concerns.", "description_method": "abstract-excerpt", "abstract": "Abstract While rich medical, behavioral, and socio-demographic data are key to modern data-driven research, their collection and use raise legitimate privacy concerns. Anonymizing datasets through de-identification and sampling before sharing them has been the main tool used to address those concerns. We here propose a generative copula-based method that can accurately estimate the likelihood of a specific person to be correctly re-identified, even in a heavily incomplete dataset. On 210 populations, our method obtains AUC scores for predicting individual uniqueness ranging from 0.84 to 0.97, with low false-discovery rate. Using our model, we find that 99.98% of Americans would be correctly re-identified in any dataset using 15 demographic attributes. Our results suggest that even heavily sampled anonymized datasets are unlikely to satisfy the modern standards for anonymization set forth by GDPR and seriously challenge the technical and legal adequacy of the de-identification release-and-forget model.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2021-04", "2022-07", "2024-02", "2025-01", "2025-06"], "last_mentioned": "2025-06-25T15:33:06Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-020-20214-z", "slug": "d7c0e9beef99e69c", "title": "Transforming carbon dioxide into jet fuel using an organic combustion-synthesized Fe-Mn-K catalyst", "authors": ["Benzhen Yao", "Tiancun Xiao", "Ofentse A. Makgae", "Xiangyu Jie", "Sergio Gonzalez-Cortes", "Shaoliang Guan", "Angus I. Kirkland", "Jonathan R. Dilworth", "Hamid A. Al-Megren", "Saeed M. Alshihri", "Peter J. Dobson", "Gari P. Owen", "John M. Thomas", "Peter P. Edwards"], "year": 2020, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-020-20214-z", "doi": "10.1038/s41467-020-20214-z", "topics": ["Other research"], "description": "Abstract With mounting concerns over climate change, the utilisation or conversion of carbon dioxide into sustainable, synthetic hydrocarbons fuels, most notably for transportation purposes, continues to attract worldwide interest.", "description_method": "abstract-excerpt", "abstract": "Abstract With mounting concerns over climate change, the utilisation or conversion of carbon dioxide into sustainable, synthetic hydrocarbons fuels, most notably for transportation purposes, continues to attract worldwide interest. This is particularly true in the search for sustainable or renewable aviation fuels. These offer considerable potential since, instead of consuming fossil crude oil, the fuels are produced from carbon dioxide using sustainable renewable hydrogen and energy. We report here a synthetic protocol to the fixation of carbon dioxide by converting it directly into aviation jet fuel using novel, inexpensive iron-based catalysts. We prepare the Fe-Mn-K catalyst by the so-called Organic Combustion Method, and the catalyst shows a carbon dioxide conversion through hydrogenation to hydrocarbons in the aviation jet fuel range of 38.2%, with a yield of 17.2%, and a selectivity of 47.8%, and with an attendant low carbon monoxide (5.6%) and methane selectivity (10.4%). The conversion reaction also produces light olefins ethylene, propylene, and butenes, totalling a yield of 8.7%, which are important raw materials for the petrochemical industry and are presently also only obtained from fossil crude oil. As this carbon dioxide is extracted from air, and re-emitted from jet fuels when combusted in flight, the overall effect is a carbon-neutral fuel. This contrasts with jet fuels produced from hydrocarbon fossil sources where the combustion process unlocks the fossil carbon and places it into the atmosphere, in longevity, as aerial carbon - carbon dioxide.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2020-12", "2022-11", "2024-01"], "last_mentioned": "2024-01-24T03:08:40Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41564-020-0771-4", "slug": "75664ea9224c0b18", "title": "Evolutionary origins of the SARS-CoV-2 sarbecovirus lineage responsible for the COVID-19 pandemic", "authors": ["Maciej F. Boni", "Philippe Lemey", "Xiaowei Jiang", "Tommy Tsan-Yuk Lam", "Blair W. Perry", "Todd A. Castoe", "Andrew Rambaut", "David L. Robertson"], "year": 2020, "venue": "Nature Microbiology", "url": "https://doi.org/10.1038/s41564-020-0771-4", "doi": "10.1038/s41564-020-0771-4", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2020-11", "2020-12", "2021-03", "2021-05", "2023-06"], "last_mentioned": "2023-06-11T10:23:03Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41567-019-0663-9", "slug": "01ae7da8a398c883", "title": "Quantum superposition of molecules beyond 25 kDa", "authors": ["Yaakov Y. Fein", "Philipp Geyer", "Patrick Zwick", "Filip Kiałka", "Sebastian Pedalino", "Marcel Mayor", "Stefan Gerlich", "Markus Arndt"], "year": 2019, "venue": "Nature Physics", "url": "https://doi.org/10.1038/s41567-019-0663-9", "doi": "10.1038/s41567-019-0663-9", "topics": ["Physics"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2019-10", "2021-12", "2022-10", "2023-01", "2025-03"], "last_mentioned": "2025-03-16T17:55:45Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41574-021-00593-z", "slug": "36d6430bcc077a3f", "title": "The health effects of vitamin D supplementation: evidence from human studies", "authors": ["Roger Bouillon", "Despoina Manousaki", "Cliff Rosen", "Katerina Trajanoska", "Fernando Rivadeneira", "J. Brent Richards"], "year": 2021, "venue": "Nature Reviews Endocrinology", "url": "https://doi.org/10.1038/s41574-021-00593-z", "doi": "10.1038/s41574-021-00593-z", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 4, "months": ["2023-09", "2024-02", "2024-04", "2024-06", "2024-08"], "last_mentioned": "2024-08-20T21:02:15Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-021-03544-w", "slug": "0575bae09d466890", "title": "A graph placement methodology for fast chip design", "authors": ["Azalia Mirhoseini", "Anna Goldie", "Mustafa Yazgan", "Joe Wenjie Jiang", "Ebrahim Songhori", "Shen Wang", "Young-Joon Lee", "Eric Johnson", "Omkar Pathak", "Azade Nova", "Jiwoo Pak", "Andy Tong", "Kavya Srinivasa", "William Hang", "Emre Tuncer", "Quoc V. Le", "James Laudon", "Richard Ho", "Roger Carpenter", "Jeff Dean"], "year": 2021, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-021-03544-w", "doi": "10.1038/s41586-021-03544-w", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2022-04", "2022-05", "2022-06", "2023-10", "2024-09"], "last_mentioned": "2024-09-27T22:08:56Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-021-01316-7", "slug": "2336722cad273060", "title": "Initial report of decreased SARS-CoV-2 viral load after inoculation with the BNT162b2 vaccine", "authors": ["Matan Levine-Tiefenbrun", "Idan Yelin", "Rachel Katz", "Esma Herzel", "Ziv Golan", "Licita Schreiber", "Tamar Wolf", "Varda Nadler", "Amir Ben-Tov", "Jacob Kuint", "Sivan Gazit", "Tal Patalon", "Gabriel Chodick", "Roy Kishony"], "year": 2021, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-021-01316-7", "doi": "10.1038/s41591-021-01316-7", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2021-07", "2021-08", "2021-10"], "last_mentioned": "2021-10-24T00:22:43Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-021-01575-4", "slug": "f4685a4f165e384f", "title": "Viral loads of Delta-variant SARS-CoV-2 breakthrough infections after vaccination and booster with BNT162b2", "authors": ["Matan Levine-Tiefenbrun", "Idan Yelin", "Hillel Alapi", "Rachel Katz", "Esma Herzel", "Jacob Kuint", "Gabriel Chodick", "Sivan Gazit", "Tal Patalon", "Roy Kishony"], "year": 2021, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-021-01575-4", "doi": "10.1038/s41591-021-01575-4", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2021-11", "2021-12", "2022-01"], "last_mentioned": "2022-01-18T01:58:51Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-024-03201-5", "slug": "67955611a318a395", "title": "The recombinant shingles vaccine is associated with lower risk of dementia", "authors": ["Maxime Taquet", "Quentin Dercon", "John A. Todd", "Paul J. Harrison"], "year": 2024, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-024-03201-5", "doi": "10.1038/s41591-024-03201-5", "topics": ["Biology & medicine"], "description": "Abstract There is emerging evidence that the live herpes zoster (shingles) vaccine might protect against dementia.", "description_method": "abstract-excerpt", "abstract": "Abstract There is emerging evidence that the live herpes zoster (shingles) vaccine might protect against dementia. However, the existing data are limited and refer only to the live vaccine, which is now discontinued in the United States and many other countries in favor of a recombinant vaccine. Whether the recombinant shingles vaccine protects against dementia remains unknown. Here we used a natural experiment opportunity created by the rapid transition from the use of live to the use of recombinant vaccines to compare the risk of dementia between vaccine types. We show that the recombinant vaccine is associated with a significantly lower risk of dementia in the 6 years post-vaccination. Specifically, receiving the recombinant vaccine is associated with a 17% increase in diagnosis-free time, translating into 164 additional days lived without a diagnosis of dementia in those subsequently affected. The recombinant shingles vaccine was also associated with lower risks of dementia than were two other vaccines commonly used in older people: influenza and tetanus–diphtheria–pertussis vaccines. The effect was robust across multiple secondary analyses, and was present in both men and women but was greater in women. These findings should stimulate studies investigating the mechanisms underpinning the protection and could facilitate the design of a large-scale randomized control trial to confirm the possible additional benefit of the recombinant shingles vaccine.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2025-02", "2025-03", "2025-06", "2025-10", "2026-02"], "last_mentioned": "2026-02-26T16:14:59Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-017-09784-z", "slug": "e9ce1b61850f2628", "title": "Toxic fluoride gas emissions from lithium-ion battery fires", "authors": ["Fredrik Larsson", "Petra Andersson", "Per Blomqvist", "Bengt-Erik Mellander"], "year": 2017, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-017-09784-z", "doi": "10.1038/s41598-017-09784-z", "topics": ["Other research"], "description": "Abstract Lithium-ion battery fires generate intense heat and considerable amounts of gas and smoke.", "description_method": "abstract-excerpt", "abstract": "Abstract Lithium-ion battery fires generate intense heat and considerable amounts of gas and smoke. Although the emission of toxic gases can be a larger threat than the heat, the knowledge of such emissions is limited. This paper presents quantitative measurements of heat release and fluoride gas emissions during battery fires for seven different types of commercial lithium-ion batteries. The results have been validated using two independent measurement techniques and show that large amounts of hydrogen fluoride (HF) may be generated, ranging between 20 and 200 mg/Wh of nominal battery energy capacity. In addition, 15–22 mg/Wh of another potentially toxic gas, phosphoryl fluoride (POF 3 ), was measured in some of the fire tests. Gas emissions when using water mist as extinguishing agent were also investigated. Fluoride gas emission can pose a serious toxic threat and the results are crucial findings for risk assessment and management, especially for large Li-ion battery packs.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2019-06", "2021-07", "2022-09", "2025-02"], "last_mentioned": "2025-02-27T16:35:45Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-018-22939-w", "slug": "73aef5aa02f549a7", "title": "Evidence that the Great Pacific Garbage Patch is rapidly accumulating plastic", "authors": ["L. Lebreton", "B. Slat", "F. Ferrari", "B. Sainte-Rose", "J. Aitken", "R. Marthouse", "S. Hajbane", "S. Cunsolo", "A. Schwarz", "A. Levivier", "K. Noble", "P. Debeljak", "H. Maral", "R. Schoeneich-Argent", "R. Brambini", "J. Reisser"], "year": 2018, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-018-22939-w", "doi": "10.1038/s41598-018-22939-w", "topics": ["Other research"], "description": "Abstract Ocean plastic can persist in sea surface waters, eventually accumulating in remote areas of the world’s oceans.", "description_method": "abstract-excerpt", "abstract": "Abstract Ocean plastic can persist in sea surface waters, eventually accumulating in remote areas of the world’s oceans. Here we characterise and quantify a major ocean plastic accumulation zone formed in subtropical waters between California and Hawaii: The Great Pacific Garbage Patch (GPGP). Our model, calibrated with data from multi-vessel and aircraft surveys, predicted at least 79 (45–129) thousand tonnes of ocean plastic are floating inside an area of 1.6 million km 2 ; a figure four to sixteen times higher than previously reported. We explain this difference through the use of more robust methods to quantify larger debris. Over three-quarters of the GPGP mass was carried by debris larger than 5 cm and at least 46% was comprised of fishing nets. Microplastics accounted for 8% of the total mass but 94% of the estimated 1.8 (1.1–3.6) trillion pieces floating in the area. Plastic collected during our study has specific characteristics such as small surface-to-volume ratio, indicating that only certain types of debris have the capacity to persist and accumulate at the surface of the GPGP. Finally, our results suggest that ocean plastic pollution within the GPGP is increasing exponentially and at a faster rate than in surrounding waters.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2019-01", "2020-01", "2021-09", "2022-05", "2022-07"], "last_mentioned": "2022-07-18T03:40:07Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-020-60661-8", "slug": "eeb57008b77a1a05", "title": "Relating Natural Language Aptitude to Individual Differences in Learning Programming Languages", "authors": ["Chantel S. Prat", "Tara M. Madhyastha", "Malayka J. Mottarella", "Chu-Hsuan Kuo"], "year": 2020, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-020-60661-8", "doi": "10.1038/s41598-020-60661-8", "topics": ["Programming languages", "Statistics & probability", "Psychology"], "description": "AbstractThis experiment employed an individual differences approach to test the hypothesis that learning modern programming languages resembles second “natural” language learning in adulthood.", "description_method": "abstract-excerpt", "abstract": "AbstractThis experiment employed an individual differences approach to test the hypothesis that learning modern programming languages resembles second “natural” language learning in adulthood. Behavioral and neural (resting-state EEG) indices of language aptitude were used along with numeracy and fluid cognitive measures (e.g., fluid reasoning, working memory, inhibitory control) as predictors. Rate of learning, programming accuracy, and post-test declarative knowledge were used as outcome measures in 36 individuals who participated in ten 45-minute Python training sessions. The resulting models explained 50–72% of the variance in learning outcomes, with language aptitude measures explaining significant variance in each outcome even when the other factors competed for variance. Across outcome variables, fluid reasoning and working-memory capacity explained 34% of the variance, followed by language aptitude (17%), resting-state EEG power in beta and low-gamma bands (10%), and numeracy (2%). These results provide a novel framework for understanding programming aptitude, suggesting that the importance of numeracy may be overestimated in modern programming education environments.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2020-07", "2020-11", "2021-09", "2024-02", "2025-05", "2025-06"], "last_mentioned": "2025-06-02T19:30:56Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-020-67211-2", "slug": "919688bc044ed520", "title": "Far-UVC light (222 nm) efficiently and safely inactivates airborne human coronaviruses", "authors": ["Manuela Buonanno", "David Welch", "Igor Shuryak", "David J. Brenner"], "year": 2020, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-020-67211-2", "doi": "10.1038/s41598-020-67211-2", "topics": ["Biology & medicine"], "description": "Abstract A direct approach to limit airborne viral transmissions is to inactivate them within a short time of their production.", "description_method": "abstract-excerpt", "abstract": "Abstract A direct approach to limit airborne viral transmissions is to inactivate them within a short time of their production. Germicidal ultraviolet light, typically at 254 nm, is effective in this context but, used directly, can be a health hazard to skin and eyes. By contrast, far-UVC light (207–222 nm) efficiently kills pathogens potentially without harm to exposed human tissues. We previously demonstrated that 222-nm far-UVC light efficiently kills airborne influenza virus and we extend those studies to explore far-UVC efficacy against airborne human coronaviruses alpha HCoV-229E and beta HCoV-OC43. Low doses of 1.7 and 1.2 mJ/cm 2 inactivated 99.9% of aerosolized coronavirus 229E and OC43, respectively. As all human coronaviruses have similar genomic sizes, far-UVC light would be expected to show similar inactivation efficiency against other human coronaviruses including SARS-CoV-2. Based on the beta-HCoV-OC43 results, continuous far-UVC exposure in occupied public locations at the current regulatory exposure limit (~3 mJ/cm 2 /hour) would result in ~90% viral inactivation in ~8 minutes, 95% in ~11 minutes, 99% in ~16 minutes and 99.9% inactivation in ~25 minutes. Thus while staying within current regulatory dose limits, low-dose-rate far-UVC exposure can potentially safely provide a major reduction in the ambient level of airborne coronaviruses in occupied public locations.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2020-10", "2022-03", "2022-05", "2024-01"], "last_mentioned": "2024-01-18T04:47:59Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-021-83040-3", "slug": "6701caa02ae691b1", "title": "Years of life lost to COVID-19 in 81 countries", "authors": ["Héctor Pifarré i Arolas", "Enrique Acosta", "Guillem López-Casasnovas", "Adeline Lo", "Catia Nicodemo", "Tim Riffe", "Mikko Myrskylä"], "year": 2021, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-021-83040-3", "doi": "10.1038/s41598-021-83040-3", "topics": ["Other research"], "description": "Abstract Understanding the mortality impact of COVID-19 requires not only counting the dead, but analyzing how premature the deaths are.", "description_method": "abstract-excerpt", "abstract": "Abstract Understanding the mortality impact of COVID-19 requires not only counting the dead, but analyzing how premature the deaths are. We calculate years of life lost (YLL) across 81 countries due to COVID-19 attributable deaths, and also conduct an analysis based on estimated excess deaths. We find that over 20.5 million years of life have been lost to COVID-19 globally. As of January 6, 2021, YLL in heavily affected countries are 2–9 times the average seasonal influenza; three quarters of the YLL result from deaths in ages below 75 and almost a third from deaths below 55; and men have lost 45% more life years than women. The results confirm the large mortality impact of COVID-19 among the elderly. They also call for heightened awareness in devising policies that protect vulnerable demographics losing the largest number of life-years.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2021-02", "2021-04", "2021-05", "2021-12"], "last_mentioned": "2021-12-22T20:27:21Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s42256-022-00465-9", "slug": "8d4ad79d152ac567", "title": "Dual use of artificial-intelligence-powered drug discovery", "authors": ["Fabio Urbina", "Filippa Lentzos", "Cédric Invernizzi", "Sean Ekins"], "year": 2022, "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-022-00465-9", "doi": "10.1038/s42256-022-00465-9", "topics": ["Psychology"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2022-05", "2023-03", "2023-04", "2023-06"], "last_mentioned": "2023-06-30T18:53:56Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.aay2400", "slug": "7b63372684654c41", "title": "Superhuman AI for multiplayer poker", "authors": ["Noam Brown", "Tuomas Sandholm"], "year": 2019, "venue": "Science", "url": "https://doi.org/10.1126/science.aay2400", "doi": "10.1126/science.aay2400", "topics": ["Psychology"], "description": "AI now masters six-player poker Computer programs have shown superiority over humans in two-player games such as chess, Go, and heads-up, no-limit Texas hold'em poker.", "description_method": "abstract-excerpt", "abstract": "AI now masters six-player poker Computer programs have shown superiority over humans in two-player games such as chess, Go, and heads-up, no-limit Texas hold'em poker. However, poker games usually include six players—a much trickier challenge for artificial intelligence than the two-player variant. Brown and Sandholm developed a program, dubbed Pluribus, that learned how to play six-player no-limit Texas hold'em by playing against five copies of itself (see the Perspective by Blair and Saffidine). When pitted against five elite professional poker players, or with five copies of Pluribus playing against one professional, the computer performed significantly better than humans over the course of 10,000 hands of poker. Science , this issue p. 885 ; see also p. 864", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2021-09", "2022-01", "2022-06", "2024-02", "2025-10"], "last_mentioned": "2025-10-28T14:55:50Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abe5017", "slug": "3fbb887e4d5172e1", "title": "Daily energy expenditure through the human life course", "authors": ["Herman Pontzer", "Yosuke Yamada", "Hiroyuki Sagayama", "Philip N. Ainslie", "Lene F. Andersen", "Liam J. Anderson", "Lenore Arab", "Issaad Baddou", "Kweku Bedu-Addo", "Ellen E. Blaak", "Stephane Blanc", "Alberto G. Bonomi", "Carlijn V. C. Bouten", "Pascal Bovet", "Maciej S. Buchowski", "Nancy F. Butte", "Stefan G. Camps", "Graeme L. Close", "Jamie A. Cooper", "Richard Cooper", "Sai Krupa Das", "Lara R. Dugas", "Ulf Ekelund", "Sonja Entringer", "Terrence Forrester", "Barry W. Fudge", "Annelies H Goris", "Michael Gurven", "Catherine Hambly", "Asmaa El Hamdouchi", "Marjije B. Hoos", "Sumei Hu", "Noorjehan Joonas", "Annemiek M. Joosen", "Peter Katzmarzyk", "Kitty P. Kempen", "Misaka Kimura", "William E. Kraus", "Robert F. Kushner", "Estelle V. Lambert", "William R. Leonard", "Nader Lessan", "Corby Martin", "Anine C. Medin", "Erwin P. Meijer", "James C. Morehen", "James P. Morton", "Marian L. Neuhouser", "Teresa A. Nicklas", "Robert M. Ojiambo", "Kirsi H. Pietiläinen", "Yannis P. Pitsiladis", "Jacob Plange-Rhule", "Guy Plasqui", "Ross L. Prentice", "Roberto A. Rabinovich", "Susan B. Racette", "David A. Raichlen", "Eric Ravussin", "Rebecca M. Reynolds", "Susan B. Roberts", "Albertine J. Schuit", "Anders M. Sjödin", "Eric Stice", "Samuel S. Urlacher", "Giulio Valenti", "Ludo M. Van Etten", "Edgar A. Van Mil", "Jonathan C. K. Wells", "George Wilson", "Brian M. Wood", "Jack Yanovski", "Tsukasa Yoshida", "Xueying Zhang", "Alexia J. Murphy-Alford", "Cornelia Loechl", "Amy H. Luke", "Jennifer Rood", "Dale A. Schoeller", "Klaas R. Westerterp", "William W. Wong", "John R. Speakman", "IAEA DLW Database Consortium§"], "year": 2021, "venue": "Science", "url": "https://doi.org/10.1126/science.abe5017", "doi": "10.1126/science.abe5017", "topics": ["Other research"], "description": "A lifetime of change Measurements of total and basal energy in a large cohort of subjects at ages spanning from before birth to old age document distinct changes that occur during a human lifetime.", "description_method": "abstract-excerpt", "abstract": "A lifetime of change Measurements of total and basal energy in a large cohort of subjects at ages spanning from before birth to old age document distinct changes that occur during a human lifetime. Pontzer et al . report that energy expenditure (adjusted for weight) in neonates was like that of adults but increased substantially in the first year of life (see the Perspective by Rhoads and Anderson). It then gradually declined until young individuals reached adult characteristics, which were maintained from age 20 to 60 years. Older individuals showed reduced energy expenditure. Tissue metabolism thus appears not to be constant but rather to undergo transitions at critical junctures. —LBR", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2022-04", "2022-07", "2022-09", "2023-10", "2024-04"], "last_mentioned": "2024-04-21T18:59:10Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abj8222", "slug": "1376fed555a2f092", "title": "Longitudinal analysis reveals high prevalence of Epstein-Barr virus associated with multiple sclerosis", "authors": ["Kjetil Bjornevik", "Marianna Cortese", "Brian C. Healy", "Jens Kuhle", "Michael J. Mina", "Yumei Leng", "Stephen J. Elledge", "David W. Niebuhr", "Ann I. Scher", "Kassandra L. Munger", "Alberto Ascherio"], "year": 2022, "venue": "Science", "url": "https://doi.org/10.1126/science.abj8222", "doi": "10.1126/science.abj8222", "topics": ["Statistics & probability", "Biology & medicine"], "description": "Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the central nervous system of unknown etiology.", "description_method": "abstract-excerpt", "abstract": "Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the central nervous system of unknown etiology. We tested the hypothesis that MS is caused by Epstein-Barr virus (EBV) in a cohort comprising more than 10 million young adults on active duty in the US military, 955 of whom were diagnosed with MS during their period of service. Risk of MS increased 32-fold after infection with EBV but was not increased after infection with other viruses, including the similarly transmitted cytomegalovirus. Serum levels of neurofilament light chain, a biomarker of neuroaxonal degeneration, increased only after EBV seroconversion. These findings cannot be explained by any known risk factor for MS and suggest EBV as the leading cause of MS.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2022-01", "2022-08", "2022-10", "2023-02", "2025-02", "2026-03"], "last_mentioned": "2026-03-31T15:45:15Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abk0063", "slug": "190720d893dc816a", "title": "Assessing ExxonMobil’s global warming projections", "authors": ["G. Supran", "S. Rahmstorf", "N. Oreskes"], "year": 2023, "venue": "Science", "url": "https://doi.org/10.1126/science.abk0063", "doi": "10.1126/science.abk0063", "topics": ["Science & research"], "description": "Climate projections by the fossil fuel industry have never been assessed.", "description_method": "abstract-excerpt", "abstract": "Climate projections by the fossil fuel industry have never been assessed. On the basis of company records, we quantitatively evaluated all available global warming projections documented by—and in many cases modeled by—Exxon and ExxonMobil Corp scientists between 1977 and 2003. We find that most of their projections accurately forecast warming that is consistent with subsequent observations. Their projections were also consistent with, and at least as skillful as, those of independent academic and government models. Exxon and ExxonMobil Corp also correctly rejected the prospect of a coming ice age, accurately predicted when human-caused global warming would first be detected, and reasonably estimated the “carbon budget” for holding warming below 2°C. On each of these points, however, the company’s public statements about climate science contradicted its own scientific data.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2023-01", "2026-03", "2026-05", "2026-07"], "last_mentioned": "2026-07-30T14:54:41Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.ade9097", "slug": "23fdfddd500fee9c", "title": "Human-level play in the game of Diplomacy by combining language models with strategic reasoning", "authors": ["Meta Fundamental AI Research Diplomacy Team (FAIR)†", "Anton Bakhtin", "Noam Brown", "Emily Dinan", "Gabriele Farina", "Colin Flaherty", "Daniel Fried", "Andrew Goff", "Jonathan Gray", "Hengyuan Hu", "Athul Paul Jacob", "Mojtaba Komeili", "Karthik Konath", "Minae Kwon", "Adam Lerer", "Mike Lewis", "Alexander H. Miller", "Sasha Mitts", "Adithya Renduchintala", "Stephen Roller", "Dirk Rowe", "Weiyan Shi", "Joe Spisak", "Alexander Wei", "David Wu", "Hugh Zhang", "Markus Zijlstra"], "year": 2022, "venue": "Science", "url": "https://doi.org/10.1126/science.ade9097", "doi": "10.1126/science.ade9097", "topics": ["Language models", "Psychology", "Machine learning"], "description": "Despite much progress in training artificial intelligence (AI) systems to imitate human language, building agents that use language to communicate intentionally with humans in interactive environments remains a major challenge.", "description_method": "abstract-excerpt", "abstract": "Despite much progress in training artificial intelligence (AI) systems to imitate human language, building agents that use language to communicate intentionally with humans in interactive environments remains a major challenge. We introduce Cicero, the first AI agent to achieve human-level performance in Diplomacy , a strategy game involving both cooperation and competition that emphasizes natural language negotiation and tactical coordination between seven players. Cicero integrates a language model with planning and reinforcement learning algorithms by inferring players’ beliefs and intentions from its conversations and generating dialogue in pursuit of its plans. Across 40 games of an anonymous online Diplomacy league, Cicero achieved more than double the average score of the human players and ranked in the top 10% of participants who played more than one game.", "score": 0.0, "mention_count": 6, "user_count": 6, "months": ["2022-11", "2022-12", "2023-11", "2024-03", "2025-10"], "last_mentioned": "2025-10-28T13:36:05Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/2911981", "slug": "a428dfe5affce3bb", "title": "Small-data computing", "authors": ["Hans-J. Boehm"], "year": 2017, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/2911981", "doi": "10.1145/2911981", "topics": ["Other research"], "description": "Rounding errors are usually avoidable, and sometimes we can afford to avoid them.", "description_method": "abstract-excerpt", "abstract": "Rounding errors are usually avoidable, and sometimes we can afford to avoid them.", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2020-03", "2020-06", "2022-04", "2024-01"], "last_mentioned": "2024-01-04T06:34:09Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3243176.3243195", "slug": "c57bfa44eb182628", "title": "Biased reference counting", "authors": ["Jiho Choi", "Thomas Shull", "Josep Torrellas"], "year": 2018, "venue": "Proceedings of the 27th International Conference on Parallel Architectures and Compilation Techniques", "url": "https://doi.org/10.1145/3243176.3243195", "doi": "10.1145/3243176.3243195", "topics": ["Psychology"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 5, "months": ["2023-04", "2023-05", "2023-06", "2024-11", "2025-07"], "last_mentioned": "2025-07-22T20:19:31Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3514221.3526055", "slug": "d461dde49067e9c5", "title": "Cloud-Native Transactions and Analytics in SingleStore", "authors": ["Adam Prout", "Szu-Po Wang", "Joseph Victor", "Zhou Sun", "Yongzhu Li", "Jack Chen", "Evan Bergeron", "Eric Hanson", "Robert Walzer", "Rodrigo Gomes", "Nikita Shamgunov"], "year": 2022, "venue": "Proceedings of the 2022 International Conference on Management of Data", "url": "https://doi.org/10.1145/3514221.3526055", "doi": "10.1145/3514221.3526055", "topics": ["Databases"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 3, "months": ["2022-09", "2022-10", "2022-11", "2023-08", "2026-04"], "last_mentioned": "2026-04-19T13:38:14Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "url:https://papers.nips.cc/paper/2017/hash/3f5ee243547dee91fbd053c1c4a845aa-Abstract.html", "slug": "578d7801c8e7db73", "title": "Attention is All you Need", "authors": ["Vaswani, Ashish", "Shazeer, Noam", "Parmar, Niki", "Uszkoreit, Jakob", "Jones, Llion", "Gomez, Aidan N.", "Kaiser, Łukasz", "Polosukhin, Illia"], "year": 2017, "venue": "Advances in Neural Information Processing Systems", "url": "https://papers.nips.cc/paper/2017/hash/3f5ee243547dee91fbd053c1c4a845aa-Abstract.html", "doi": null, "topics": ["Language models"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 6, "user_count": 4, "months": ["2022-03", "2023-01", "2023-02", "2023-07", "2024-02"], "last_mentioned": "2024-02-15T14:04:39Z", "metadata_source": "publisher-citation-tags", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0907.0455", "slug": "723c50bd8372fe5d", "title": "The Peter Principle Revisited: A Computational Study", "authors": ["Alessandro Pluchino", "Andrea Rapisarda", "Cesare Garofalo"], "year": 2009, "venue": "arXiv", "url": "https://arxiv.org/abs/0907.0455", "doi": "10.1016/j.physa.2009.09.045", "topics": ["Psychology", "Economics"], "description": "In the late sixties the Canadian psychologist Laurence J.", "description_method": "abstract-excerpt", "abstract": "In the late sixties the Canadian psychologist Laurence J. Peter advanced an apparently paradoxical principle, named since then after him, which can be summarized as follows: {\\it 'Every new member in a hierarchical organization climbs the hierarchy until he/she reaches his/her level of maximum incompetence'}. Despite its apparent unreasonableness, such a principle would realistically act in any organization where the mechanism of promotion rewards the best members and where the mechanism at their new level in the hierarchical structure does not depend on the competence they had at the previous level, usually because the tasks of the levels are very different to each other. Here we show, by means of agent based simulations, that if the latter two features actually hold in a given model of an organization with a hierarchical structure, then not only is the Peter principle unavoidable, but also it yields in turn a significant reduction of the global efficiency of the organization. Within a game theory-like approach, we explore different promotion strategies and we find, counterintuitively, that in order to avoid such an effect the best ways for improving the efficiency of a given organization are either to promote each time an agent at random or to promote randomly the best and the worst members in terms of competence.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2009-07", "2010-11", "2011-04", "2017-11", "2022-12"], "last_mentioned": "2022-12-04T01:40:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1102.1889", "slug": "ca6a271ae84b4de3", "title": "Ologs: a categorical framework for knowledge representation", "authors": ["David I. Spivak", "Robert E. Kent"], "year": 2011, "venue": "arXiv", "url": "https://arxiv.org/abs/1102.1889", "doi": "10.1371/journal.pone.0024274", "topics": ["Databases", "Mathematics"], "description": "In this paper we introduce the olog, or ontology log, a category-theoretic model for knowledge representation (KR).", "description_method": "abstract-excerpt", "abstract": "In this paper we introduce the olog, or ontology log, a category-theoretic model for knowledge representation (KR). Grounded in formal mathematics, ologs can be rigorously formulated and cross-compared in ways that other KR models (such as semantic networks) cannot. An olog is similar to a relational database schema; in fact an olog can serve as a data repository if desired. Unlike database schemas, which are generally difficult to create or modify, ologs are designed to be user-friendly enough that authoring or reconfiguring an olog is a matter of course rather than a difficult chore. It is hoped that learning to author ologs is much simpler than learning a database definition language, despite their similarity. We describe ologs carefully and illustrate with many examples. As an application we show that any primitive recursive function can be described by an olog. We also show that ologs can be aligned or connected together into a larger network using functors. The various methods of information flow and institutions can then be used to integrate local and global world-views. We finish by providing several different avenues for future research.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2017-06", "2017-11", "2018-07", "2018-12", "2019-08"], "last_mentioned": "2019-08-25T14:04:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1105.3796", "slug": "b854a76fd56e6c2d", "title": "The Multiverse Interpretation of Quantum Mechanics", "authors": ["Raphael Bousso", "Leonard Susskind"], "year": 2011, "venue": "arXiv", "url": "https://arxiv.org/abs/1105.3796", "doi": "10.1103/PhysRevD.85.045007", "topics": ["Physics", "Statistics & probability", "Mathematics"], "description": "We argue that the many-worlds of quantum mechanics and the many worlds of the multiverse are the same thing, and that the multiverse is necessary to give exact operational meaning to probabilistic predictions from quantum mechanics.", "description_method": "abstract-excerpt", "abstract": "We argue that the many-worlds of quantum mechanics and the many worlds of the multiverse are the same thing, and that the multiverse is necessary to give exact operational meaning to probabilistic predictions from quantum mechanics. Decoherence - the modern version of wave-function collapse - is subjective in that it depends on the choice of a set of unmonitored degrees of freedom, the \"environment\". In fact decoherence is absent in the complete description of any region larger than the future light-cone of a measurement event. However, if one restricts to the causal diamond - the largest region that can be causally probed - then the boundary of the diamond acts as a one-way membrane and thus provides a preferred choice of environment. We argue that the global multiverse is a representation of the many-worlds (all possible decoherent causal diamond histories) in a single geometry. We propose that it must be possible in principle to verify quantum-mechanical predictions exactly. This requires not only the existence of exact observables but two additional postulates: a single observer within the universe can access infinitely many identical experiments; and the outcome of each experiment must be completely definite. In causal diamonds with finite surface area, holographic entropy bounds imply that no exact observables exist, and both postulates fail: experiments cannot be repeated infinitely many times; and decoherence is not completely irreversible, so outcomes are not definite. We argue that our postulates can be satisfied in \"hats\" (supersymmetric multiverse regions with vanishing cosmological constant). We propose a complementarity principle that relates the approximate observables associated with finite causal diamonds to exact observables in the hat.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2011-07", "2017-01", "2018-09", "2018-11"], "last_mentioned": "2018-11-30T03:04:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1112.6209", "slug": "d37ed24b5fc384c2", "title": "Building high-level features using large scale unsupervised learning", "authors": ["Quoc V. Le", "Marc'Aurelio Ranzato", "Rajat Monga", "Matthieu Devin", "Kai Chen", "Greg S. Corrado", "Jeff Dean", "Andrew Y. Ng"], "year": 2011, "venue": "arXiv", "url": "https://arxiv.org/abs/1112.6209", "doi": null, "topics": ["Machine learning", "Networks"], "description": "We consider the problem of building high-level, class-specific feature detectors from only unlabeled data.", "description_method": "abstract-excerpt", "abstract": "We consider the problem of building high-level, class-specific feature detectors from only unlabeled data. For example, is it possible to learn a face detector using only unlabeled images? To answer this, we train a 9-layered locally connected sparse autoencoder with pooling and local contrast normalization on a large dataset of images (the model has 1 billion connections, the dataset has 10 million 200x200 pixel images downloaded from the Internet). We train this network using model parallelism and asynchronous SGD on a cluster with 1,000 machines (16,000 cores) for three days. Contrary to what appears to be a widely-held intuition, our experimental results reveal that it is possible to train a face detector without having to label images as containing a face or not. Control experiments show that this feature detector is robust not only to translation but also to scaling and out-of-plane rotation. We also find that the same network is sensitive to other high-level concepts such as cat faces and human bodies. Starting with these learned features, we trained our network to obtain 15.8% accuracy in recognizing 20,000 object categories from ImageNet, a leap of 70% relative improvement over the previous state-of-the-art.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2012-06", "2013-12", "2015-08", "2023-03"], "last_mentioned": "2023-03-19T04:19:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1210.1847", "slug": "0931521d0bdc791e", "title": "Constraints on the Universe as a Numerical Simulation", "authors": ["Silas R. Beane", "Zohreh Davoudi", "Martin J. Savage"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1210.1847", "doi": "10.1140/epja/i2014-14148-0", "topics": ["Statistics & probability"], "description": "Observable consequences of the hypothesis that the observed universe is a numerical simulation performed on a cubic space-time lattice or grid are explored.", "description_method": "abstract-excerpt", "abstract": "Observable consequences of the hypothesis that the observed universe is a numerical simulation performed on a cubic space-time lattice or grid are explored. The simulation scenario is first motivated by extrapolating current trends in computational resource requirements for lattice QCD into the future. Using the historical development of lattice gauge theory technology as a guide, we assume that our universe is an early numerical simulation with unimproved Wilson fermion discretization and investigate potentially-observable consequences. Among the observables that are considered are the muon g-2 and the current differences between determinations of alpha, but the most stringent bound on the inverse lattice spacing of the universe, b^(-1) >~ 10^(11) GeV, is derived from the high-energy cut off of the cosmic ray spectrum. The numerical simulation scenario could reveal itself in the distributions of the highest energy cosmic rays exhibiting a degree of rotational symmetry breaking that reflects the structure of the underlying lattice.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2012-12", "2014-02", "2014-07", "2015-12", "2021-07"], "last_mentioned": "2021-07-20T09:18:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1309.4838", "slug": "ea478eec91f600c9", "title": "Mass Extinction And The Structure Of The Milky Way", "authors": ["M. D. Filipović", "J. Horner", "E. J. Crawford", "N. F. H. Tothill"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1309.4838", "doi": "10.2298/SAJ130819005F", "topics": ["Statistics & probability"], "description": "We use the most up to date Milky Way model and solar orbit data in order to test the hypothesis that the Sun's galactic spiral arm crossings cause mass extinction events on Earth.", "description_method": "abstract-excerpt", "abstract": "We use the most up to date Milky Way model and solar orbit data in order to test the hypothesis that the Sun's galactic spiral arm crossings cause mass extinction events on Earth. To do this, we created a new model of the Milky Way's spiral arms by combining a large quantity of data from several surveys. We then combined this model with a recently derived solution for the solar orbit to determine the timing of the Sun's historical passages through the Galaxy's spiral arms. Our new model was designed with a symmetrical appearance, with the major alteration being the addition of a spur at the far side of the Galaxy. A correlation was found between the times at which the Sun crosses the spiral arms and six known mass extinction events. Furthermore, we identify five additional historical mass extinction events that might be explained by the motion of the Sun around our Galaxy. These five additional significant drops in marine genera that we find include significant reductions in diversity at 415, 322, 300, 145 and 33 Myr ago. Our simulations indicate that the Sun has spent ~60% of its time passing through our Galaxy's various spiral arms. Also, we briefly discuss and combine previous work on the Galactic Habitable Zone with the new Milky Way model.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2017-01", "2017-05", "2018-03", "2018-10", "2020-05"], "last_mentioned": "2020-05-28T13:27:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.7152", "slug": "345c953d76962c88", "title": "twister - a P2P microblogging platform", "authors": ["Miguel Freitas"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.7152", "doi": null, "topics": ["Security & cryptography", "Distributed systems", "Economics"], "description": "This paper proposes a new microblogging architecture based on peer-to-peer networks overlays.", "description_method": "abstract-excerpt", "abstract": "This paper proposes a new microblogging architecture based on peer-to-peer networks overlays. The proposed platform is comprised of three mostly independent overlay networks. The first provides distributed user registration and authentication and is based on the Bitcoin protocol. The second one is a Distributed Hash Table (DHT) overlay network providing key/value storage for user resources and tracker location for the third network. The last network is a collection of possibly disjoint \"swarms\" of followers, based on the Bittorrent protocol, which can be used for efficient near-instant notification delivery to many users. By leveraging from existing and proven technologies, twister provides a new microblogging platform offering security, scalability and privacy features. A mechanism provides incentive for entities that contribute processing time to run the user registration network, rewarding such entities with the privilege of sending a single unsolicited (\"promoted\") message to the entire network. The number of unsolicited messages per day is defined in order to not upset users.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2013-12", "2015-01", "2015-03", "2018-06"], "last_mentioned": "2018-06-03T20:11:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1401.4356", "slug": "e122f1348fafdeac", "title": "Why bouncing droplets are a pretty good model of quantum mechanics", "authors": ["Robert Brady", "Ross Anderson"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1401.4356", "doi": null, "topics": ["Physics", "Psychology"], "description": "In 2005, Couder, Protiere, Fort and Badouad showed that oil droplets bouncing on a vibrating tray of oil can display nonlocal interactions reminiscent of the particle-wave associations in quantum mechanics; in particular they can move, attract, repel and orbit each other.", "description_method": "abstract-excerpt", "abstract": "In 2005, Couder, Protiere, Fort and Badouad showed that oil droplets bouncing on a vibrating tray of oil can display nonlocal interactions reminiscent of the particle-wave associations in quantum mechanics; in particular they can move, attract, repel and orbit each other. Subsequent experimental work by Couder, Fort, Protiere, Eddi, Sultan, Moukhtar, Rossi, Molacek, Bush and Sbitnev has established that bouncing drops exhibit single-slit and double-slit diffraction, tunnelling, quantised energy levels, Anderson localisation and the creation/annihilation of droplet/bubble pairs. In this paper we explain why. We show first that the surface waves guiding the droplets are Lorentz covariant with the characteristic speed c of the surface waves; second, that pairs of bouncing droplets experience an inverse-square force of attraction or repulsion according to their relative phase, and an analogue of the magnetic force; third, that bouncing droplets are governed by an analogue of Schroedinger's equation where Planck's constant is replaced by an appropriate constant of the motion; and fourth, that orbiting droplet pairs exhibit spin-half symmetry and align antisymmetrically as in the Pauli exclusion principle. Our analysis explains the similarities between bouncing-droplet experiments and the behaviour of quantum-mechanical particles. It also enables us to highlight some differences, and to predict some surprising phenomena that can be tested in feasible experiments.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2014-06", "2014-07", "2015-05", "2016-05", "2017-08"], "last_mentioned": "2017-08-30T15:30:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1412.5567", "slug": "125f3b5e459dbaef", "title": "Deep Speech: Scaling up end-to-end speech recognition", "authors": ["Awni Hannun", "Carl Case", "Jared Casper", "Bryan Catanzaro", "Greg Diamos", "Erich Elsen", "Ryan Prenger", "Sanjeev Satheesh", "Shubho Sengupta", "Adam Coates", "Andrew Y. Ng"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1412.5567", "doi": null, "topics": ["Psychology", "Machine learning", "Economics"], "description": "We present a state-of-the-art speech recognition system developed using end-to-end deep learning.", "description_method": "abstract-excerpt", "abstract": "We present a state-of-the-art speech recognition system developed using end-to-end deep learning. Our architecture is significantly simpler than traditional speech systems, which rely on laboriously engineered processing pipelines; these traditional systems also tend to perform poorly when used in noisy environments. In contrast, our system does not need hand-designed components to model background noise, reverberation, or speaker variation, but instead directly learns a function that is robust to such effects. We do not need a phoneme dictionary, nor even the concept of a \"phoneme.\" Key to our approach is a well-optimized RNN training system that uses multiple GPUs, as well as a set of novel data synthesis techniques that allow us to efficiently obtain a large amount of varied data for training. Our system, called Deep Speech, outperforms previously published results on the widely studied Switchboard Hub5'00, achieving 16.0% error on the full test set. Deep Speech also handles challenging noisy environments better than widely used, state-of-the-art commercial speech systems.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2015-06", "2015-10", "2017-02", "2017-12", "2021-08"], "last_mentioned": "2021-08-05T14:13:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1501.00092", "slug": "d7d22edafff4cebf", "title": "Image Super-Resolution Using Deep Convolutional Networks", "authors": ["Chao Dong", "Chen Change Loy", "Kaiming He", "Xiaoou Tang"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1501.00092", "doi": null, "topics": ["Computer vision", "Machine learning"], "description": "We propose a deep learning method for single image super-resolution (SR).", "description_method": "abstract-excerpt", "abstract": "We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low/high-resolution images. The mapping is represented as a deep convolutional neural network (CNN) that takes the low-resolution image as the input and outputs the high-resolution one. We further show that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network. But unlike traditional methods that handle each component separately, our method jointly optimizes all layers. Our deep CNN has a lightweight structure, yet demonstrates state-of-the-art restoration quality, and achieves fast speed for practical on-line usage. We explore different network structures and parameter settings to achieve trade-offs between performance and speed. Moreover, we extend our network to cope with three color channels simultaneously, and show better overall reconstruction quality.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2015-08", "2016-11", "2017-01", "2017-02"], "last_mentioned": "2017-02-01T03:30:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.05698", "slug": "b430d4cbb86f7f47", "title": "Towards AI-Complete Question Answering: A Set of Prerequisite Toy Tasks", "authors": ["Jason Weston", "Antoine Bordes", "Sumit Chopra", "Alexander M. Rush", "Bart van Merriënboer", "Armand Joulin", "Tomas Mikolov"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.05698", "doi": null, "topics": ["Machine learning"], "description": "One long-term goal of machine learning research is to produce methods that are applicable to reasoning and natural language, in particular building an intelligent dialogue agent.", "description_method": "abstract-excerpt", "abstract": "One long-term goal of machine learning research is to produce methods that are applicable to reasoning and natural language, in particular building an intelligent dialogue agent. To measure progress towards that goal, we argue for the usefulness of a set of proxy tasks that evaluate reading comprehension via question answering. Our tasks measure understanding in several ways: whether a system is able to answer questions via chaining facts, simple induction, deduction and many more. The tasks are designed to be prerequisites for any system that aims to be capable of conversing with a human. We believe many existing learning systems can currently not solve them, and hence our aim is to classify these tasks into skill sets, so that researchers can identify (and then rectify) the failings of their systems. We also extend and improve the recently introduced Memory Networks model, and show it is able to solve some, but not all, of the tasks.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2015-06", "2015-08", "2015-09", "2016-06", "2025-10"], "last_mentioned": "2025-10-09T09:19:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1503.01007", "slug": "0a2092e6f9ec5d5a", "title": "Inferring Algorithmic Patterns with Stack-Augmented Recurrent Nets", "authors": ["Armand Joulin", "Tomas Mikolov"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1503.01007", "doi": null, "topics": ["Algorithms", "Machine learning", "Psychology"], "description": "Despite the recent achievements in machine learning, we are still very far from achieving real artificial intelligence.", "description_method": "abstract-excerpt", "abstract": "Despite the recent achievements in machine learning, we are still very far from achieving real artificial intelligence. In this paper, we discuss the limitations of standard deep learning approaches and show that some of these limitations can be overcome by learning how to grow the complexity of a model in a structured way. Specifically, we study the simplest sequence prediction problems that are beyond the scope of what is learnable with standard recurrent networks, algorithmically generated sequences which can only be learned by models which have the capacity to count and to memorize sequences. We show that some basic algorithms can be learned from sequential data using a recurrent network associated with a trainable memory.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2015-05", "2015-09", "2016-01", "2023-01"], "last_mentioned": "2023-01-23T18:21:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1603.03720", "slug": "b0100899e27d5198", "title": "Unexpected biases in the distribution of consecutive primes", "authors": ["Robert J. Lemke Oliver", "Kannan Soundararajan"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1603.03720", "doi": "10.1073/pnas.1605366113", "topics": ["Psychology", "Distributed systems"], "description": "While the sequence of primes is very well distributed in the reduced residue classes (mod $q$), the distribution of pairs of consecutive primes among the permissible $φ(q)^2$ pairs of reduced residue classes (mod $q$) is surprisingly erratic.", "description_method": "abstract-excerpt", "abstract": "While the sequence of primes is very well distributed in the reduced residue classes (mod $q$), the distribution of pairs of consecutive primes among the permissible $φ(q)^2$ pairs of reduced residue classes (mod $q$) is surprisingly erratic. This paper proposes a conjectural explanation for this phenomenon, based on the Hardy-Littlewood conjectures. The conjectures are then compared to numerical data, and the observed fit is very good.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2016-03", "2016-06", "2020-02"], "last_mentioned": "2020-02-16T16:50:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1606.01868", "slug": "9ce61bea57f6220f", "title": "Unifying Count-Based Exploration and Intrinsic Motivation", "authors": ["Marc G. Bellemare", "Sriram Srinivasan", "Georg Ostrovski", "Tom Schaul", "David Saxton", "Remi Munos"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1606.01868", "doi": null, "topics": ["Algorithms", "Machine learning"], "description": "We consider an agent's uncertainty about its environment and the problem of generalizing this uncertainty across observations.", "description_method": "abstract-excerpt", "abstract": "We consider an agent's uncertainty about its environment and the problem of generalizing this uncertainty across observations. Specifically, we focus on the problem of exploration in non-tabular reinforcement learning. Drawing inspiration from the intrinsic motivation literature, we use density models to measure uncertainty, and propose a novel algorithm for deriving a pseudo-count from an arbitrary density model. This technique enables us to generalize count-based exploration algorithms to the non-tabular case. We apply our ideas to Atari 2600 games, providing sensible pseudo-counts from raw pixels. We transform these pseudo-counts into intrinsic rewards and obtain significantly improved exploration in a number of hard games, including the infamously difficult Montezuma's Revenge.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2016-06", "2017-09", "2018-07"], "last_mentioned": "2018-07-04T19:27:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1608.06993", "slug": "aa1dd159db6bb160", "title": "Densely Connected Convolutional Networks", "authors": ["Gao Huang", "Zhuang Liu", "Laurens van der Maaten", "Kilian Q. Weinberger"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1608.06993", "doi": null, "topics": ["Computer vision", "Psychology", "Machine learning"], "description": "Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those close to the output.", "description_method": "abstract-excerpt", "abstract": "Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those close to the output. In this paper, we embrace this observation and introduce the Dense Convolutional Network (DenseNet), which connects each layer to every other layer in a feed-forward fashion. Whereas traditional convolutional networks with L layers have L connections - one between each layer and its subsequent layer - our network has L(L+1)/2 direct connections. For each layer, the feature-maps of all preceding layers are used as inputs, and its own feature-maps are used as inputs into all subsequent layers. DenseNets have several compelling advantages: they alleviate the vanishing-gradient problem, strengthen feature propagation, encourage feature reuse, and substantially reduce the number of parameters. We evaluate our proposed architecture on four highly competitive object recognition benchmark tasks (CIFAR-10, CIFAR-100, SVHN, and ImageNet). DenseNets obtain significant improvements over the state-of-the-art on most of them, whilst requiring less computation to achieve high performance. Code and pre-trained models are available at https://github.com/liuzhuang13/DenseNet .", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2016-11", "2017-07", "2017-11", "2021-12", "2025-09"], "last_mentioned": "2025-09-18T23:08:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.09359", "slug": "fa84704c290bf74a", "title": "Don't Skype & Type! Acoustic Eavesdropping in Voice-Over-IP", "authors": ["Alberto Compagno", "Mauro Conti", "Daniele Lain", "Gene Tsudik"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.09359", "doi": null, "topics": ["Security & cryptography", "Networks"], "description": "Acoustic emanations of computer keyboards represent a serious privacy issue.", "description_method": "abstract-excerpt", "abstract": "Acoustic emanations of computer keyboards represent a serious privacy issue. As demonstrated in prior work, physical properties of keystroke sounds might reveal what a user is typing. However, previous attacks assumed relatively strong adversary models that are not very practical in many real-world settings. Such strong models assume: (i) adversary's physical proximity to the victim, (ii) precise profiling of the victim's typing style and keyboard, and/or (iii) significant amount of victim's typed information (and its corresponding sounds) available to the adversary. This paper presents and explores a new keyboard acoustic eavesdropping attack that involves Voice-over-IP (VoIP), called Skype & Type (S&T), while avoiding prior strong adversary assumptions. This work is motivated by the simple observation that people often engage in secondary activities (including typing) while participating in VoIP calls. As expected, VoIP software acquires and faithfully transmits all sounds, including emanations of pressed keystrokes, which can include passwords and other sensitive information. We show that one very popular VoIP software (Skype) conveys enough audio information to reconstruct the victim's input -- keystrokes typed on the remote keyboard. Our results demonstrate that, given some knowledge on the victim's typing style and keyboard model, the attacker attains top-5 accuracy of 91.7% in guessing a random key pressed by the victim. Furthermore, we demonstrate that S&T is robust to various VoIP issues (e.g., Internet bandwidth fluctuations and presence of voice over keystrokes), thus confirming feasibility of this attack. Finally, it applies to other popular VoIP software, such as Google Hangouts.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2018-09", "2019-08", "2020-04", "2022-08", "2023-12"], "last_mentioned": "2023-12-09T23:59:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1611.05763", "slug": "740664389e5fcc22", "title": "Learning to reinforcement learn", "authors": ["Jane X Wang", "Zeb Kurth-Nelson", "Dhruva Tirumala", "Hubert Soyer", "Joel Z Leibo", "Remi Munos", "Charles Blundell", "Dharshan Kumaran", "Matt Botvinick"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1611.05763", "doi": null, "topics": ["Machine learning", "Algorithms", "Neuroscience"], "description": "In recent years deep reinforcement learning (RL) systems have attained superhuman performance in a number of challenging task domains.", "description_method": "abstract-excerpt", "abstract": "In recent years deep reinforcement learning (RL) systems have attained superhuman performance in a number of challenging task domains. However, a major limitation of such applications is their demand for massive amounts of training data. A critical present objective is thus to develop deep RL methods that can adapt rapidly to new tasks. In the present work we introduce a novel approach to this challenge, which we refer to as deep meta-reinforcement learning. Previous work has shown that recurrent networks can support meta-learning in a fully supervised context. We extend this approach to the RL setting. What emerges is a system that is trained using one RL algorithm, but whose recurrent dynamics implement a second, quite separate RL procedure. This second, learned RL algorithm can differ from the original one in arbitrary ways. Importantly, because it is learned, it is configured to exploit structure in the training domain. We unpack these points in a series of seven proof-of-concept experiments, each of which examines a key aspect of deep meta-RL. We consider prospects for extending and scaling up the approach, and also point out some potentially important implications for neuroscience.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2016-12", "2017-02", "2018-06", "2019-03"], "last_mentioned": "2019-03-28T21:24:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1611.07612", "slug": "9ef4f48e02b1482a", "title": "Faster Population Counts Using AVX2 Instructions", "authors": ["Wojciech Muła", "Nathan Kurz", "Daniel Lemire"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1611.07612", "doi": "10.1093/comjnl/bxx046", "topics": ["Computer systems", "Security & cryptography", "Programming languages"], "description": "Counting the number of ones in a binary stream is a common operation in database, information-retrieval, cryptographic and machine-learning applications.", "description_method": "abstract-excerpt", "abstract": "Counting the number of ones in a binary stream is a common operation in database, information-retrieval, cryptographic and machine-learning applications. Most processors have dedicated instructions to count the number of ones in a word (e.g., popcnt on x64 processors). Maybe surprisingly, we show that a vectorized approach using SIMD instructions can be twice as fast as using the dedicated instructions on recent Intel processors. The benefits can be even greater for applications such as similarity measures (e.g., the Jaccard index) that require additional Boolean operations. Our approach has been adopted by LLVM: it is used by its popular C compiler (clang).", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2016-12", "2018-05", "2019-06", "2019-09", "2021-06"], "last_mentioned": "2021-06-11T18:27:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1701.07875", "slug": "d41c5e4c5f68eb82", "title": "Wasserstein GAN", "authors": ["Martin Arjovsky", "Soumith Chintala", "Léon Bottou"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1701.07875", "doi": null, "topics": ["Algorithms"], "description": "We introduce a new algorithm named WGAN, an alternative to traditional GAN training.", "description_method": "abstract-excerpt", "abstract": "We introduce a new algorithm named WGAN, an alternative to traditional GAN training. In this new model, we show that we can improve the stability of learning, get rid of problems like mode collapse, and provide meaningful learning curves useful for debugging and hyperparameter searches. Furthermore, we show that the corresponding optimization problem is sound, and provide extensive theoretical work highlighting the deep connections to other distances between distributions.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2017-05", "2017-08", "2018-04", "2018-12", "2019-04"], "last_mentioned": "2019-04-10T02:58:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1703.00064", "slug": "d6e198545d593b5c", "title": "Ending the Anomaly: Achieving Low Latency and Airtime Fairness in WiFi", "authors": ["Toke Høiland-Jørgensen", "Michał Kazior", "Dave Täht", "Per Hurtig", "Anna Brunstrom"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1703.00064", "doi": null, "topics": ["Networks", "Computer systems"], "description": "With more devices connected, delays and jitter at the WiFi hop become more prevalent, and correct functioning during network congestion becomes more important.", "description_method": "abstract-excerpt", "abstract": "With more devices connected, delays and jitter at the WiFi hop become more prevalent, and correct functioning during network congestion becomes more important. However, two important performance issues prevent modern WiFi from reaching its potential: Increased latency under load caused by excessive queueing (i.e. bufferbloat) and the 802.11 performance anomaly. To remedy these issues, we present a novel two-part solution: We design a new queueing scheme that eliminates bufferbloat in the wireless setting. Leveraging this queueing scheme, we then design an airtime fairness scheduler that operates at the access point and doesn't require any changes to clients. We evaluate our solution using both a theoretical model and experiments in a testbed environment, formulating a suitable analytical model in the process. We show that our solution achieves an order of magnitude reduction in latency under load, large improvements in multi-station throughput, and nearly perfect airtime fairness for both TCP and downstream UDP traffic. Further experiments with application traffic confirm that the solution provides significant performance gains for real-world traffic.We develop a production quality implementation of our solution in the Linux kernel, the platform powering most access points outside of the managed enterprise setting. The implementation has been accepted into the mainline kernel distribution, making it available for deployment on billions of devices running Linux today.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2017-03", "2018-07", "2018-08", "2022-01"], "last_mentioned": "2022-01-31T19:28:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1703.10987", "slug": "bf688e337eb3c413", "title": "On the Impossibility of Supersized Machines", "authors": ["Ben Garfinkel", "Miles Brundage", "Daniel Filan", "Carrick Flynn", "Jelena Luketina", "Michael Page", "Anders Sandberg", "Andrew Snyder-Beattie", "Max Tegmark"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1703.10987", "doi": null, "topics": ["Physics"], "description": "In recent years, a number of prominent computer scientists, along with academics in fields such as philosophy and physics, have lent credence to the notion that machines may one day become as large as humans.", "description_method": "abstract-excerpt", "abstract": "In recent years, a number of prominent computer scientists, along with academics in fields such as philosophy and physics, have lent credence to the notion that machines may one day become as large as humans. Many have further argued that machines could even come to exceed human size by a significant margin. However, there are at least seven distinct arguments that preclude this outcome. We show that it is not only implausible that machines will ever exceed human size, but in fact impossible.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2017-04", "2018-11", "2021-08", "2022-08", "2024-11"], "last_mentioned": "2024-11-18T02:24:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1704.01444", "slug": "4e9989a69908601b", "title": "Learning to Generate Reviews and Discovering Sentiment", "authors": ["Alec Radford", "Rafal Jozefowicz", "Ilya Sutskever"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1704.01444", "doi": null, "topics": ["Language models"], "description": "We explore the properties of byte-level recurrent language models.", "description_method": "abstract-excerpt", "abstract": "We explore the properties of byte-level recurrent language models. When given sufficient amounts of capacity, training data, and compute time, the representations learned by these models include disentangled features corresponding to high-level concepts. Specifically, we find a single unit which performs sentiment analysis. These representations, learned in an unsupervised manner, achieve state of the art on the binary subset of the Stanford Sentiment Treebank. They are also very data efficient. When using only a handful of labeled examples, our approach matches the performance of strong baselines trained on full datasets. We also demonstrate the sentiment unit has a direct influence on the generative process of the model. Simply fixing its value to be positive or negative generates samples with the corresponding positive or negative sentiment.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2017-04", "2017-05", "2023-09", "2024-02", "2024-06"], "last_mentioned": "2024-06-02T04:51:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1706.03741", "slug": "73d0f49d618fabfa", "title": "Deep reinforcement learning from human preferences", "authors": ["Paul Christiano", "Jan Leike", "Tom B. Brown", "Miljan Martic", "Shane Legg", "Dario Amodei"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1706.03741", "doi": null, "topics": ["Machine learning", "Robotics", "Psychology"], "description": "For sophisticated reinforcement learning (RL) systems to interact usefully with real-world environments, we need to communicate complex goals to these systems.", "description_method": "abstract-excerpt", "abstract": "For sophisticated reinforcement learning (RL) systems to interact usefully with real-world environments, we need to communicate complex goals to these systems. In this work, we explore goals defined in terms of (non-expert) human preferences between pairs of trajectory segments. We show that this approach can effectively solve complex RL tasks without access to the reward function, including Atari games and simulated robot locomotion, while providing feedback on less than one percent of our agent's interactions with the environment. This reduces the cost of human oversight far enough that it can be practically applied to state-of-the-art RL systems. To demonstrate the flexibility of our approach, we show that we can successfully train complex novel behaviors with about an hour of human time. These behaviors and environments are considerably more complex than any that have been previously learned from human feedback.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2018-07", "2019-03", "2019-10", "2023-03", "2025-08"], "last_mentioned": "2025-08-11T01:11:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1706.05137", "slug": "92cd221f5bf22115", "title": "One Model To Learn Them All", "authors": ["Lukasz Kaiser", "Aidan N. Gomez", "Noam Shazeer", "Ashish Vaswani", "Niki Parmar", "Llion Jones", "Jakob Uszkoreit"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1706.05137", "doi": null, "topics": ["Psychology", "Computer vision", "Machine learning"], "description": "Deep learning yields great results across many fields, from speech recognition, image classification, to translation.", "description_method": "abstract-excerpt", "abstract": "Deep learning yields great results across many fields, from speech recognition, image classification, to translation. But for each problem, getting a deep model to work well involves research into the architecture and a long period of tuning. We present a single model that yields good results on a number of problems spanning multiple domains. In particular, this single model is trained concurrently on ImageNet, multiple translation tasks, image captioning (COCO dataset), a speech recognition corpus, and an English parsing task. Our model architecture incorporates building blocks from multiple domains. It contains convolutional layers, an attention mechanism, and sparsely-gated layers. Each of these computational blocks is crucial for a subset of the tasks we train on. Interestingly, even if a block is not crucial for a task, we observe that adding it never hurts performance and in most cases improves it on all tasks. We also show that tasks with less data benefit largely from joint training with other tasks, while performance on large tasks degrades only slightly if at all.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2017-06", "2017-07", "2017-11", "2018-02", "2019-03"], "last_mentioned": "2019-03-27T19:16:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1707.07012", "slug": "274200f5345e6b43", "title": "Learning Transferable Architectures for Scalable Image Recognition", "authors": ["Barret Zoph", "Vijay Vasudevan", "Jonathon Shlens", "Quoc V. Le"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1707.07012", "doi": null, "topics": ["Computer vision", "Psychology", "Machine learning"], "description": "Developing neural network image classification models often requires significant architecture engineering.", "description_method": "abstract-excerpt", "abstract": "Developing neural network image classification models often requires significant architecture engineering. In this paper, we study a method to learn the model architectures directly on the dataset of interest. As this approach is expensive when the dataset is large, we propose to search for an architectural building block on a small dataset and then transfer the block to a larger dataset. The key contribution of this work is the design of a new search space (the \"NASNet search space\") which enables transferability. In our experiments, we search for the best convolutional layer (or \"cell\") on the CIFAR-10 dataset and then apply this cell to the ImageNet dataset by stacking together more copies of this cell, each with their own parameters to design a convolutional architecture, named \"NASNet architecture\". We also introduce a new regularization technique called ScheduledDropPath that significantly improves generalization in the NASNet models. On CIFAR-10 itself, NASNet achieves 2.4% error rate, which is state-of-the-art. On ImageNet, NASNet achieves, among the published works, state-of-the-art accuracy of 82.7% top-1 and 96.2% top-5 on ImageNet. Our model is 1.2% better in top-1 accuracy than the best human-invented architectures while having 9 billion fewer FLOPS - a reduction of 28% in computational demand from the previous state-of-the-art model. When evaluated at different levels of computational cost, accuracies of NASNets exceed those of the state-of-the-art human-designed models. For instance, a small version of NASNet also achieves 74% top-1 accuracy, which is 3.1% better than equivalently-sized, state-of-the-art models for mobile platforms. Finally, the learned features by NASNet used with the Faster-RCNN framework surpass state-of-the-art by 4.0% achieving 43.1% mAP on the COCO dataset.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2017-08", "2017-09", "2017-11", "2018-07", "2018-10"], "last_mentioned": "2018-10-30T19:26:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1708.04748", "slug": "782b5ef7e1798958", "title": "When the cookie meets the blockchain: Privacy risks of web payments via cryptocurrencies", "authors": ["Steven Goldfeder", "Harry Kalodner", "Dillon Reisman", "Arvind Narayanan"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1708.04748", "doi": null, "topics": ["Security & cryptography", "Databases"], "description": "We show how third-party web trackers can deanonymize users of cryptocurrencies.", "description_method": "abstract-excerpt", "abstract": "We show how third-party web trackers can deanonymize users of cryptocurrencies. We present two distinct but complementary attacks. On most shopping websites, third party trackers receive information about user purchases for purposes of advertising and analytics. We show that, if the user pays using a cryptocurrency, trackers typically possess enough information about the purchase to uniquely identify the transaction on the blockchain, link it to the user's cookie, and further to the user's real identity. Our second attack shows that if the tracker is able to link two purchases of the same user to the blockchain in this manner, it can identify the user's entire cluster of addresses and transactions on the blockchain, even if the user employs blockchain anonymity techniques such as CoinJoin. The attacks are passive and hence can be retroactively applied to past purchases. We discuss several mitigations, but none are perfect.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2017-08", "2017-09", "2018-03", "2022-06"], "last_mentioned": "2022-06-12T15:23:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1708.09537", "slug": "790f61124477a92f", "title": "DolphinAtack: Inaudible Voice Commands", "authors": ["Guoming Zhang", "Chen Yan", "Xiaoyu Ji", "Taimin Zhang", "Tianchen Zhang", "Wenyuan Xu"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1708.09537", "doi": "10.1145/3133956.3134052", "topics": ["Security & cryptography", "Psychology", "Human–computer interaction"], "description": "Speech recognition (SR) systems such as Siri or Google Now have become an increasingly popular human-computer interaction method, and have turned various systems into voice controllable systems(VCS).", "description_method": "abstract-excerpt", "abstract": "Speech recognition (SR) systems such as Siri or Google Now have become an increasingly popular human-computer interaction method, and have turned various systems into voice controllable systems(VCS). Prior work on attacking VCS shows that the hidden voice commands that are incomprehensible to people can control the systems. Hidden voice commands, though hidden, are nonetheless audible. In this work, we design a completely inaudible attack, DolphinAttack, that modulates voice commands on ultrasonic carriers (e.g., f > 20 kHz) to achieve inaudibility. By leveraging the nonlinearity of the microphone circuits, the modulated low frequency audio commands can be successfully demodulated, recovered, and more importantly interpreted by the speech recognition systems. We validate DolphinAttack on popular speech recognition systems, including Siri, Google Now, Samsung S Voice, Huawei HiVoice, Cortana and Alexa. By injecting a sequence of inaudible voice commands, we show a few proof-of-concept attacks, which include activating Siri to initiate a FaceTime call on iPhone, activating Google Now to switch the phone to the airplane mode, and even manipulating the navigation system in an Audi automobile. We propose hardware and software defense solutions. We validate that it is feasible to detect DolphinAttack by classifying the audios using supported vector machine (SVM), and suggest to re-design voice controllable systems to be resilient to inaudible voice command attacks.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2017-09", "2018-05", "2019-01", "2020-02"], "last_mentioned": "2020-02-16T12:42:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1710.08781", "slug": "77ab91c6401682c9", "title": "A partisan districting protocol with provably nonpartisan outcomes", "authors": ["Wesley Pegden", "Ariel D. Procaccia", "Dingli Yu"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1710.08781", "doi": null, "topics": ["Other research"], "description": "We design and analyze a protocol for dividing a state into districts, where parties take turns proposing a division, and freezing a district from the other party's proposed division.", "description_method": "abstract-excerpt", "abstract": "We design and analyze a protocol for dividing a state into districts, where parties take turns proposing a division, and freezing a district from the other party's proposed division. We show that our protocol has predictable and provable guarantees for both the number of districts in which each party has a majority of supporters, and the extent to which either party has the power to pack a specific population into a single district.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2021-01", "2022-11", "2023-06", "2024-02", "2026-06"], "last_mentioned": "2026-06-18T18:20:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1710.08864", "slug": "a6d637332591ce7f", "title": "One pixel attack for fooling deep neural networks", "authors": ["Jiawei Su", "Danilo Vasconcellos Vargas", "Sakurai Kouichi"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1710.08864", "doi": "10.1109/TEVC.2019.2890858", "topics": ["Security & cryptography", "Machine learning"], "description": "Recent research has revealed that the output of Deep Neural Networks (DNN) can be easily altered by adding relatively small perturbations to the input vector.", "description_method": "abstract-excerpt", "abstract": "Recent research has revealed that the output of Deep Neural Networks (DNN) can be easily altered by adding relatively small perturbations to the input vector. In this paper, we analyze an attack in an extremely limited scenario where only one pixel can be modified. For that we propose a novel method for generating one-pixel adversarial perturbations based on differential evolution (DE). It requires less adversarial information (a black-box attack) and can fool more types of networks due to the inherent features of DE. The results show that 67.97% of the natural images in Kaggle CIFAR-10 test dataset and 16.04% of the ImageNet (ILSVRC 2012) test images can be perturbed to at least one target class by modifying just one pixel with 74.03% and 22.91% confidence on average. We also show the same vulnerability on the original CIFAR-10 dataset. Thus, the proposed attack explores a different take on adversarial machine learning in an extreme limited scenario, showing that current DNNs are also vulnerable to such low dimension attacks. Besides, we also illustrate an important application of DE (or broadly speaking, evolutionary computation) in the domain of adversarial machine learning: creating tools that can effectively generate low-cost adversarial attacks against neural networks for evaluating robustness.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2017-10", "2017-11", "2018-10"], "last_mentioned": "2018-10-25T17:25:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1710.09756", "slug": "2bccf00b6c207799", "title": "Linear Haskell: practical linearity in a higher-order polymorphic language", "authors": ["Jean-Philippe Bernardy", "Mathieu Boespflug", "Ryan R. Newton", "Simon Peyton Jones", "Arnaud Spiwack"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1710.09756", "doi": "10.1145/3158093", "topics": ["Programming languages"], "description": "Linear type systems have a long and storied history, but not a clear path forward to integrate with existing languages such as OCaml or Haskell.", "description_method": "abstract-excerpt", "abstract": "Linear type systems have a long and storied history, but not a clear path forward to integrate with existing languages such as OCaml or Haskell. In this paper, we study a linear type system designed with two crucial properties in mind: backwards-compatibility and code reuse across linear and non-linear users of a library. Only then can the benefits of linear types permeate conventional functional programming. Rather than bifurcate types into linear and non-linear counterparts, we instead attach linearity to function arrows. Linear functions can receive inputs from linearly-bound values, but can also operate over unrestricted, regular values. To demonstrate the efficacy of our linear type system - both how easy it can be integrated in an existing language implementation and how streamlined it makes it to write programs with linear types - we implemented our type system in GHC, the leading Haskell compiler, and demonstrate two kinds of applications of linear types: mutable data with pure interfaces; and enforcing protocols in I/O-performing functions.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2018-10", "2020-09", "2020-12"], "last_mentioned": "2020-12-31T11:30:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1712.09665", "slug": "36a7f0acf5ffc767", "title": "Adversarial Patch", "authors": ["Tom B. Brown", "Dandelion Mané", "Aurko Roy", "Martín Abadi", "Justin Gilmer"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1712.09665", "doi": null, "topics": ["Machine learning", "Security & cryptography"], "description": "We present a method to create universal, robust, targeted adversarial image patches in the real world.", "description_method": "abstract-excerpt", "abstract": "We present a method to create universal, robust, targeted adversarial image patches in the real world. The patches are universal because they can be used to attack any scene, robust because they work under a wide variety of transformations, and targeted because they can cause a classifier to output any target class. These adversarial patches can be printed, added to any scene, photographed, and presented to image classifiers; even when the patches are small, they cause the classifiers to ignore the other items in the scene and report a chosen target class. To reproduce the results from the paper, our code is available at https://github.com/tensorflow/cleverhans/tree/master/examples/adversarial_patch", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2018-05", "2019-08", "2020-01", "2020-04", "2024-03"], "last_mentioned": "2024-03-22T22:15:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1801.06146", "slug": "804d47449f5005d2", "title": "Universal Language Model Fine-tuning for Text Classification", "authors": ["Jeremy Howard", "Sebastian Ruder"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1801.06146", "doi": null, "topics": ["Language models", "Computer vision"], "description": "Inductive transfer learning has greatly impacted computer vision, but existing approaches in NLP still require task-specific modifications and training from scratch.", "description_method": "abstract-excerpt", "abstract": "Inductive transfer learning has greatly impacted computer vision, but existing approaches in NLP still require task-specific modifications and training from scratch. We propose Universal Language Model Fine-tuning (ULMFiT), an effective transfer learning method that can be applied to any task in NLP, and introduce techniques that are key for fine-tuning a language model. Our method significantly outperforms the state-of-the-art on six text classification tasks, reducing the error by 18-24% on the majority of datasets. Furthermore, with only 100 labeled examples, it matches the performance of training from scratch on 100x more data. We open-source our pretrained models and code.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2018-02", "2018-05", "2018-06", "2018-10", "2020-07"], "last_mentioned": "2020-07-07T16:49:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1802.03426", "slug": "e6291087d00249c5", "title": "UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction", "authors": ["Leland McInnes", "John Healy", "James Melville"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1802.03426", "doi": null, "topics": ["Mathematics", "Algorithms", "Machine learning"], "description": "UMAP (Uniform Manifold Approximation and Projection) is a novel manifold learning technique for dimension reduction.", "description_method": "abstract-excerpt", "abstract": "UMAP (Uniform Manifold Approximation and Projection) is a novel manifold learning technique for dimension reduction. UMAP is constructed from a theoretical framework based in Riemannian geometry and algebraic topology. The result is a practical scalable algorithm that applies to real world data. The UMAP algorithm is competitive with t-SNE for visualization quality, and arguably preserves more of the global structure with superior run time performance. Furthermore, UMAP has no computational restrictions on embedding dimension, making it viable as a general purpose dimension reduction technique for machine learning.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2018-08", "2019-01", "2019-04", "2024-10"], "last_mentioned": "2024-10-25T17:26:39Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1802.08232", "slug": "5d1a373008dbdbb5", "title": "The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networks", "authors": ["Nicholas Carlini", "Chang Liu", "Úlfar Erlingsson", "Jernej Kos", "Dawn Song"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1802.08232", "doi": null, "topics": ["Machine learning", "Security & cryptography"], "description": "This paper describes a testing methodology for quantitatively assessing the risk that rare or unique training-data sequences are unintentionally memorized by generative sequence models---a common type of machine-learning model.", "description_method": "abstract-excerpt", "abstract": "This paper describes a testing methodology for quantitatively assessing the risk that rare or unique training-data sequences are unintentionally memorized by generative sequence models---a common type of machine-learning model. Because such models are sometimes trained on sensitive data (e.g., the text of users' private messages), this methodology can benefit privacy by allowing deep-learning practitioners to select means of training that minimize such memorization. In experiments, we show that unintended memorization is a persistent, hard-to-avoid issue that can have serious consequences. Specifically, for models trained without consideration of memorization, we describe new, efficient procedures that can extract unique, secret sequences, such as credit card numbers. We show that our testing strategy is a practical and easy-to-use first line of defense, e.g., by describing its application to quantitatively limit data exposure in Google's Smart Compose, a commercial text-completion neural network trained on millions of users' email messages.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2019-05", "2019-10", "2019-11", "2021-01"], "last_mentioned": "2021-01-06T05:06:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.07770", "slug": "16af0e83a7c8bc61", "title": "Emergence of grid-like representations by training recurrent neural networks to perform spatial localization", "authors": ["Christopher J. Cueva", "Xue-Xin Wei"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.07770", "doi": null, "topics": ["Biology & medicine", "Neuroscience", "Machine learning"], "description": "Decades of research on the neural code underlying spatial navigation have revealed a diverse set of neural response properties.", "description_method": "abstract-excerpt", "abstract": "Decades of research on the neural code underlying spatial navigation have revealed a diverse set of neural response properties. The Entorhinal Cortex (EC) of the mammalian brain contains a rich set of spatial correlates, including grid cells which encode space using tessellating patterns. However, the mechanisms and functional significance of these spatial representations remain largely mysterious. As a new way to understand these neural representations, we trained recurrent neural networks (RNNs) to perform navigation tasks in 2D arenas based on velocity inputs. Surprisingly, we find that grid-like spatial response patterns emerge in trained networks, along with units that exhibit other spatial correlates, including border cells and band-like cells. All these different functional types of neurons have been observed experimentally. The order of the emergence of grid-like and border cells is also consistent with observations from developmental studies. Together, our results suggest that grid cells, border cells and others as observed in EC may be a natural solution for representing space efficiently given the predominant recurrent connections in the neural circuits.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2018-05", "2019-02", "2022-04", "2024-06"], "last_mentioned": "2024-06-16T14:12:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1806.04558", "slug": "8b206e72f90cb5e5", "title": "Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis", "authors": ["Ye Jia", "Yu Zhang", "Ron J. Weiss", "Quan Wang", "Jonathan Shen", "Fei Ren", "Zhifeng Chen", "Patrick Nguyen", "Ruoming Pang", "Ignacio Lopez Moreno", "Yonghui Wu"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1806.04558", "doi": null, "topics": ["Machine learning"], "description": "We describe a neural network-based system for text-to-speech (TTS) synthesis that is able to generate speech audio in the voice of many different speakers, including those unseen during training.", "description_method": "abstract-excerpt", "abstract": "We describe a neural network-based system for text-to-speech (TTS) synthesis that is able to generate speech audio in the voice of many different speakers, including those unseen during training. Our system consists of three independently trained components: (1) a speaker encoder network, trained on a speaker verification task using an independent dataset of noisy speech from thousands of speakers without transcripts, to generate a fixed-dimensional embedding vector from seconds of reference speech from a target speaker; (2) a sequence-to-sequence synthesis network based on Tacotron 2, which generates a mel spectrogram from text, conditioned on the speaker embedding; (3) an auto-regressive WaveNet-based vocoder that converts the mel spectrogram into a sequence of time domain waveform samples. We demonstrate that the proposed model is able to transfer the knowledge of speaker variability learned by the discriminatively-trained speaker encoder to the new task, and is able to synthesize natural speech from speakers that were not seen during training. We quantify the importance of training the speaker encoder on a large and diverse speaker set in order to obtain the best generalization performance. Finally, we show that randomly sampled speaker embeddings can be used to synthesize speech in the voice of novel speakers dissimilar from those used in training, indicating that the model has learned a high quality speaker representation.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2020-06", "2021-01", "2021-09"], "last_mentioned": "2021-09-25T08:59:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1810.07951", "slug": "ac21c14fb65ce921", "title": "Don't Unroll Adjoint: Differentiating SSA-Form Programs", "authors": ["Michael Innes"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1810.07951", "doi": null, "topics": ["Programming languages", "Algorithms", "Human–computer interaction"], "description": "This paper presents reverse-mode algorithmic differentiation (AD) based on source code transformation, in particular of the Static Single Assignment (SSA) form used by modern compilers.", "description_method": "abstract-excerpt", "abstract": "This paper presents reverse-mode algorithmic differentiation (AD) based on source code transformation, in particular of the Static Single Assignment (SSA) form used by modern compilers. The approach can support control flow, nesting, mutation, recursion, data structures, higher-order functions, and other language constructs, and the output is given to an existing compiler to produce highly efficient differentiated code. Our implementation is a new AD tool for the Julia language, called Zygote, which presents high-level dynamic semantics while transparently compiling adjoint code under the hood. We discuss the benefits of this approach to both the usability and performance of AD tools.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2019-07", "2020-02", "2020-10", "2021-07", "2021-12"], "last_mentioned": "2021-12-25T16:30:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1810.11490", "slug": "5c98958aed7d6f78", "title": "Could Solar Radiation Pressure Explain 'Oumuamua's Peculiar Acceleration?", "authors": ["Shmuel Bialy", "Abraham Loeb"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1810.11490", "doi": "10.3847/2041-8213/aaeda8", "topics": ["Statistics & probability"], "description": "`Oumuamua (1I/2017 U1) is the first object of interstellar origin observed in the Solar System.", "description_method": "abstract-excerpt", "abstract": "`Oumuamua (1I/2017 U1) is the first object of interstellar origin observed in the Solar System. Recently, \\citet{Micheli2018} reported that `Oumuamua showed deviations from a Keplerian orbit at a high statistical significance. The observed trajectory is best explained by an excess radial acceleration $Δa \\propto r^{-2}$, where $r$ is the distance of `Oumuamua from the Sun. Such an acceleration is naturally expected for comets, driven by the evaporating material. However, recent observational and theoretical studies imply that `Oumuamua is not an active comet. We explore the possibility that the excess acceleration results from Solar radiation pressure. The required mass-to-area ratio is $(m/A)\\approx 0.1$ g cm$^{-2}$. For a thin sheet this requires a thickness of $\\approx 0.3-0.9$ mm. We find that although extremely thin, such an object would survive an interstellar travel over Galactic distances of $\\sim 5$ kpc, withstanding collisions with gas and dust-grains as well as stresses from rotation and tidal forces. We discuss the possible origins of such an object. Our general results apply to any light probes designed for interstellar travel.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2018-11", "2019-01", "2019-02", "2021-01"], "last_mentioned": "2021-01-21T23:03:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1811.10154", "slug": "62d3d7b3b4d020d2", "title": "Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead", "authors": ["Cynthia Rudin"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1811.10154", "doi": null, "topics": ["Machine learning", "Psychology", "Computer vision"], "description": "Black box machine learning models are currently being used for high stakes decision-making throughout society, causing problems throughout healthcare, criminal justice, and in other domains.", "description_method": "abstract-excerpt", "abstract": "Black box machine learning models are currently being used for high stakes decision-making throughout society, causing problems throughout healthcare, criminal justice, and in other domains. People have hoped that creating methods for explaining these black box models will alleviate some of these problems, but trying to \\textit{explain} black box models, rather than creating models that are \\textit{interpretable} in the first place, is likely to perpetuate bad practices and can potentially cause catastrophic harm to society. There is a way forward -- it is to design models that are inherently interpretable. This manuscript clarifies the chasm between explaining black boxes and using inherently interpretable models, outlines several key reasons why explainable black boxes should be avoided in high-stakes decisions, identifies challenges to interpretable machine learning, and provides several example applications where interpretable models could potentially replace black box models in criminal justice, healthcare, and computer vision.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2020-04", "2021-10", "2022-06", "2024-04"], "last_mentioned": "2024-04-04T13:06:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1812.04948", "slug": "5bff703902010a2e", "title": "A Style-Based Generator Architecture for Generative Adversarial Networks", "authors": ["Tero Karras", "Samuli Laine", "Timo Aila"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1812.04948", "doi": null, "topics": ["Statistics & probability"], "description": "We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature.", "description_method": "abstract-excerpt", "abstract": "We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on human faces) and stochastic variation in the generated images (e.g., freckles, hair), and it enables intuitive, scale-specific control of the synthesis. The new generator improves the state-of-the-art in terms of traditional distribution quality metrics, leads to demonstrably better interpolation properties, and also better disentangles the latent factors of variation. To quantify interpolation quality and disentanglement, we propose two new, automated methods that are applicable to any generator architecture. Finally, we introduce a new, highly varied and high-quality dataset of human faces.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2019-02", "2021-06", "2021-10"], "last_mentioned": "2021-10-21T06:41:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1902.10565", "slug": "4b5a2e9d48fad5f9", "title": "Accelerating Self-Play Learning in Go", "authors": ["David J. Wu"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1902.10565", "doi": null, "topics": ["Statistics & probability", "Distributed systems"], "description": "By introducing several improvements to the AlphaZero process and architecture, we greatly accelerate self-play learning in Go, achieving a 50x reduction in computation over comparable methods.", "description_method": "abstract-excerpt", "abstract": "By introducing several improvements to the AlphaZero process and architecture, we greatly accelerate self-play learning in Go, achieving a 50x reduction in computation over comparable methods. Like AlphaZero and replications such as ELF OpenGo and Leela Zero, our bot KataGo only learns from neural-net-guided Monte Carlo tree search self-play. But whereas AlphaZero required thousands of TPUs over several days and ELF required thousands of GPUs over two weeks, KataGo surpasses ELF's final model after only 19 days on fewer than 30 GPUs. Much of the speedup involves non-domain-specific improvements that might directly transfer to other problems. Further gains from domain-specific techniques reveal the remaining efficiency gap between the best methods and purely general methods such as AlphaZero. Our work is a step towards making learning in state spaces as large as Go possible without large-scale computational resources.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2019-11", "2019-12", "2020-06", "2022-11"], "last_mentioned": "2022-11-03T17:30:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1902.10811", "slug": "6c04dc731c6ee13d", "title": "Do ImageNet Classifiers Generalize to ImageNet?", "authors": ["Benjamin Recht", "Rebecca Roelofs", "Ludwig Schmidt", "Vaishaal Shankar"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1902.10811", "doi": null, "topics": ["Machine learning"], "description": "We build new test sets for the CIFAR-10 and ImageNet datasets.", "description_method": "abstract-excerpt", "abstract": "We build new test sets for the CIFAR-10 and ImageNet datasets. Both benchmarks have been the focus of intense research for almost a decade, raising the danger of overfitting to excessively re-used test sets. By closely following the original dataset creation processes, we test to what extent current classification models generalize to new data. We evaluate a broad range of models and find accuracy drops of 3% - 15% on CIFAR-10 and 11% - 14% on ImageNet. However, accuracy gains on the original test sets translate to larger gains on the new test sets. Our results suggest that the accuracy drops are not caused by adaptivity, but by the models' inability to generalize to slightly \"harder\" images than those found in the original test sets.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2019-09", "2020-09", "2021-02", "2021-04", "2021-12"], "last_mentioned": "2021-12-06T17:10:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1907.05607", "slug": "f2665c28beefe8cd", "title": "A strong no-go theorem on the Wigner's friend paradox", "authors": ["Kok-Wei Bong", "Aníbal Utreras-Alarcón", "Farzad Ghafari", "Yeong-Cherng Liang", "Nora Tischler", "Eric G. Cavalcanti", "Geoff J. Pryde", "Howard M. Wiseman"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1907.05607", "doi": "10.1038/s41567-020-0990-x", "topics": ["Physics", "Mathematics"], "description": "Does quantum theory apply at all scales, including that of observers?", "description_method": "abstract-excerpt", "abstract": "Does quantum theory apply at all scales, including that of observers? New light on this fundamental question has recently been shed through a resurgence of interest in the long-standing Wigner's friend paradox. This is a thought experiment addressing the quantum measurement problem -- the difficulty of reconciling the (unitary, deterministic) evolution of isolated systems and the (non-unitary, probabilistic) state update after a measurement. Here, by building on a scenario with two separated but entangled friends introduced by Brukner, we prove that if quantum evolution is controllable on the scale of an observer, then one of 'No-Superdeterminism', 'Locality' or 'Absoluteness of Observed Events' -- that every observed event exists absolutely, not relatively -- must be false. We show that although the violation of Bell-type inequalities in such scenarios is not in general sufficient to demonstrate the contradiction between those three assumptions, new inequalities can be derived in a theory-independent manner, that are violated by quantum correlations. This is demonstrated in a proof-of-principle experiment where a photon's path is deemed an observer. We discuss how this new theorem places strictly stronger constraints on physical reality than Bell's theorem.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2020-08", "2020-09", "2021-12", "2024-07"], "last_mentioned": "2024-07-31T03:45:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1908.02499", "slug": "760fba39667028d9", "title": "The Argument against Quantum Computers", "authors": ["Gil Kalai"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1908.02499", "doi": null, "topics": ["Physics", "Algorithms", "Statistics & probability"], "description": "We give a computational complexity argument against the feasibility of quantum computers.", "description_method": "abstract-excerpt", "abstract": "We give a computational complexity argument against the feasibility of quantum computers. We identify a very low complexity class of probability distributions described by noisy intermediate-scale quantum computers, and explain why it will allow neither good-quality quantum error-correction nor a demonstration of \"quantum supremacy.\" Some general principles governing the behavior of noisy quantum systems are derived. Our work supports the \"physical Church thesis\" studied by Pitowsky (1990) and follows his vision of using abstract ideas about computation to study the performance of actual physical computers.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2019-10", "2019-12", "2021-10", "2024-12"], "last_mentioned": "2024-12-11T21:27:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1909.11090", "slug": "e40c8e7fe8caa26c", "title": "What if Planet 9 is a Primordial Black Hole?", "authors": ["Jakub Scholtz", "James Unwin"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1909.11090", "doi": "10.1103/PhysRevLett.125.051103", "topics": ["Physics", "Statistics & probability"], "description": "We highlight that the anomalous orbits of Trans-Neptunian Objects (TNOs) and an excess in microlensing events in the 5-year OGLE dataset can be simultaneously explained by a new population of astrophysical bodies with mass several times that of Earth ($M_\\oplus$).", "description_method": "abstract-excerpt", "abstract": "We highlight that the anomalous orbits of Trans-Neptunian Objects (TNOs) and an excess in microlensing events in the 5-year OGLE dataset can be simultaneously explained by a new population of astrophysical bodies with mass several times that of Earth ($M_\\oplus$). We take these objects to be primordial black holes (PBHs) and point out the orbits of TNOs would be altered if one of these PBHs was captured by the Solar System, inline with the Planet 9 hypothesis. Capture of a free floating planet is a leading explanation for the origin of Planet 9 and we show that the probability of capturing a PBH instead is comparable. The observational constraints on a PBH in the outer Solar System significantly differ from the case of a new ninth planet. This scenario could be confirmed through annihilation signals from the dark matter microhalo around the PBH.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2019-10", "2020-07", "2024-04", "2024-05", "2025-05"], "last_mentioned": "2025-05-05T12:40:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1910.06709", "slug": "e454d2b1fd650e8e", "title": "A Simple Proof of the Quadratic Formula", "authors": ["Po-Shen Loh"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1910.06709", "doi": null, "topics": ["Other research"], "description": "This article provides a simple proof of the quadratic formula, which also produces an efficient and natural method for solving general quadratic equations.", "description_method": "abstract-excerpt", "abstract": "This article provides a simple proof of the quadratic formula, which also produces an efficient and natural method for solving general quadratic equations. The derivation is computationally light and conceptually natural, and has the potential to demystify quadratic equations for students worldwide.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2021-11", "2022-10", "2024-04", "2026-05"], "last_mentioned": "2026-05-08T12:54:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1912.01412", "slug": "7e3429eb87167d3f", "title": "Deep Learning for Symbolic Mathematics", "authors": ["Guillaume Lample", "François Charton"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1912.01412", "doi": null, "topics": ["Mathematics", "Machine learning", "Statistics & probability"], "description": "Neural networks have a reputation for being better at solving statistical or approximate problems than at performing calculations or working with symbolic data.", "description_method": "abstract-excerpt", "abstract": "Neural networks have a reputation for being better at solving statistical or approximate problems than at performing calculations or working with symbolic data. In this paper, we show that they can be surprisingly good at more elaborated tasks in mathematics, such as symbolic integration and solving differential equations. We propose a syntax for representing mathematical problems, and methods for generating large datasets that can be used to train sequence-to-sequence models. We achieve results that outperform commercial Computer Algebra Systems such as Matlab or Mathematica.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2020-05", "2020-08", "2022-05", "2023-03"], "last_mentioned": "2023-03-21T20:31:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1912.04958", "slug": "e91cdf9a498aed1a", "title": "Analyzing and Improving the Image Quality of StyleGAN", "authors": ["Tero Karras", "Samuli Laine", "Miika Aittala", "Janne Hellsten", "Jaakko Lehtinen", "Timo Aila"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1912.04958", "doi": null, "topics": ["Other research"], "description": "The style-based GAN architecture (StyleGAN) yields state-of-the-art results in data-driven unconditional generative image modeling.", "description_method": "abstract-excerpt", "abstract": "The style-based GAN architecture (StyleGAN) yields state-of-the-art results in data-driven unconditional generative image modeling. We expose and analyze several of its characteristic artifacts, and propose changes in both model architecture and training methods to address them. In particular, we redesign the generator normalization, revisit progressive growing, and regularize the generator to encourage good conditioning in the mapping from latent codes to images. In addition to improving image quality, this path length regularizer yields the additional benefit that the generator becomes significantly easier to invert. This makes it possible to reliably attribute a generated image to a particular network. We furthermore visualize how well the generator utilizes its output resolution, and identify a capacity problem, motivating us to train larger models for additional quality improvements. Overall, our improved model redefines the state of the art in unconditional image modeling, both in terms of existing distribution quality metrics as well as perceived image quality.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2020-02", "2021-02", "2021-06", "2022-07", "2024-10"], "last_mentioned": "2024-10-28T07:01:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1912.13321", "slug": "cc59449cbef75efd", "title": "OTEANN: Estimating the Transparency of Orthographies with an Artificial Neural Network", "authors": ["Xavier Marjou"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1912.13321", "doi": "10.18653/v1/2021.sigtyp-1.1", "topics": ["Machine learning"], "description": "To transcribe spoken language to written medium, most alphabets enable an unambiguous sound-to-letter rule.", "description_method": "abstract-excerpt", "abstract": "To transcribe spoken language to written medium, most alphabets enable an unambiguous sound-to-letter rule. However, some writing systems have distanced themselves from this simple concept and little work exists in Natural Language Processing (NLP) on measuring such distance. In this study, we use an Artificial Neural Network (ANN) model to evaluate the transparency between written words and their pronunciation, hence its name Orthographic Transparency Estimation with an ANN (OTEANN). Based on datasets derived from Wikimedia dictionaries, we trained and tested this model to score the percentage of correct predictions in phoneme-to-grapheme and grapheme-to-phoneme translation tasks. The scores obtained on 17 orthographies were in line with the estimations of other studies. Interestingly, the model also provided insight into typical mistakes made by learners who only consider the phonemic rule in reading and writing.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2020-11", "2023-05", "2025-07"], "last_mentioned": "2025-07-07T20:45:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2001.04383", "slug": "b26ca583212c1c85", "title": "MIP*=RE", "authors": ["Zhengfeng Ji", "Anand Natarajan", "Thomas Vidick", "John Wright", "Henry Yuen"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2001.04383", "doi": null, "topics": ["Physics", "Mathematics"], "description": "We show that the class MIP* of languages that can be decided by a classical verifier interacting with multiple all-powerful quantum provers sharing entanglement is equal to the class RE of recursively enumerable languages.", "description_method": "abstract-excerpt", "abstract": "We show that the class MIP* of languages that can be decided by a classical verifier interacting with multiple all-powerful quantum provers sharing entanglement is equal to the class RE of recursively enumerable languages. Our proof builds upon the quantum low-degree test of (Natarajan and Vidick, FOCS 2018) and the classical low-individual degree test of (Ji, et al., 2020) by integrating recent developments from (Natarajan and Wright, FOCS 2019) and combining them with the recursive compression framework of (Fitzsimons et al., STOC 2019). An immediate byproduct of our result is that there is an efficient reduction from the Halting Problem to the problem of deciding whether a two-player nonlocal game has entangled value $1$ or at most $1/2$. Using a known connection, undecidability of the entangled value implies a negative answer to Tsirelson's problem: we show, by providing an explicit example, that the closure $C_{qa}$ of the set of quantum tensor product correlations is strictly included in the set $C_{qc}$ of quantum commuting correlations. Following work of (Fritz, Rev. Math. Phys. 2012) and (Junge et al., J. Math. Phys. 2011) our results provide a refutation of Connes' embedding conjecture from the theory of von Neumann algebras.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2020-03", "2020-04", "2023-03", "2023-08"], "last_mentioned": "2023-08-19T04:04:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2006.04182", "slug": "6aa3bf72f5f0d8b1", "title": "Predictive Coding Approximates Backprop along Arbitrary Computation Graphs", "authors": ["Beren Millidge", "Alexander Tschantz", "Christopher L. Buckley"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2006.04182", "doi": null, "topics": ["Machine learning", "Biology & medicine", "Algorithms"], "description": "Backpropagation of error (backprop) is a powerful algorithm for training machine learning architectures through end-to-end differentiation.", "description_method": "abstract-excerpt", "abstract": "Backpropagation of error (backprop) is a powerful algorithm for training machine learning architectures through end-to-end differentiation. However, backprop is often criticised for lacking biological plausibility. Recently, it has been shown that backprop in multilayer-perceptrons (MLPs) can be approximated using predictive coding, a biologically-plausible process theory of cortical computation which relies only on local and Hebbian updates. The power of backprop, however, lies not in its instantiation in MLPs, but rather in the concept of automatic differentiation which allows for the optimisation of any differentiable program expressed as a computation graph. Here, we demonstrate that predictive coding converges asymptotically (and in practice rapidly) to exact backprop gradients on arbitrary computation graphs using only local learning rules. We apply this result to develop a straightforward strategy to translate core machine learning architectures into their predictive coding equivalents. We construct predictive coding CNNs, RNNs, and the more complex LSTMs, which include a non-layer-like branching internal graph structure and multiplicative interactions. Our models perform equivalently to backprop on challenging machine learning benchmarks, while utilising only local and (mostly) Hebbian plasticity. Our method raises the potential that standard machine learning algorithms could in principle be directly implemented in neural circuitry, and may also contribute to the development of completely distributed neuromorphic architectures.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2021-04", "2023-09", "2024-05"], "last_mentioned": "2024-05-12T13:24:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2006.07125", "slug": "427e375c153e0477", "title": "Breaking the Warp Barrier: Hyper-Fast Solitons in Einstein-Maxwell-Plasma Theory", "authors": ["Erik W. Lentz"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2006.07125", "doi": null, "topics": ["Physics"], "description": "Solitons in space--time capable of transporting time-like observers at superluminal speeds have long been tied to violations of the weak, strong, and dominant energy conditions of general relativity.", "description_method": "abstract-excerpt", "abstract": "Solitons in space--time capable of transporting time-like observers at superluminal speeds have long been tied to violations of the weak, strong, and dominant energy conditions of general relativity. The negative-energy sources required for these solitons must be created through energy-intensive uncertainty principle processes as no such classical source is known in particle physics. This paper overcomes this barrier by constructing a class of soliton solutions that are capable of superluminal motion and sourced by purely positive energy densities. The solitons are also shown to be capable of being sourced from the stress-energy of a conducting plasma and classical electromagnetic fields. This is the first example of hyper-fast solitons resulting from known and familiar sources, reopening the discussion of superluminal mechanisms rooted in conventional physics.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2021-03", "2021-05", "2021-09", "2022-01", "2023-03"], "last_mentioned": "2023-03-18T17:32:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2006.11239", "slug": "78296905fc8ebbf4", "title": "Denoising Diffusion Probabilistic Models", "authors": ["Jonathan Ho", "Ajay Jain", "Pieter Abbeel"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2006.11239", "doi": null, "topics": ["Physics", "Machine learning"], "description": "We present high quality image synthesis results using diffusion probabilistic models, a class of latent variable models inspired by considerations from nonequilibrium thermodynamics.", "description_method": "abstract-excerpt", "abstract": "We present high quality image synthesis results using diffusion probabilistic models, a class of latent variable models inspired by considerations from nonequilibrium thermodynamics. Our best results are obtained by training on a weighted variational bound designed according to a novel connection between diffusion probabilistic models and denoising score matching with Langevin dynamics, and our models naturally admit a progressive lossy decompression scheme that can be interpreted as a generalization of autoregressive decoding. On the unconditional CIFAR10 dataset, we obtain an Inception score of 9.46 and a state-of-the-art FID score of 3.17. On 256x256 LSUN, we obtain sample quality similar to ProgressiveGAN. Our implementation is available at https://github.com/hojonathanho/diffusion", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2022-09", "2023-01", "2023-07", "2024-02", "2024-05"], "last_mentioned": "2024-05-26T21:33:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2104.12871", "slug": "b9a0a164bb58154b", "title": "Why AI is Harder Than We Think", "authors": ["Melanie Mitchell"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2104.12871", "doi": null, "topics": ["Psychology", "Robotics", "Algorithms"], "description": "Since its beginning in the 1950s, the field of artificial intelligence has cycled several times between periods of optimistic predictions and massive investment (\"AI spring\") and periods of disappointment, loss of confidence, and reduced funding (\"AI winter\").", "description_method": "abstract-excerpt", "abstract": "Since its beginning in the 1950s, the field of artificial intelligence has cycled several times between periods of optimistic predictions and massive investment (\"AI spring\") and periods of disappointment, loss of confidence, and reduced funding (\"AI winter\"). Even with today's seemingly fast pace of AI breakthroughs, the development of long-promised technologies such as self-driving cars, housekeeping robots, and conversational companions has turned out to be much harder than many people expected. One reason for these repeating cycles is our limited understanding of the nature and complexity of intelligence itself. In this paper I describe four fallacies in common assumptions made by AI researchers, which can lead to overconfident predictions about the field. I conclude by discussing the open questions spurred by these fallacies, including the age-old challenge of imbuing machines with humanlike common sense.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2021-04", "2021-05", "2023-01", "2025-09"], "last_mentioned": "2025-09-11T06:31:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2105.03824", "slug": "b81a5e974f1390a6", "title": "FNet: Mixing Tokens with Fourier Transforms", "authors": ["James Lee-Thorp", "Joshua Ainslie", "Ilya Eckstein", "Santiago Ontanon"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2105.03824", "doi": null, "topics": ["Language models"], "description": "We show that Transformer encoder architectures can be sped up, with limited accuracy costs, by replacing the self-attention sublayers with simple linear transformations that \"mix\" input tokens.", "description_method": "abstract-excerpt", "abstract": "We show that Transformer encoder architectures can be sped up, with limited accuracy costs, by replacing the self-attention sublayers with simple linear transformations that \"mix\" input tokens. These linear mixers, along with standard nonlinearities in feed-forward layers, prove competent at modeling semantic relationships in several text classification tasks. Most surprisingly, we find that replacing the self-attention sublayer in a Transformer encoder with a standard, unparameterized Fourier Transform achieves 92-97% of the accuracy of BERT counterparts on the GLUE benchmark, but trains 80% faster on GPUs and 70% faster on TPUs at standard 512 input lengths. At longer input lengths, our FNet model is significantly faster: when compared to the \"efficient\" Transformers on the Long Range Arena benchmark, FNet matches the accuracy of the most accurate models, while outpacing the fastest models across all sequence lengths on GPUs (and across relatively shorter lengths on TPUs). Finally, FNet has a light memory footprint and is particularly efficient at smaller model sizes; for a fixed speed and accuracy budget, small FNet models outperform Transformer counterparts.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2024-05", "2025-02", "2026-03"], "last_mentioned": "2026-03-03T19:22:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2107.10356", "slug": "48f5ef120015cc64", "title": "Reading Race: AI Recognises Patient's Racial Identity In Medical Images", "authors": ["Imon Banerjee", "Ananth Reddy Bhimireddy", "John L. Burns", "Leo Anthony Celi", "Li-Ching Chen", "Ramon Correa", "Natalie Dullerud", "Marzyeh Ghassemi", "Shih-Cheng Huang", "Po-Chih Kuo", "Matthew P Lungren", "Lyle Palmer", "Brandon J Price", "Saptarshi Purkayastha", "Ayis Pyrros", "Luke Oakden-Rayner", "Chima Okechukwu", "Laleh Seyyed-Kalantari", "Hari Trivedi", "Ryan Wang", "Zachary Zaiman", "Haoran Zhang", "Judy W Gichoya"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2107.10356", "doi": "10.1016/S2589-7500(22)00063-2", "topics": ["Biology & medicine", "Machine learning"], "description": "Background: In medical imaging, prior studies have demonstrated disparate AI performance by race, yet there is no known correlation for race on medical imaging that would be obvious to the human expert interpreting the images.", "description_method": "abstract-excerpt", "abstract": "Background: In medical imaging, prior studies have demonstrated disparate AI performance by race, yet there is no known correlation for race on medical imaging that would be obvious to the human expert interpreting the images. Methods: Using private and public datasets we evaluate: A) performance quantification of deep learning models to detect race from medical images, including the ability of these models to generalize to external environments and across multiple imaging modalities, B) assessment of possible confounding anatomic and phenotype population features, such as disease distribution and body habitus as predictors of race, and C) investigation into the underlying mechanism by which AI models can recognize race. Findings: Standard deep learning models can be trained to predict race from medical images with high performance across multiple imaging modalities. Our findings hold under external validation conditions, as well as when models are optimized to perform clinically motivated tasks. We demonstrate this detection is not due to trivial proxies or imaging-related surrogate covariates for race, such as underlying disease distribution. Finally, we show that performance persists over all anatomical regions and frequency spectrum of the images suggesting that mitigation efforts will be challenging and demand further study. Interpretation: We emphasize that model ability to predict self-reported race is itself not the issue of importance. However, our findings that AI can trivially predict self-reported race -- even from corrupted, cropped, and noised medical images -- in a setting where clinical experts cannot, creates an enormous risk for all model deployments in medical imaging: if an AI model secretly used its knowledge of self-reported race to misclassify all Black patients, radiologists would not be able to tell using the same data the model has access to.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2021-08", "2021-09", "2021-12"], "last_mentioned": "2021-12-11T15:30:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2112.00114", "slug": "3ec28a4f81c78aea", "title": "Show Your Work: Scratchpads for Intermediate Computation with Language Models", "authors": ["Maxwell Nye", "Anders Johan Andreassen", "Guy Gur-Ari", "Henryk Michalewski", "Jacob Austin", "David Bieber", "David Dohan", "Aitor Lewkowycz", "Maarten Bosma", "David Luan", "Charles Sutton", "Augustus Odena"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2112.00114", "doi": null, "topics": ["Language models"], "description": "Large pre-trained language models perform remarkably well on tasks that can be done \"in one pass\", such as generating realistic text or synthesizing computer programs.", "description_method": "abstract-excerpt", "abstract": "Large pre-trained language models perform remarkably well on tasks that can be done \"in one pass\", such as generating realistic text or synthesizing computer programs. However, they struggle with tasks that require unbounded multi-step computation, such as adding integers or executing programs. Surprisingly, we find that these same models are able to perform complex multi-step computations -- even in the few-shot regime -- when asked to perform the operation \"step by step\", showing the results of intermediate computations. In particular, we train transformers to perform multi-step computations by asking them to emit intermediate computation steps into a \"scratchpad\". On a series of increasingly complex tasks ranging from long addition to the execution of arbitrary programs, we show that scratchpads dramatically improve the ability of language models to perform multi-step computations.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2022-02", "2023-03", "2025-11", "2026-04"], "last_mentioned": "2026-04-05T13:12:44Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2202.03629", "slug": "d2703890beb32edf", "title": "Survey of Hallucination in Natural Language Generation", "authors": ["Ziwei Ji", "Nayeon Lee", "Rita Frieske", "Tiezheng Yu", "Dan Su", "Yan Xu", "Etsuko Ishii", "Yejin Bang", "Delong Chen", "Wenliang Dai", "Ho Shu Chan", "Andrea Madotto", "Pascale Fung"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2202.03629", "doi": "10.1145/3571730", "topics": ["Language models", "Machine learning", "Economics"], "description": "Natural Language Generation (NLG) has improved exponentially in recent years thanks to the development of sequence-to-sequence deep learning technologies such as Transformer-based language models.", "description_method": "abstract-excerpt", "abstract": "Natural Language Generation (NLG) has improved exponentially in recent years thanks to the development of sequence-to-sequence deep learning technologies such as Transformer-based language models. This advancement has led to more fluent and coherent NLG, leading to improved development in downstream tasks such as abstractive summarization, dialogue generation and data-to-text generation. However, it is also apparent that deep learning based generation is prone to hallucinate unintended text, which degrades the system performance and fails to meet user expectations in many real-world scenarios. To address this issue, many studies have been presented in measuring and mitigating hallucinated texts, but these have never been reviewed in a comprehensive manner before. In this survey, we thus provide a broad overview of the research progress and challenges in the hallucination problem in NLG. The survey is organized into two parts: (1) a general overview of metrics, mitigation methods, and future directions; (2) an overview of task-specific research progress on hallucinations in the following downstream tasks, namely abstractive summarization, dialogue generation, generative question answering, data-to-text generation, machine translation, and visual-language generation; and (3) hallucinations in large language models (LLMs). This survey serves to facilitate collaborative efforts among researchers in tackling the challenge of hallucinated texts in NLG.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2023-02", "2023-03", "2023-09", "2024-02"], "last_mentioned": "2024-02-26T05:46:57Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2205.05982", "slug": "35a53661c985321d", "title": "Vectorized and performance-portable Quicksort", "authors": ["Mark Blacher", "Joachim Giesen", "Peter Sanders", "Jan Wassenberg"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2205.05982", "doi": null, "topics": ["Algorithms"], "description": "Recent works showed that implementations of Quicksort using vector CPU instructions can outperform the non-vectorized algorithms in widespread use.", "description_method": "abstract-excerpt", "abstract": "Recent works showed that implementations of Quicksort using vector CPU instructions can outperform the non-vectorized algorithms in widespread use. However, these implementations are typically single-threaded, implemented for a particular instruction set, and restricted to a small set of key types. We lift these three restrictions: our proposed 'vqsort' algorithm integrates into the state-of-the-art parallel sorter 'ips4o', with a geometric mean speedup of 1.59. The same implementation works on seven instruction sets (including SVE and RISC-V V) across four platforms. It also supports floating-point and 16-128 bit integer keys. To the best of our knowledge, this is the fastest sort for non-tuple keys on CPUs, up to 20 times as fast as the sorting algorithms implemented in standard libraries. This paper focuses on the practical engineering aspects enabling the speed and portability, which we have not yet seen demonstrated for a Quicksort implementation. Furthermore, we introduce compact and transpose-free sorting networks for in-register sorting of small arrays, and a vector-friendly pivot sampling strategy that is robust against adversarial input.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2022-06", "2022-12", "2023-01", "2024-03"], "last_mentioned": "2024-03-01T19:38:33Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2205.11487", "slug": "0dc901faf0b59bd4", "title": "Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding", "authors": ["Chitwan Saharia", "William Chan", "Saurabh Saxena", "Lala Li", "Jay Whang", "Emily Denton", "Seyed Kamyar Seyed Ghasemipour", "Burcu Karagol Ayan", "S. Sara Mahdavi", "Rapha Gontijo Lopes", "Tim Salimans", "Jonathan Ho", "David J Fleet", "Mohammad Norouzi"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2205.11487", "doi": null, "topics": ["Physics", "Machine learning", "Language models"], "description": "We present Imagen, a text-to-image diffusion model with an unprecedented degree of photorealism and a deep level of language understanding.", "description_method": "abstract-excerpt", "abstract": "We present Imagen, a text-to-image diffusion model with an unprecedented degree of photorealism and a deep level of language understanding. Imagen builds on the power of large transformer language models in understanding text and hinges on the strength of diffusion models in high-fidelity image generation. Our key discovery is that generic large language models (e.g. T5), pretrained on text-only corpora, are surprisingly effective at encoding text for image synthesis: increasing the size of the language model in Imagen boosts both sample fidelity and image-text alignment much more than increasing the size of the image diffusion model. Imagen achieves a new state-of-the-art FID score of 7.27 on the COCO dataset, without ever training on COCO, and human raters find Imagen samples to be on par with the COCO data itself in image-text alignment. To assess text-to-image models in greater depth, we introduce DrawBench, a comprehensive and challenging benchmark for text-to-image models. With DrawBench, we compare Imagen with recent methods including VQ-GAN+CLIP, Latent Diffusion Models, and DALL-E 2, and find that human raters prefer Imagen over other models in side-by-side comparisons, both in terms of sample quality and image-text alignment. See https://imagen.research.google/ for an overview of the results.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-02", "2023-03", "2023-04", "2023-05", "2023-07"], "last_mentioned": "2023-07-26T20:24:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2206.10257", "slug": "1e4e265bade7b8be", "title": "Satoshi Nakamoto and the Origins of Bitcoin -- The Profile of a 1-in-a-Billion Genius", "authors": ["Jens Ducrée"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2206.10257", "doi": null, "topics": ["Security & cryptography", "Economics"], "description": "The mystery about the ingenious creator of Bitcoin concealing behind the pseudonym Satoshi Nakamoto has been fascinating the global public for more than a decade.", "description_method": "abstract-excerpt", "abstract": "The mystery about the ingenious creator of Bitcoin concealing behind the pseudonym Satoshi Nakamoto has been fascinating the global public for more than a decade. Suddenly jumping out of the dark in 2008, this persona hurled the decentralized electronic cash system \"Bitcoin\", which has reached a peak market capitalization in the region of 1 trillion USD. In a purposely agnostic, and meticulous \"lea-ving no stone unturned\" approach, this study presents new hard facts, which evidently slipped through Satoshi Nakamoto's elaborate privacy shield, and derives meaningful pointers that are primarily inferred from Bitcoin's whitepaper, its blockchain parameters, and data that were widely up to his discretion. This ample stack of established and novel evidence is systematically categorized, analyzed, and then connected to its related, real-world ambient, like relevant locations and happenings in the past, and at the time. Evidence compounds towards a substantial role of the Benelux cryptography ecosystem, with strong transatlantic links, in the creation of Bitcoin. A consistent biography, a psychogram, and gripping story of an ingenious, multi-talented, autodidactic, reticent, and capricious polymath transpire, which are absolutely unique from a history of science and technology perspective. A cohort of previously fielded and best matches emerging from the investigations are probed against an unprecedently restrictive, multi-stage exclusion filter, which can, with maximum certainty, rule out most \"Satoshi Nakamoto\" candidates, while some of them remain to be confirmed. With this article, you will be able to decide who is not, or highly unlikely to be Satoshi Nakamoto, be equipped with an ample stack of systematically categorized evidence and efficient methodologies to find suitable candidates, and can possibly unveil the real identity of the creator of Bitcoin - if you want.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2024-02", "2024-04", "2024-10", "2026-04"], "last_mentioned": "2026-04-09T01:22:39Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2206.14858", "slug": "bdccb0ae064f833f", "title": "Solving Quantitative Reasoning Problems with Language Models", "authors": ["Aitor Lewkowycz", "Anders Andreassen", "David Dohan", "Ethan Dyer", "Henryk Michalewski", "Vinay Ramasesh", "Ambrose Slone", "Cem Anil", "Imanol Schlag", "Theo Gutman-Solo", "Yuhuai Wu", "Behnam Neyshabur", "Guy Gur-Ari", "Vedant Misra"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2206.14858", "doi": null, "topics": ["Language models", "Physics", "Mathematics"], "description": "Language models have achieved remarkable performance on a wide range of tasks that require natural language understanding.", "description_method": "abstract-excerpt", "abstract": "Language models have achieved remarkable performance on a wide range of tasks that require natural language understanding. Nevertheless, state-of-the-art models have generally struggled with tasks that require quantitative reasoning, such as solving mathematics, science, and engineering problems at the college level. To help close this gap, we introduce Minerva, a large language model pretrained on general natural language data and further trained on technical content. The model achieves state-of-the-art performance on technical benchmarks without the use of external tools. We also evaluate our model on over two hundred undergraduate-level problems in physics, biology, chemistry, economics, and other sciences that require quantitative reasoning, and find that the model can correctly answer nearly a third of them.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2022-10", "2022-12", "2023-02", "2023-11", "2025-01"], "last_mentioned": "2025-01-28T17:53:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2207.06431", "slug": "f78a589f928f39d1", "title": "Suppressing quantum errors by scaling a surface code logical qubit", "authors": ["Rajeev Acharya", "Igor Aleiner", "Richard Allen", "Trond I. Andersen", "Markus Ansmann", "Frank Arute", "Kunal Arya", "Abraham Asfaw", "Juan Atalaya", "Ryan Babbush", "Dave Bacon", "Joseph C. Bardin", "Joao Basso", "Andreas Bengtsson", "Sergio Boixo", "Gina Bortoli", "Alexandre Bourassa", "Jenna Bovaird", "Leon Brill", "Michael Broughton", "Bob B. Buckley", "David A. Buell", "Tim Burger", "Brian Burkett", "Nicholas Bushnell", "Yu Chen", "Zijun Chen", "Ben Chiaro", "Josh Cogan", "Roberto Collins", "Paul Conner", "William Courtney", "Alexander L. Crook", "Ben Curtin", "Dripto M. Debroy", "Alexander Del Toro Barba", "Sean Demura", "Andrew Dunsworth", "Daniel Eppens", "Catherine Erickson", "Lara Faoro", "Edward Farhi", "Reza Fatemi", "Leslie Flores Burgos", "Ebrahim Forati", "Austin G. Fowler", "Brooks Foxen", "William Giang", "Craig Gidney", "Dar Gilboa", "Marissa Giustina", "Alejandro Grajales Dau", "Jonathan A. Gross", "Steve Habegger", "Michael C. Hamilton", "Matthew P. Harrigan", "Sean D. Harrington", "Oscar Higgott", "Jeremy Hilton", "Markus Hoffmann", "Sabrina Hong", "Trent Huang", "Ashley Huff", "William J. Huggins", "Lev B. Ioffe", "Sergei V. Isakov", "Justin Iveland", "Evan Jeffrey", "Zhang Jiang", "Cody Jones", "Pavol Juhas", "Dvir Kafri", "Kostyantyn Kechedzhi", "Julian Kelly", "Tanuj Khattar", "Mostafa Khezri", "Mária Kieferová", "Seon Kim", "Alexei Kitaev", "Paul V. Klimov", "Andrey R. Klots", "Alexander N. Korotkov", "Fedor Kostritsa", "John Mark Kreikebaum", "David Landhuis", "Pavel Laptev", "Kim-Ming Lau", "Lily Laws", "Joonho Lee", "Kenny Lee", "Brian J. Lester", "Alexander Lill", "Wayne Liu", "Aditya Locharla", "Erik Lucero", "Fionn D. Malone", "Jeffrey Marshall", "Orion Martin", "Jarrod R. McClean", "Trevor Mccourt", "Matt McEwen", "Anthony Megrant", "Bernardo Meurer Costa", "Xiao Mi", "Kevin C. Miao", "Masoud Mohseni", "Shirin Montazeri", "Alexis Morvan", "Emily Mount", "Wojciech Mruczkiewicz", "Ofer Naaman", "Matthew Neeley", "Charles Neill", "Ani Nersisyan", "Hartmut Neven", "Michael Newman", "Jiun How Ng", "Anthony Nguyen", "Murray Nguyen", "Murphy Yuezhen Niu", "Thomas E. O'Brien", "Alex Opremcak", "John Platt", "Andre Petukhov", "Rebecca Potter", "Leonid P. Pryadko", "Chris Quintana", "Pedram Roushan", "Nicholas C. Rubin", "Negar Saei", "Daniel Sank", "Kannan Sankaragomathi", "Kevin J. Satzinger", "Henry F. Schurkus", "Christopher Schuster", "Michael J. Shearn", "Aaron Shorter", "Vladimir Shvarts", "Jindra Skruzny", "Vadim Smelyanskiy", "W. Clarke Smith", "George Sterling", "Doug Strain", "Marco Szalay", "Alfredo Torres", "Guifre Vidal", "Benjamin Villalonga", "Catherine Vollgraff Heidweiller", "Theodore White", "Cheng Xing", "Z. Jamie Yao", "Ping Yeh", "Juhwan Yoo", "Grayson Young", "Adam Zalcman", "Yaxing Zhang", "Ningfeng Zhu"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2207.06431", "doi": "10.1038/s41586-022-05434-1", "topics": ["Physics", "Statistics & probability", "Algorithms"], "description": "Practical quantum computing will require error rates that are well below what is achievable with physical qubits.", "description_method": "abstract-excerpt", "abstract": "Practical quantum computing will require error rates that are well below what is achievable with physical qubits. Quantum error correction offers a path to algorithmically-relevant error rates by encoding logical qubits within many physical qubits, where increasing the number of physical qubits enhances protection against physical errors. However, introducing more qubits also increases the number of error sources, so the density of errors must be sufficiently low in order for logical performance to improve with increasing code size. Here, we report the measurement of logical qubit performance scaling across multiple code sizes, and demonstrate that our system of superconducting qubits has sufficient performance to overcome the additional errors from increasing qubit number. We find our distance-5 surface code logical qubit modestly outperforms an ensemble of distance-3 logical qubits on average, both in terms of logical error probability over 25 cycles and logical error per cycle ($2.914\\%\\pm 0.016\\%$ compared to $3.028\\%\\pm 0.023\\%$). To investigate damaging, low-probability error sources, we run a distance-25 repetition code and observe a $1.7\\times10^{-6}$ logical error per round floor set by a single high-energy event ($1.6\\times10^{-7}$ when excluding this event). We are able to accurately model our experiment, and from this model we can extract error budgets that highlight the biggest challenges for future systems. These results mark the first experimental demonstration where quantum error correction begins to improve performance with increasing qubit number, illuminating the path to reaching the logical error rates required for computation.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2022-07", "2023-02", "2023-12", "2026-04"], "last_mentioned": "2026-04-01T16:31:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2209.01188", "slug": "580c01117b25c03e", "title": "Petals: Collaborative Inference and Fine-tuning of Large Models", "authors": ["Alexander Borzunov", "Dmitry Baranchuk", "Tim Dettmers", "Max Ryabinin", "Younes Belkada", "Artem Chumachenko", "Pavel Samygin", "Colin Raffel"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2209.01188", "doi": null, "topics": ["Economics", "Language models"], "description": "Many NLP tasks benefit from using large language models (LLMs) that often have more than 100 billion parameters.", "description_method": "abstract-excerpt", "abstract": "Many NLP tasks benefit from using large language models (LLMs) that often have more than 100 billion parameters. With the release of BLOOM-176B and OPT-175B, everyone can download pretrained models of this scale. Still, using these models requires high-end hardware unavailable to many researchers. In some cases, LLMs can be used more affordably via RAM offloading or hosted APIs. However, these techniques have innate limitations: offloading is too slow for interactive inference, while APIs are not flexible enough for research that requires access to weights, attention or logits. In this work, we propose Petals - a system for inference and fine-tuning of large models collaboratively by joining the resources of multiple parties. We demonstrate that this strategy outperforms offloading for very large models, running inference of BLOOM-176B on consumer GPUs with $\\approx$ 1 step per second, which is enough for many interactive LLM applications. Unlike most inference APIs, Petals also natively exposes hidden states of served models, allowing to train and share custom model extensions based on efficient fine-tuning methods.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2023-01", "2023-02", "2023-03"], "last_mentioned": "2023-03-06T08:16:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2209.15162", "slug": "3848e901a2eb5a8e", "title": "Linearly Mapping from Image to Text Space", "authors": ["Jack Merullo", "Louis Castricato", "Carsten Eickhoff", "Ellie Pavlick"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2209.15162", "doi": null, "topics": ["Language models", "Statistics & probability"], "description": "The extent to which text-only language models (LMs) learn to represent features of the non-linguistic world is an open question.", "description_method": "abstract-excerpt", "abstract": "The extent to which text-only language models (LMs) learn to represent features of the non-linguistic world is an open question. Prior work has shown that pretrained LMs can be taught to caption images when a vision model's parameters are optimized to encode images in the language space. We test a stronger hypothesis: that the conceptual representations learned by frozen text-only models and vision-only models are similar enough that this can be achieved with a linear map. We show that the image representations from vision models can be transferred as continuous prompts to frozen LMs by training only a single linear projection. Using these to prompt the LM achieves competitive performance on captioning and visual question answering tasks compared to models that tune both the image encoder and text decoder (such as the MAGMA model). We compare three image encoders with increasing amounts of linguistic supervision seen during pretraining: BEIT (no linguistic information), NF-ResNET (lexical category information), and CLIP (full natural language descriptions). We find that all three encoders perform equally well at transferring visual property information to the language model (e.g., whether an animal is large or small), but that image encoders pretrained with linguistic supervision more saliently encode category information (e.g., distinguishing hippo vs. elephant) and thus perform significantly better on benchmark language-and-vision tasks. Our results indicate that LMs encode conceptual information structurally similarly to vision-based models, even those that are solely trained on images. Code is available here: https://github.com/jmerullo/limber", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2022-10", "2023-04", "2023-06", "2023-12"], "last_mentioned": "2023-12-07T10:27:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2211.03622", "slug": "891ac8dcdf559ad3", "title": "Do Users Write More Insecure Code with AI Assistants?", "authors": ["Neil Perry", "Megha Srivastava", "Deepak Kumar", "Dan Boneh"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2211.03622", "doi": "10.1145/3576915.3623157", "topics": ["Security & cryptography", "Psychology", "Programming languages"], "description": "We conduct the first large-scale user study examining how users interact with an AI Code assistant to solve a variety of security related tasks across different programming languages.", "description_method": "abstract-excerpt", "abstract": "We conduct the first large-scale user study examining how users interact with an AI Code assistant to solve a variety of security related tasks across different programming languages. Overall, we find that participants who had access to an AI assistant based on OpenAI's codex-davinci-002 model wrote significantly less secure code than those without access. Additionally, participants with access to an AI assistant were more likely to believe they wrote secure code than those without access to the AI assistant. Furthermore, we find that participants who trusted the AI less and engaged more with the language and format of their prompts (e.g. re-phrasing, adjusting temperature) provided code with fewer security vulnerabilities. Finally, in order to better inform the design of future AI-based Code assistants, we provide an in-depth analysis of participants' language and interaction behavior, as well as release our user interface as an instrument to conduct similar studies in the future.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2023-02", "2023-03", "2025-09", "2026-02", "2026-05"], "last_mentioned": "2026-05-18T23:04:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.10562", "slug": "31a1b4bf5e460e88", "title": "Character-Aware Models Improve Visual Text Rendering", "authors": ["Rosanne Liu", "Dan Garrette", "Chitwan Saharia", "William Chan", "Adam Roberts", "Sharan Narang", "Irina Blok", "RJ Mical", "Mohammad Norouzi", "Noah Constant"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.10562", "doi": null, "topics": ["Computer vision"], "description": "Current image generation models struggle to reliably produce well-formed visual text.", "description_method": "abstract-excerpt", "abstract": "Current image generation models struggle to reliably produce well-formed visual text. In this paper, we investigate a key contributing factor: popular text-to-image models lack character-level input features, making it much harder to predict a word's visual makeup as a series of glyphs. To quantify this effect, we conduct a series of experiments comparing character-aware vs. character-blind text encoders. In the text-only domain, we find that character-aware models provide large gains on a novel spelling task (WikiSpell). Applying our learnings to the visual domain, we train a suite of image generation models, and show that character-aware variants outperform their character-blind counterparts across a range of novel text rendering tasks (our DrawText benchmark). Our models set a much higher state-of-the-art on visual spelling, with 30+ point accuracy gains over competitors on rare words, despite training on far fewer examples.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2023-02", "2023-04", "2024-03"], "last_mentioned": "2024-03-17T18:51:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2304.06835", "slug": "17b918fc192ce66b", "title": "Automated Translation and Accelerated Solving of Differential Equations on Multiple GPU Platforms", "authors": ["Utkarsh Utkarsh", "Valentin Churavy", "Yingbo Ma", "Tim Besard", "Prakitr Srisuma", "Tim Gymnich", "Adam R. Gerlach", "Alan Edelman", "George Barbastathis", "Richard D. Braatz", "Christopher Rackauckas"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2304.06835", "doi": "10.1016/j.cma.2023.116591", "topics": ["Statistics & probability", "Distributed systems", "Computer systems"], "description": "We demonstrate a high-performance vendor-agnostic method for massively parallel solving of ensembles of ordinary differential equations (ODEs) and stochastic differential equations (SDEs) on GPUs.", "description_method": "abstract-excerpt", "abstract": "We demonstrate a high-performance vendor-agnostic method for massively parallel solving of ensembles of ordinary differential equations (ODEs) and stochastic differential equations (SDEs) on GPUs. The method is integrated with a widely used differential equation solver library in a high-level language (Julia's DifferentialEquations.jl) and enables GPU acceleration without requiring code changes by the user. Our approach achieves state-of-the-art performance compared to hand-optimized CUDA-C++ kernels while performing 20--100$\\times$ faster than the vectorizing map (vmap) approach implemented in JAX and PyTorch. Performance evaluation on NVIDIA, AMD, Intel, and Apple GPUs demonstrates performance portability and vendor-agnosticism. We show composability with MPI to enable distributed multi-GPU workflows. The implemented solvers are fully featured -- supporting event handling, automatic differentiation, and incorporation of datasets via the GPU's texture memory -- allowing scientists to take advantage of GPU acceleration on all major current architectures without changing their model code and without loss of performance. We distribute the software as an open-source library https://github.com/SciML/DiffEqGPU.jl", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2023-05", "2023-12", "2024-03", "2024-08"], "last_mentioned": "2024-08-30T14:17:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.07185", "slug": "06a390ec2b9a9bce", "title": "MEGABYTE: Predicting Million-byte Sequences with Multiscale Transformers", "authors": ["Lili Yu", "Dániel Simig", "Colin Flaherty", "Armen Aghajanyan", "Luke Zettlemoyer", "Mike Lewis"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.07185", "doi": null, "topics": ["Language models"], "description": "Autoregressive transformers are spectacular models for short sequences but scale poorly to long sequences such as high-resolution images, podcasts, code, or books.", "description_method": "abstract-excerpt", "abstract": "Autoregressive transformers are spectacular models for short sequences but scale poorly to long sequences such as high-resolution images, podcasts, code, or books. We proposed Megabyte, a multi-scale decoder architecture that enables end-to-end differentiable modeling of sequences of over one million bytes. Megabyte segments sequences into patches and uses a local submodel within patches and a global model between patches. This enables sub-quadratic self-attention, much larger feedforward layers for the same compute, and improved parallelism during decoding -- unlocking better performance at reduced cost for both training and generation. Extensive experiments show that Megabyte allows byte-level models to perform competitively with subword models on long context language modeling, achieve state-of-the-art density estimation on ImageNet, and model audio from raw files. Together, these results establish the viability of tokenization-free autoregressive sequence modeling at scale.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-05", "2023-06", "2024-02", "2025-10"], "last_mentioned": "2025-10-22T07:42:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.11169", "slug": "e97959e92d8f826f", "title": "Emergent Representations of Program Semantics in Language Models Trained on Programs", "authors": ["Charles Jin", "Martin Rinard"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.11169", "doi": null, "topics": ["Language models", "Statistics & probability", "Psychology"], "description": "We present evidence that language models (LMs) of code can learn to represent the formal semantics of programs, despite being trained only to perform next-token prediction.", "description_method": "abstract-excerpt", "abstract": "We present evidence that language models (LMs) of code can learn to represent the formal semantics of programs, despite being trained only to perform next-token prediction. Specifically, we train a Transformer model on a synthetic corpus of programs written in a domain-specific language for navigating 2D grid world environments. Each program in the corpus is preceded by a (partial) specification in the form of several input-output grid world states. Despite providing no further inductive biases, we find that a probing classifier is able to extract increasingly accurate representations of the unobserved, intermediate grid world states from the LM hidden states over the course of training, suggesting the LM acquires an emergent ability to interpret programs in the formal sense. We also develop a novel interventional baseline that enables us to disambiguate what is represented by the LM as opposed to learned by the probe. We anticipate that this technique may be generally applicable to a broad range of semantic probing experiments. In summary, this paper does not propose any new techniques for training LMs of code, but develops an experimental framework for and provides insights into the acquisition and representation of formal semantics in statistical models of code. Our code is available at https://github.com/charlesjin/emergent-semantics.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2023-05", "2023-06", "2024-08", "2026-08"], "last_mentioned": "2026-08-19T22:05:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.13048", "slug": "9e5faaeb47dfc7d3", "title": "RWKV: Reinventing RNNs for the Transformer Era", "authors": ["Bo Peng", "Eric Alcaide", "Quentin Anthony", "Alon Albalak", "Samuel Arcadinho", "Stella Biderman", "Huanqi Cao", "Xin Cheng", "Michael Chung", "Matteo Grella", "Kranthi Kiran GV", "Xuzheng He", "Haowen Hou", "Jiaju Lin", "Przemyslaw Kazienko", "Jan Kocon", "Jiaming Kong", "Bartlomiej Koptyra", "Hayden Lau", "Krishna Sri Ipsit Mantri", "Ferdinand Mom", "Atsushi Saito", "Guangyu Song", "Xiangru Tang", "Bolun Wang", "Johan S. Wind", "Stanislaw Wozniak", "Ruichong Zhang", "Zhenyuan Zhang", "Qihang Zhao", "Peng Zhou", "Qinghua Zhou", "Jian Zhu", "Rui-Jie Zhu"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.13048", "doi": null, "topics": ["Language models", "Algorithms", "Machine learning"], "description": "Transformers have revolutionized almost all natural language processing (NLP) tasks but suffer from memory and computational complexity that scales quadratically with sequence length.", "description_method": "abstract-excerpt", "abstract": "Transformers have revolutionized almost all natural language processing (NLP) tasks but suffer from memory and computational complexity that scales quadratically with sequence length. In contrast, recurrent neural networks (RNNs) exhibit linear scaling in memory and computational requirements but struggle to match the same performance as Transformers due to limitations in parallelization and scalability. We propose a novel model architecture, Receptance Weighted Key Value (RWKV), that combines the efficient parallelizable training of transformers with the efficient inference of RNNs. Our approach leverages a linear attention mechanism and allows us to formulate the model as either a Transformer or an RNN, thus parallelizing computations during training and maintains constant computational and memory complexity during inference. We scale our models as large as 14 billion parameters, by far the largest dense RNN ever trained, and find RWKV performs on par with similarly sized Transformers, suggesting future work can leverage this architecture to create more efficient models. This work presents a significant step towards reconciling trade-offs between computational efficiency and model performance in sequence processing tasks.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-07", "2023-11", "2024-01", "2026-02"], "last_mentioned": "2026-02-04T00:28:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.15717", "slug": "b6b8d171897aa6a2", "title": "The False Promise of Imitating Proprietary LLMs", "authors": ["Arnav Gudibande", "Eric Wallace", "Charlie Snell", "Xinyang Geng", "Hao Liu", "Pieter Abbeel", "Sergey Levine", "Dawn Song"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.15717", "doi": null, "topics": ["Language models"], "description": "An emerging method to cheaply improve a weaker language model is to finetune it on outputs from a stronger model, such as a proprietary system like ChatGPT (e.g., Alpaca, Self-Instruct, and others).", "description_method": "abstract-excerpt", "abstract": "An emerging method to cheaply improve a weaker language model is to finetune it on outputs from a stronger model, such as a proprietary system like ChatGPT (e.g., Alpaca, Self-Instruct, and others). This approach looks to cheaply imitate the proprietary model's capabilities using a weaker open-source model. In this work, we critically analyze this approach. We first finetune a series of LMs that imitate ChatGPT using varying base model sizes (1.5B--13B), data sources, and imitation data amounts (0.3M--150M tokens). We then evaluate the models using crowd raters and canonical NLP benchmarks. Initially, we were surprised by the output quality of our imitation models -- they appear far better at following instructions, and crowd workers rate their outputs as competitive with ChatGPT. However, when conducting more targeted automatic evaluations, we find that imitation models close little to none of the gap from the base LM to ChatGPT on tasks that are not heavily supported in the imitation data. We show that these performance discrepancies may slip past human raters because imitation models are adept at mimicking ChatGPT's style but not its factuality. Overall, we conclude that model imitation is a false promise: there exists a substantial capabilities gap between open and closed LMs that, with current methods, can only be bridged using an unwieldy amount of imitation data or by using more capable base LMs. In turn, we argue that the highest leverage action for improving open-source models is to tackle the difficult challenge of developing better base LMs, rather than taking the shortcut of imitating proprietary systems.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-05", "2023-06", "2026-06"], "last_mentioned": "2026-06-25T01:21:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.16264", "slug": "4deab4bd6f90bad7", "title": "Scaling Data-Constrained Language Models", "authors": ["Niklas Muennighoff", "Alexander M. Rush", "Boaz Barak", "Teven Le Scao", "Aleksandra Piktus", "Nouamane Tazi", "Sampo Pyysalo", "Thomas Wolf", "Colin Raffel"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.16264", "doi": null, "topics": ["Language models", "Networks"], "description": "The current trend of scaling language models involves increasing both parameter count and training dataset size.", "description_method": "abstract-excerpt", "abstract": "The current trend of scaling language models involves increasing both parameter count and training dataset size. Extrapolating this trend suggests that training dataset size may soon be limited by the amount of text data available on the internet. Motivated by this limit, we investigate scaling language models in data-constrained regimes. Specifically, we run a large set of experiments varying the extent of data repetition and compute budget, ranging up to 900 billion training tokens and 9 billion parameter models. We find that with constrained data for a fixed compute budget, training with up to 4 epochs of repeated data yields negligible changes to loss compared to having unique data. However, with more repetition, the value of adding compute eventually decays to zero. We propose and empirically validate a scaling law for compute optimality that accounts for the decreasing value of repeated tokens and excess parameters. Finally, we experiment with approaches mitigating data scarcity, including augmenting the training dataset with code data or removing commonly used filters. Models and datasets from our 400 training runs are freely available at https://github.com/huggingface/datablations.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-09", "2024-08", "2025-01"], "last_mentioned": "2025-01-20T13:51:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.10072", "slug": "d93e19b898b0ce65", "title": "Shor's Algorithm Does Not Factor Large Integers in the Presence of Noise", "authors": ["Jin-Yi Cai"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.10072", "doi": "10.1007/s11432-023-3961-3", "topics": ["Algorithms", "Physics", "Statistics & probability"], "description": "We consider Shor's quantum factoring algorithm in the setting of noisy quantum gates.", "description_method": "abstract-excerpt", "abstract": "We consider Shor's quantum factoring algorithm in the setting of noisy quantum gates. Under a generic model of random noise for (controlled) rotation gates, we prove that the algorithm does not factor integers of the form $pq$ when the noise exceeds a vanishingly small level in terms of $n$ -- the number of bits of the integer to be factored, where $p$ and $q$ are from a well-defined set of primes of positive density. We further prove that with probability $1 - o(1)$ over random prime pairs $(p,q)$, Shor's factoring algorithm does not factor numbers of the form $pq$, with the same level of random noise present.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2024-04", "2024-05", "2024-12", "2025-07"], "last_mentioned": "2025-07-15T10:37:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.15595", "slug": "ab51cf6c0bf0c998", "title": "Extending Context Window of Large Language Models via Positional Interpolation", "authors": ["Shouyuan Chen", "Sherman Wong", "Liangjian Chen", "Yuandong Tian"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.15595", "doi": null, "topics": ["Language models", "Algorithms"], "description": "We present Position Interpolation (PI) that extends the context window sizes of RoPE-based pretrained LLMs such as LLaMA models to up to 32768 with minimal fine-tuning (within 1000 steps), while demonstrating strong empirical results on various tasks that require long context, including passkey retrieval, language modeling, and long docum", "description_method": "abstract-excerpt", "abstract": "We present Position Interpolation (PI) that extends the context window sizes of RoPE-based pretrained LLMs such as LLaMA models to up to 32768 with minimal fine-tuning (within 1000 steps), while demonstrating strong empirical results on various tasks that require long context, including passkey retrieval, language modeling, and long document summarization from LLaMA 7B to 65B. Meanwhile, the extended model by Position Interpolation preserve quality relatively well on tasks within its original context window. To achieve this goal, Position Interpolation linearly down-scales the input position indices to match the original context window size, rather than extrapolating beyond the trained context length which may lead to catastrophically high attention scores that completely ruin the self-attention mechanism. Our theoretical study shows that the upper bound of interpolation is at least $\\sim 600 \\times$ smaller than that of extrapolation, further demonstrating its stability. Models extended via Position Interpolation retain its original architecture and can reuse most pre-existing optimization and infrastructure.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-06", "2023-07", "2023-11", "2024-01"], "last_mentioned": "2024-01-12T18:43:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2307.08621", "slug": "5c1a9189498eb840", "title": "Retentive Network: A Successor to Transformer for Large Language Models", "authors": ["Yutao Sun", "Li Dong", "Shaohan Huang", "Shuming Ma", "Yuqing Xia", "Jilong Xue", "Jianyong Wang", "Furu Wei"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2307.08621", "doi": null, "topics": ["Language models", "Algorithms"], "description": "In this work, we propose Retentive Network (RetNet) as a foundation architecture for large language models, simultaneously achieving training parallelism, low-cost inference, and good performance.", "description_method": "abstract-excerpt", "abstract": "In this work, we propose Retentive Network (RetNet) as a foundation architecture for large language models, simultaneously achieving training parallelism, low-cost inference, and good performance. We theoretically derive the connection between recurrence and attention. Then we propose the retention mechanism for sequence modeling, which supports three computation paradigms, i.e., parallel, recurrent, and chunkwise recurrent. Specifically, the parallel representation allows for training parallelism. The recurrent representation enables low-cost $O(1)$ inference, which improves decoding throughput, latency, and GPU memory without sacrificing performance. The chunkwise recurrent representation facilitates efficient long-sequence modeling with linear complexity, where each chunk is encoded parallelly while recurrently summarizing the chunks. Experimental results on language modeling show that RetNet achieves favorable scaling results, parallel training, low-cost deployment, and efficient inference. The intriguing properties make RetNet a strong successor to Transformer for large language models. Code will be available at https://aka.ms/retnet.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-09", "2023-12", "2024-01", "2025-02"], "last_mentioned": "2025-02-15T03:04:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2308.01074", "slug": "24e13786c4b56d26", "title": "A Practical Deep Learning-Based Acoustic Side Channel Attack on Keyboards", "authors": ["Joshua Harrison", "Ehsan Toreini", "Maryam Mehrnezhad"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2308.01074", "doi": "10.1109/EuroSPW59978.2023.00034", "topics": ["Security & cryptography", "Machine learning", "Language models"], "description": "With recent developments in deep learning, the ubiquity of micro-phones and the rise in online services via personal devices, acoustic side channel attacks present a greater threat to keyboards than ever.", "description_method": "abstract-excerpt", "abstract": "With recent developments in deep learning, the ubiquity of micro-phones and the rise in online services via personal devices, acoustic side channel attacks present a greater threat to keyboards than ever. This paper presents a practical implementation of a state-of-the-art deep learning model in order to classify laptop keystrokes, using a smartphone integrated microphone. When trained on keystrokes recorded by a nearby phone, the classifier achieved an accuracy of 95%, the highest accuracy seen without the use of a language model. When trained on keystrokes recorded using the video-conferencing software Zoom, an accuracy of 93% was achieved, a new best for the medium. Our results prove the practicality of these side channel attacks via off-the-shelf equipment and algorithms. We discuss a series of mitigation methods to protect users against these series of attacks.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-08", "2023-10", "2023-12", "2024-11", "2026-03"], "last_mentioned": "2026-03-21T17:51:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.04153", "slug": "7ebc1c079a8bdc4f", "title": "Fair coins tend to land on the same side they started: Evidence from 350,757 flips", "authors": ["František Bartoš", "Alexandra Sarafoglou", "Henrik R. Godmann", "Amir Sahrani", "David Klein Leunk", "Pierre Y. Gui", "David Voss", "Kaleem Ullah", "Malte J. Zoubek", "Franziska Nippold", "Frederik Aust", "Felipe F. Vieira", "Chris-Gabriel Islam", "Anton J. Zoubek", "Sara Shabani", "Jonas Petter", "Ingeborg B. Roos", "Adam Finnemann", "Aaron B. Lob", "Madlen F. Hoffstadt", "Jason Nak", "Jill de Ron", "Koen Derks", "Karoline Huth", "Sjoerd Terpstra", "Thomas Bastelica", "Magda Matetovici", "Vincent L. Ott", "Andreea S. Zetea", "Katharina Karnbach", "Michelle C. Donzallaz", "Arne John", "Roy M. Moore", "Franziska Assion", "Riet van Bork", "Theresa E. Leidinger", "Xiaochang Zhao", "Adrian Karami Motaghi", "Ting Pan", "Hannah Armstrong", "Tianqi Peng", "Mara Bialas", "Joyce Y. -C. Pang", "Bohan Fu", "Shujun Yang", "Xiaoyi Lin", "Dana Sleiffer", "Miklos Bognar", "Balazs Aczel", "Eric-Jan Wagenmakers"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.04153", "doi": "10.1080/01621459.2025.2516210", "topics": ["Psychology", "Statistics & probability", "Physics"], "description": "Many people have flipped coins but few have stopped to ponder the statistical and physical intricacies of the process.", "description_method": "abstract-excerpt", "abstract": "Many people have flipped coins but few have stopped to ponder the statistical and physical intricacies of the process. We collected $350{,}757$ coin flips to test the counterintuitive prediction from a physics model of human coin tossing developed by Diaconis, Holmes, and Montgomery (DHM; 2007). The model asserts that when people flip an ordinary coin, it tends to land on the same side it started -- DHM estimated the probability of a same-side outcome to be about 51\\%. Our data lend strong support to this precise prediction: the coins landed on the same side more often than not, $\\text{Pr}(\\text{same side}) = 0.508$, 95\\% credible interval (CI) [$0.506$, $0.509$], $\\text{BF}_{\\text{same-side bias}} = 2359$. Furthermore, the data revealed considerable between-people variation in the degree of this same-side bias. Our data also confirmed the generic prediction that when people flip an ordinary coin -- with the initial side-up randomly determined -- it is equally likely to land heads or tails: $\\text{Pr}(\\text{heads}) = 0.500$, 95\\% CI [$0.498$, $0.502$], $\\text{BF}_{\\text{heads-tails bias}} = 0.182$. Furthermore, this lack of heads-tails bias does not appear to vary across coins. Additional analyses revealed that the within-people same-side bias decreased as more coins were flipped, an effect that is consistent with the possibility that practice makes people flip coins in a less wobbly fashion. Our data therefore provide strong evidence that when some (but not all) people flip a fair coin, it tends to land on the same side it started.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-10", "2023-12", "2024-01"], "last_mentioned": "2024-01-31T19:41:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.09423", "slug": "ab7ba26d715563fd", "title": "QUIC is not Quick Enough over Fast Internet", "authors": ["Xumiao Zhang", "Shuowei Jin", "Yi He", "Ahmad Hassan", "Z. Morley Mao", "Feng Qian", "Zhi-Li Zhang"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.09423", "doi": "10.1145/3589334.3645323", "topics": ["Networks", "Computer systems", "Biology & medicine"], "description": "QUIC is expected to be a game-changer in improving web application performance.", "description_method": "abstract-excerpt", "abstract": "QUIC is expected to be a game-changer in improving web application performance. In this paper, we conduct a systematic examination of QUIC's performance over high-speed networks. We find that over fast Internet, the UDP+QUIC+HTTP/3 stack suffers a data rate reduction of up to 45.2% compared to the TCP+TLS+HTTP/2 counterpart. Moreover, the performance gap between QUIC and HTTP/2 grows as the underlying bandwidth increases. We observe this issue on lightweight data transfer clients and major web browsers (Chrome, Edge, Firefox, Opera), on different hosts (desktop, mobile), and over diverse networks (wired broadband, cellular). It affects not only file transfers, but also various applications such as video streaming (up to 9.8% video bitrate reduction) and web browsing. Through rigorous packet trace analysis and kernel- and user-space profiling, we identify the root cause to be high receiver-side processing overhead, in particular, excessive data packets and QUIC's user-space ACKs. We make concrete recommendations for mitigating the observed performance issues.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2024-02", "2024-09", "2025-03", "2025-08"], "last_mentioned": "2025-08-22T22:50:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2311.17035", "slug": "7a7dcd59b06c0d57", "title": "Scalable Extraction of Training Data from (Production) Language Models", "authors": ["Milad Nasr", "Nicholas Carlini", "Jonathan Hayase", "Matthew Jagielski", "A. Feder Cooper", "Daphne Ippolito", "Christopher A. Choquette-Choo", "Eric Wallace", "Florian Tramèr", "Katherine Lee"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2311.17035", "doi": null, "topics": ["Security & cryptography", "Language models", "Machine learning"], "description": "This paper studies extractable memorization: training data that an adversary can efficiently extract by querying a machine learning model without prior knowledge of the training dataset.", "description_method": "abstract-excerpt", "abstract": "This paper studies extractable memorization: training data that an adversary can efficiently extract by querying a machine learning model without prior knowledge of the training dataset. We show an adversary can extract gigabytes of training data from open-source language models like Pythia or GPT-Neo, semi-open models like LLaMA or Falcon, and closed models like ChatGPT. Existing techniques from the literature suffice to attack unaligned models; in order to attack the aligned ChatGPT, we develop a new divergence attack that causes the model to diverge from its chatbot-style generations and emit training data at a rate 150x higher than when behaving properly. Our methods show practical attacks can recover far more data than previously thought, and reveal that current alignment techniques do not eliminate memorization.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-12", "2024-01", "2024-02", "2025-10", "2025-11"], "last_mentioned": "2025-11-28T03:18:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.03300", "slug": "636fb871904bc06f", "title": "DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models", "authors": ["Zhihong Shao", "Peiyi Wang", "Qihao Zhu", "Runxin Xu", "Junxiao Song", "Xiao Bi", "Haowei Zhang", "Mingchuan Zhang", "Y. K. Li", "Y. Wu", "Daya Guo"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.03300", "doi": null, "topics": ["Mathematics", "Language models", "Algorithms"], "description": "Mathematical reasoning poses a significant challenge for language models due to its complex and structured nature.", "description_method": "abstract-excerpt", "abstract": "Mathematical reasoning poses a significant challenge for language models due to its complex and structured nature. In this paper, we introduce DeepSeekMath 7B, which continues pre-training DeepSeek-Coder-Base-v1.5 7B with 120B math-related tokens sourced from Common Crawl, together with natural language and code data. DeepSeekMath 7B has achieved an impressive score of 51.7% on the competition-level MATH benchmark without relying on external toolkits and voting techniques, approaching the performance level of Gemini-Ultra and GPT-4. Self-consistency over 64 samples from DeepSeekMath 7B achieves 60.9% on MATH. The mathematical reasoning capability of DeepSeekMath is attributed to two key factors: First, we harness the significant potential of publicly available web data through a meticulously engineered data selection pipeline. Second, we introduce Group Relative Policy Optimization (GRPO), a variant of Proximal Policy Optimization (PPO), that enhances mathematical reasoning abilities while concurrently optimizing the memory usage of PPO.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2025-01", "2025-02", "2025-05", "2025-11", "2026-01"], "last_mentioned": "2026-01-21T06:03:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.14531", "slug": "b7b510147318f959", "title": "Should We Respect LLMs? A Cross-Lingual Study on the Influence of Prompt Politeness on LLM Performance", "authors": ["Ziqi Yin", "Hao Wang", "Kaito Horio", "Daisuke Kawahara", "Satoshi Sekine"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.14531", "doi": null, "topics": ["Language models", "Psychology"], "description": "We investigate the impact of politeness levels in prompts on the performance of large language models (LLMs).", "description_method": "abstract-excerpt", "abstract": "We investigate the impact of politeness levels in prompts on the performance of large language models (LLMs). Polite language in human communications often garners more compliance and effectiveness, while rudeness can cause aversion, impacting response quality. We consider that LLMs mirror human communication traits, suggesting they align with human cultural norms. We assess the impact of politeness in prompts on LLMs across English, Chinese, and Japanese tasks. We observed that impolite prompts often result in poor performance, but overly polite language does not guarantee better outcomes. The best politeness level is different according to the language. This phenomenon suggests that LLMs not only reflect human behavior but are also influenced by language, particularly in different cultural contexts. Our findings highlight the need to factor in politeness for cross-cultural natural language processing and LLM usage.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2024-03", "2025-09", "2026-02", "2026-04"], "last_mentioned": "2026-04-05T13:33:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2406.17642", "slug": "8b2e47d3ce9a7515", "title": "Banishing LLM Hallucinations Requires Rethinking Generalization", "authors": ["Johnny Li", "Saksham Consul", "Eda Zhou", "James Wong", "Naila Farooqui", "Yuxin Ye", "Nithyashree Manohar", "Zhuxiaona Wei", "Tian Wu", "Ben Echols", "Sharon Zhou", "Gregory Diamos"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2406.17642", "doi": null, "topics": ["Language models", "Machine learning", "Networks"], "description": "Despite their powerful chat, coding, and reasoning abilities, Large Language Models (LLMs) frequently hallucinate.", "description_method": "abstract-excerpt", "abstract": "Despite their powerful chat, coding, and reasoning abilities, Large Language Models (LLMs) frequently hallucinate. Conventional wisdom suggests that hallucinations are a consequence of a balance between creativity and factuality, which can be mitigated, but not eliminated, by grounding the LLM in external knowledge sources. Through extensive systematic experiments, we show that these traditional approaches fail to explain why LLMs hallucinate in practice. Specifically, we show that LLMs augmented with a massive Mixture of Memory Experts (MoME) can easily memorize large datasets of random numbers. We corroborate these experimental findings with a theoretical construction showing that simple neural networks trained to predict the next token hallucinate when the training loss is above a threshold as it usually does in practice when training on internet scale data. We interpret our findings by comparing against traditional retrieval methods for mitigating hallucinations. We use our findings to design a first generation model for removing hallucinations -- Lamini-1 -- that stores facts in a massive mixture of millions of memory experts that are retrieved dynamically.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2024-08", "2024-09", "2024-11", "2025-01"], "last_mentioned": "2025-01-27T21:51:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2409.05746", "slug": "e3782c46fe3be5e3", "title": "LLMs Will Always Hallucinate, and We Need to Live With This", "authors": ["Sourav Banerjee", "Ayushi Agarwal", "Saloni Singla"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2409.05746", "doi": null, "topics": ["Mathematics", "Language models", "Statistics & probability"], "description": "As Large Language Models become more ubiquitous across domains, it becomes important to examine their inherent limitations critically.", "description_method": "abstract-excerpt", "abstract": "As Large Language Models become more ubiquitous across domains, it becomes important to examine their inherent limitations critically. This work argues that hallucinations in language models are not just occasional errors but an inevitable feature of these systems. We demonstrate that hallucinations stem from the fundamental mathematical and logical structure of LLMs. It is, therefore, impossible to eliminate them through architectural improvements, dataset enhancements, or fact-checking mechanisms. Our analysis draws on computational theory and Godel's First Incompleteness Theorem, which references the undecidability of problems like the Halting, Emptiness, and Acceptance Problems. We demonstrate that every stage of the LLM process-from training data compilation to fact retrieval, intent classification, and text generation-will have a non-zero probability of producing hallucinations. This work introduces the concept of Structural Hallucination as an intrinsic nature of these systems. By establishing the mathematical certainty of hallucinations, we challenge the prevailing notion that they can be fully mitigated.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2024-09", "2024-10", "2025-06"], "last_mentioned": "2025-06-22T18:52:39Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2410.00037", "slug": "5337192312556d3d", "title": "Moshi: a speech-text foundation model for real-time dialogue", "authors": ["Alexandre Défossez", "Laurent Mazaré", "Manu Orsini", "Amélie Royer", "Patrick Pérez", "Hervé Jégou", "Edouard Grave", "Neil Zeghidour"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2410.00037", "doi": null, "topics": ["Psychology", "Language models", "Computer vision"], "description": "We introduce Moshi, a speech-text foundation model and full-duplex spoken dialogue framework.", "description_method": "abstract-excerpt", "abstract": "We introduce Moshi, a speech-text foundation model and full-duplex spoken dialogue framework. Current systems for spoken dialogue rely on pipelines of independent components, namely voice activity detection, speech recognition, textual dialogue and text-to-speech. Such frameworks cannot emulate the experience of real conversations. First, their complexity induces a latency of several seconds between interactions. Second, text being the intermediate modality for dialogue, non-linguistic information that modifies meaning -- such as emotion or non-speech sounds -- is lost in the interaction. Finally, they rely on a segmentation into speaker turns, which does not take into account overlapping speech, interruptions and interjections. Moshi solves these independent issues altogether by casting spoken dialogue as speech-to-speech generation. Starting from a text language model backbone, Moshi generates speech as tokens from the residual quantizer of a neural audio codec, while modeling separately its own speech and that of the user into parallel streams. This allows for the removal of explicit speaker turns, and the modeling of arbitrary conversational dynamics. We moreover extend the hierarchical semantic-to-acoustic token generation of previous work to first predict time-aligned text tokens as a prefix to audio tokens. Not only this \"Inner Monologue\" method significantly improves the linguistic quality of generated speech, but we also illustrate how it can provide streaming speech recognition and text-to-speech. Our resulting model is the first real-time full-duplex spoken large language model, with a theoretical latency of 160ms, 200ms in practice, and is available at https://github.com/kyutai-labs/moshi.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2025-01", "2025-10", "2026-03", "2026-07"], "last_mentioned": "2026-07-08T18:04:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2412.05265", "slug": "194acd4ccc81a7f5", "title": "Reinforcement Learning: An Overview", "authors": ["Kevin Murphy"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2412.05265", "doi": null, "topics": ["Machine learning", "Psychology"], "description": "This manuscript gives a big-picture, up-to-date overview of the field of (deep) reinforcement learning and sequential decision making, covering value-based methods, policy-based methods, model-based methods, multi-agent RL, LLMs and RL, and various other topics (e.g., offline RL, hierarchical RL, intrinsic reward).", "description_method": "abstract-excerpt", "abstract": "This manuscript gives a big-picture, up-to-date overview of the field of (deep) reinforcement learning and sequential decision making, covering value-based methods, policy-based methods, model-based methods, multi-agent RL, LLMs and RL, and various other topics (e.g., offline RL, hierarchical RL, intrinsic reward). It also includes some code snippets for training LLMs with RL.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2025-02", "2025-03", "2025-11", "2026-06"], "last_mentioned": "2026-06-16T04:57:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2502.09992", "slug": "a4863ed6f639560b", "title": "Large Language Diffusion Models", "authors": ["Shen Nie", "Fengqi Zhu", "Zebin You", "Xiaolu Zhang", "Jingyang Ou", "Jun Hu", "Jun Zhou", "Yankai Lin", "Ji-Rong Wen", "Chongxuan Li"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2502.09992", "doi": null, "topics": ["Language models", "Physics", "Machine learning"], "description": "The capabilities of large language models (LLMs) are widely regarded as relying on autoregressive models (ARMs).", "description_method": "abstract-excerpt", "abstract": "The capabilities of large language models (LLMs) are widely regarded as relying on autoregressive models (ARMs). We challenge this notion by introducing LLaDA, a diffusion model trained from scratch under the pre-training and supervised fine-tuning (SFT) paradigm. LLaDA employs a forward data masking process and a reverse generation process, parameterized by a Transformer to predict masked tokens. It provides a principled generative approach for probabilistic inference by optimizing a likelihood lower bound. Across extensive benchmarks on general tasks, math, code, and so on, LLaDA demonstrates strong scalability and performs comparably to our self-constructed ARM baselines. Remarkably, LLaDA 8B is competitive with strong LLMs like LLaMA3 8B in in-context learning and, after SFT, exhibits impressive instruction-following abilities in case studies such as multi-turn dialogue. Moreover, LLaDA addresses the reversal curse, surpassing GPT-4o in a reversal poem completion task. Our findings show the promise of diffusion models for language modeling at scale and challenge the common assumption that core LLM capabilities discussed above inherently depend on ARMs. Project page and codes: https://ml-gsai.github.io/LLaDA-demo/.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2025-02", "2025-03", "2025-04"], "last_mentioned": "2025-04-05T16:16:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2504.17033", "slug": "79f3ed02eb586e77", "title": "Breaking the Sorting Barrier for Directed Single-Source Shortest Paths", "authors": ["Ran Duan", "Jiayi Mao", "Xiao Mao", "Xinkai Shu", "Longhui Yin"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2504.17033", "doi": null, "topics": ["Algorithms"], "description": "We give a deterministic $O(m\\log^{2/3}n)$-time algorithm for single-source shortest paths (SSSP) on directed graphs with real non-negative edge weights in the comparison-addition model.", "description_method": "abstract-excerpt", "abstract": "We give a deterministic $O(m\\log^{2/3}n)$-time algorithm for single-source shortest paths (SSSP) on directed graphs with real non-negative edge weights in the comparison-addition model. This is the first result to break the $O(m+n\\log n)$ time bound of Dijkstra's algorithm on sparse graphs, showing that Dijkstra's algorithm is not optimal for SSSP.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2025-08", "2025-10", "2026-02"], "last_mentioned": "2026-02-23T16:14:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2506.09501", "slug": "d570f4e8117213c8", "title": "Understanding and Mitigating Numerical Sources of Nondeterminism in LLM Inference", "authors": ["Jiayi Yuan", "Hao Li", "Xinheng Ding", "Wenya Xie", "Yu-Jhe Li", "Wentian Zhao", "Kun Wan", "Jing Shi", "Xia Hu", "Zirui Liu"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2506.09501", "doi": null, "topics": ["Science & research", "Language models"], "description": "Large Language Models (LLMs) are now integral across various domains and have demonstrated impressive performance.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) are now integral across various domains and have demonstrated impressive performance. Progress, however, rests on the premise that benchmark scores are both accurate and reproducible. We demonstrate that the reproducibility of LLM performance is fragile: changing system configuration, such as evaluation batch size, GPU count, and GPU version, can introduce significant differences in the generated responses. This issue is especially pronounced in reasoning models, where minor rounding differences in early tokens can cascade into divergent chains of thought, ultimately affecting accuracy. For instance, under bfloat16 precision with greedy decoding, a reasoning model like DeepSeek-R1-Distill-Qwen-7B can exhibit up to 9% variation in accuracy and 9,000 tokens difference in response length due to differences in GPU count, type, and evaluation batch size. We trace the root cause of this variability to the non-associative nature of floating-point arithmetic under limited numerical precision. This work presents the first systematic investigation into how numerical precision affects reproducibility in LLM inference. Through carefully controlled experiments across various hardware, software, and precision settings, we quantify when and how model outputs diverge. Our analysis reveals that floating-point precision - while critical for reproducibility - is often neglected in evaluation practices. Inspired by this, we develop a lightweight inference pipeline, dubbed LayerCast, that stores weights in 16-bit precision but performs all computations in FP32, balancing memory efficiency with numerical stability. Code is available at https://github.com/nanomaoli/llm_reproducibility.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2025-06", "2025-07", "2026-02", "2026-03"], "last_mentioned": "2026-03-14T04:26:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2506.13131", "slug": "cfc5bf7a6b82596d", "title": "AlphaEvolve: A coding agent for scientific and algorithmic discovery", "authors": ["Alexander Novikov", "Ngân Vũ", "Marvin Eisenberger", "Emilien Dupont", "Po-Sen Huang", "Adam Zsolt Wagner", "Sergey Shirobokov", "Borislav Kozlovskii", "Francisco J. R. Ruiz", "Abbas Mehrabian", "M. Pawan Kumar", "Abigail See", "Swarat Chaudhuri", "George Holland", "Alex Davies", "Sebastian Nowozin", "Pushmeet Kohli", "Matej Balog"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2506.13131", "doi": null, "topics": ["Algorithms", "Science & research", "Robotics"], "description": "In this white paper, we present AlphaEvolve, an evolutionary coding agent that substantially enhances capabilities of state-of-the-art LLMs on highly challenging tasks such as tackling open scientific problems or optimizing critical pieces of computational infrastructure.", "description_method": "abstract-excerpt", "abstract": "In this white paper, we present AlphaEvolve, an evolutionary coding agent that substantially enhances capabilities of state-of-the-art LLMs on highly challenging tasks such as tackling open scientific problems or optimizing critical pieces of computational infrastructure. AlphaEvolve orchestrates an autonomous pipeline of LLMs, whose task is to improve an algorithm by making direct changes to the code. Using an evolutionary approach, continuously receiving feedback from one or more evaluators, AlphaEvolve iteratively improves the algorithm, potentially leading to new scientific and practical discoveries. We demonstrate the broad applicability of this approach by applying it to a number of important computational problems. When applied to optimizing critical components of large-scale computational stacks at Google, AlphaEvolve developed a more efficient scheduling algorithm for data centers, found a functionally equivalent simplification in the circuit design of hardware accelerators, and accelerated the training of the LLM underpinning AlphaEvolve itself. Furthermore, AlphaEvolve discovered novel, provably correct algorithms that surpass state-of-the-art solutions on a spectrum of problems in mathematics and computer science, significantly expanding the scope of prior automated discovery methods (Romera-Paredes et al., 2023). Notably, AlphaEvolve developed a search algorithm that found a procedure to multiply two $4 \\times 4$ complex-valued matrices using $48$ scalar multiplications; offering the first improvement, after 56 years, over Strassen's algorithm in this setting. We believe AlphaEvolve and coding agents like it can have a significant impact in improving solutions of problems across many areas of science and computation.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2025-12", "2026-04", "2026-05", "2026-07"], "last_mentioned": "2026-07-01T13:48:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2509.04664", "slug": "86283b1402da0188", "title": "Why Language Models Hallucinate", "authors": ["Adam Tauman Kalai", "Ofir Nachum", "Santosh S. Vempala", "Edwin Zhang"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2509.04664", "doi": null, "topics": ["Language models", "Statistics & probability"], "description": "Like students facing hard exam questions, large language models sometimes guess when uncertain, producing plausible yet incorrect statements instead of admitting uncertainty.", "description_method": "abstract-excerpt", "abstract": "Like students facing hard exam questions, large language models sometimes guess when uncertain, producing plausible yet incorrect statements instead of admitting uncertainty. Such \"hallucinations\" persist even in state-of-the-art systems and undermine trust. We argue that language models hallucinate because the training and evaluation procedures reward guessing over acknowledging uncertainty, and we analyze the statistical causes of hallucinations in the modern training pipeline. Hallucinations need not be mysterious -- they originate simply as errors in binary classification. If incorrect statements cannot be distinguished from facts, then hallucinations in pretrained language models will arise through natural statistical pressures. We then argue that hallucinations persist due to the way most evaluations are graded -- language models are optimized to be good test-takers, and guessing when uncertain improves test performance. This \"epidemic\" of penalizing uncertain responses can only be addressed through a socio-technical mitigation: modifying the scoring of existing benchmarks that are misaligned but dominate leaderboards, rather than introducing additional hallucination evaluations. This change may steer the field toward more trustworthy AI systems.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2025-09", "2025-12", "2026-02", "2026-05"], "last_mentioned": "2026-05-26T20:56:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2510.04950", "slug": "6a8eeadd8873d680", "title": "Mind Your Tone: Investigating How Prompt Politeness Affects LLM Accuracy (short paper)", "authors": ["Om Dobariya", "Akhil Kumar"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2510.04950", "doi": null, "topics": ["Language models", "Statistics & probability", "Mathematics"], "description": "The wording of natural language prompts has been shown to influence the performance of large language models (LLMs), yet the role of politeness and tone remains underexplored.", "description_method": "abstract-excerpt", "abstract": "The wording of natural language prompts has been shown to influence the performance of large language models (LLMs), yet the role of politeness and tone remains underexplored. In this study, we investigate how varying levels of prompt politeness affect model accuracy on multiple-choice questions. We created a dataset of 50 base questions spanning mathematics, science, and history, each rewritten into five tone variants: Very Polite, Polite, Neutral, Rude, and Very Rude, yielding 250 unique prompts. Using ChatGPT 4o, we evaluated responses across these conditions and applied paired sample t-tests to assess statistical significance. Contrary to expectations, impolite prompts consistently outperformed polite ones, with accuracy ranging from 80.8% for Very Polite prompts to 84.8% for Very Rude prompts. These findings differ from earlier studies that associated rudeness with poorer outcomes, suggesting that newer LLMs may respond differently to tonal variation. Our results highlight the importance of studying pragmatic aspects of prompting and raise broader questions about the social dimensions of human-AI interaction.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2026-05", "2026-06", "2026-07", "2026-08"], "last_mentioned": "2026-08-20T21:18:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2601.20245", "slug": "426a3ed7604fcdf6", "title": "How AI Impacts Skill Formation", "authors": ["Judy Hanwen Shen", "Alex Tamkin"], "year": 2026, "venue": "arXiv", "url": "https://arxiv.org/abs/2601.20245", "doi": null, "topics": ["Programming languages", "Neuroscience"], "description": "AI assistance produces significant productivity gains across professional domains, particularly for novice workers.", "description_method": "abstract-excerpt", "abstract": "AI assistance produces significant productivity gains across professional domains, particularly for novice workers. Yet how this assistance affects the development of skills required to effectively supervise AI remains unclear. Novice workers who rely heavily on AI to complete unfamiliar tasks may compromise their own skill acquisition in the process. We conduct randomized experiments to study how developers gained mastery of a new asynchronous programming library with and without the assistance of AI. We find that AI use impairs conceptual understanding, code reading, and debugging abilities, without delivering significant efficiency gains on average. Participants who fully delegated coding tasks showed some productivity improvements, but at the cost of learning the library. We identify six distinct AI interaction patterns, three of which involve cognitive engagement and preserve learning outcomes even when participants receive AI assistance. Our findings suggest that AI-enhanced productivity is not a shortcut to competence and AI assistance should be carefully adopted into workflows to preserve skill formation -- particularly in safety-critical domains.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2026-01", "2026-02", "2026-05", "2026-06"], "last_mentioned": "2026-06-19T18:51:44Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:astro-ph/0305457", "slug": "e97b779d3bc3e871", "title": "New varying speed of light theories", "authors": ["Joao Magueijo"], "year": 2003, "venue": "arXiv", "url": "https://arxiv.org/abs/astro-ph/0305457", "doi": "10.1088/0034-4885/66/11/R04", "topics": ["Physics", "Distributed systems"], "description": "We review recent work on the possibility of a varying speed of light (VSL).", "description_method": "abstract-excerpt", "abstract": "We review recent work on the possibility of a varying speed of light (VSL). We start by discussing the physical meaning of a varying $c$, dispelling the myth that the constancy of $c$ is a matter of logical consistency. We then summarize the main VSL mechanisms proposed so far: hard breaking of Lorentz invariance; bimetric theories (where the speeds of gravity and light are not the same); locally Lorentz invariant VSL theories; theories exhibiting a color dependent speed of light; varying $c$ induced by extra dimensions (e.g. in the brane-world scenario); and field theories where VSL results from vacuum polarization or CPT violation. We show how VSL scenarios may solve the cosmological problems usually tackled by inflation, and also how they may produce a scale-invariant spectrum of Gaussian fluctuations, capable of explaining the WMAP data. We then review the connection between VSL and theories of quantum gravity, showing how ``doubly special'' relativity has emerged as a VSL effective model of quantum space-time, with observational implications for ultra high energy cosmic rays and gamma ray bursts. Some recent work on the physics of ``black'' holes and other compact objects in VSL theories is also described, highlighting phenomena associated with spatial (as opposed to temporal) variations in $c$. Finally we describe the observational status of the theory. The evidence is currently slim -- redshift dependence in the atomic fine structure, anomalies with ultra high energy cosmic rays, and (to a much lesser extent) the acceleration of the universe and the WMAP data. The constraints (e.g. those arising from nucleosynthesis or geological bounds) are tight, but not insurmountable. We conclude with the observational predictions of the theory, and the prospects for its refutation or vindication.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2013-11", "2014-08", "2021-10", "2022-07", "2025-06"], "last_mentioned": "2025-06-12T09:23:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:cs/0312044", "slug": "02b13c3533cd44ad", "title": "Clustering by compression", "authors": ["Rudi Cilibrasi", "Paul Vitanyi"], "year": 2003, "venue": "arXiv", "url": "https://arxiv.org/abs/cs/0312044", "doi": null, "topics": ["Statistics & probability", "Biology & medicine", "Algorithms"], "description": "We present a new method for clustering based on compression.", "description_method": "abstract-excerpt", "abstract": "We present a new method for clustering based on compression. The method doesn't use subject-specific features or background knowledge, and works as follows: First, we determine a universal similarity distance, the normalized compression distance or NCD, computed from the lengths of compressed data files (singly and in pairwise concatenation). Second, we apply a hierarchical clustering method. The NCD is universal in that it is not restricted to a specific application area, and works across application area boundaries. A theoretical precursor, the normalized information distance, co-developed by one of the authors, is provably optimal but uses the non-computable notion of Kolmogorov complexity. We propose precise notions of similarity metric, normal compressor, and show that the NCD based on a normal compressor is a similarity metric that approximates universality. To extract a hierarchy of clusters from the distance matrix, we determine a dendrogram (binary tree) by a new quartet method and a fast heuristic to implement it. The method is implemented and available as public software, and is robust under choice of different compressors. To substantiate our claims of universality and robustness, we report evidence of successful application in areas as diverse as genomics, virology, languages, literature, music, handwritten digits, astronomy, and combinations of objects from completely different domains, using statistical, dictionary, and block sorting compressors. In genomics we presented new evidence for major questions in Mammalian evolution, based on whole-mitochondrial genomic analysis: the Eutherian orders and the Marsupionta hypothesis against the Theria hypothesis.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2018-08", "2018-12", "2021-05", "2023-07", "2024-07"], "last_mentioned": "2024-07-11T22:25:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:cs/0610105", "slug": "0594b2ab1c759b93", "title": "How To Break Anonymity of the Netflix Prize Dataset", "authors": ["Arvind Narayanan", "Vitaly Shmatikov"], "year": 2006, "venue": "arXiv", "url": "https://arxiv.org/abs/cs/0610105", "doi": null, "topics": ["Databases", "Statistics & probability", "Security & cryptography"], "description": "We present a new class of statistical de-anonymization attacks against high-dimensional micro-data, such as individual preferences, recommendations, transaction records and so on.", "description_method": "abstract-excerpt", "abstract": "We present a new class of statistical de-anonymization attacks against high-dimensional micro-data, such as individual preferences, recommendations, transaction records and so on. Our techniques are robust to perturbation in the data and tolerate some mistakes in the adversary's background knowledge. We apply our de-anonymization methodology to the Netflix Prize dataset, which contains anonymous movie ratings of 500,000 subscribers of Netflix, the world's largest online movie rental service. We demonstrate that an adversary who knows only a little bit about an individual subscriber can easily identify this subscriber's record in the dataset. Using the Internet Movie Database as the source of background knowledge, we successfully identified the Netflix records of known users, uncovering their apparent political preferences and other potentially sensitive information.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2011-01", "2023-07", "2025-09", "2026-04"], "last_mentioned": "2026-04-23T22:43:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:hep-th/0503158", "slug": "8baede71e080dfad", "title": "The Casimir Effect and the Quantum Vacuum", "authors": ["R. L. Jaffe"], "year": 2005, "venue": "arXiv", "url": "https://arxiv.org/abs/hep-th/0503158", "doi": "10.1103/PhysRevD.72.021301", "topics": ["Physics"], "description": "In discussions of the cosmological constant, the Casimir effect is often invoked as decisive evidence that the zero point energies of quantum fields are \"real''.", "description_method": "abstract-excerpt", "abstract": "In discussions of the cosmological constant, the Casimir effect is often invoked as decisive evidence that the zero point energies of quantum fields are \"real''. On the contrary, Casimir effects can be formulated and Casimir forces can be computed without reference to zero point energies. They are relativistic, quantum forces between charges and currents. The Casimir force (per unit area) between parallel plates vanishes as α, the fine structure constant, goes to zero, and the standard result, which appears to be independent of α, corresponds to the α\\to\\infty limit.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2019-03", "2020-11", "2021-12", "2024-04", "2025-10"], "last_mentioned": "2025-10-11T09:30:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:quant-ph/9609013", "slug": "c092da9d95ebd6d9", "title": "The Ithaca Interpretation of Quantum Mechanics", "authors": ["N. David Mermin"], "year": 1996, "venue": "arXiv", "url": "https://arxiv.org/abs/quant-ph/9609013", "doi": "10.1007/BF02827447", "topics": ["Physics", "Mathematics"], "description": "I list several strong requirements for what I would consider a sensible interpretation of quantum mechanics and I discuss two simple theorems.", "description_method": "abstract-excerpt", "abstract": "I list several strong requirements for what I would consider a sensible interpretation of quantum mechanics and I discuss two simple theorems. One, as far as I know, is new; the other was only noted a few years ago. Both have important implications for such a sensible interpretation. My talk will not clear everything up; indeed, you may conclude that it has not cleared anything up. But I hope it will provide a different perspective from which to view some old and vexing puzzles (or, if you believe nothing needs to be cleared up, some ancient verities.)", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2011-04", "2015-09", "2015-10", "2025-10"], "last_mentioned": "2025-10-30T13:26:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:quant-ph/9907009", "slug": "85fa22785a61b85e", "title": "The importance of quantum decoherence in brain processes", "authors": ["Max Tegmark"], "year": 1999, "venue": "arXiv", "url": "https://arxiv.org/abs/quant-ph/9907009", "doi": "10.1103/PhysRevE.61.4194", "topics": ["Neuroscience", "Physics", "Machine learning"], "description": "Based on a calculation of neural decoherence rates, we argue that that the degrees of freedom of the human brain that relate to cognitive processes should be thought of as a classical rather than quantum system, i.e., that there is nothing fundamentally wrong with the current classical approach to neural network simulations.", "description_method": "abstract-excerpt", "abstract": "Based on a calculation of neural decoherence rates, we argue that that the degrees of freedom of the human brain that relate to cognitive processes should be thought of as a classical rather than quantum system, i.e., that there is nothing fundamentally wrong with the current classical approach to neural network simulations. We find that the decoherence timescales ~10^{-13}-10^{-20} seconds are typically much shorter than the relevant dynamical timescales (~0.001-0.1 seconds), both for regular neuron firing and for kink-like polarization excitations in microtubules. This conclusion disagrees with suggestions by Penrose and others that the brain acts as a quantum computer, and that quantum coherence is related to consciousness in a fundamental way.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2012-03", "2013-02", "2013-08", "2018-06", "2018-11"], "last_mentioned": "2018-11-07T12:20:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/35014679", "slug": "8c1a4338bcc7b338", "title": "Catching crumbs from the table", "authors": ["Ted Chiang"], "year": 2000, "venue": "Nature", "url": "https://doi.org/10.1038/35014679", "doi": "10.1038/35014679", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2019-07", "2020-05", "2021-01", "2021-03", "2025-03"], "last_mentioned": "2025-03-27T23:36:04Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature.2015.18787", "slug": "d3cdf421eed9cbcb", "title": "Engineered bat virus stirs debate over risky research", "authors": ["Declan Butler"], "year": 2015, "venue": "Nature", "url": "https://doi.org/10.1038/nature.2015.18787", "doi": "10.1038/nature.2015.18787", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2021-06", "2022-02", "2022-09"], "last_mentioned": "2022-09-06T15:03:12Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature07276", "slug": "690739deb761fca4", "title": "Old-growth forests as global carbon sinks", "authors": ["Sebastiaan Luyssaert", "E. -Detlef Schulze", "Annett Börner", "Alexander Knohl", "Dominik Hessenmöller", "Beverly E. Law", "Philippe Ciais", "John Grace"], "year": 2008, "venue": "Nature", "url": "https://doi.org/10.1038/nature07276", "doi": "10.1038/nature07276", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2019-02", "2019-07", "2021-04", "2022-03", "2023-09"], "last_mentioned": "2023-09-04T19:50:20Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature13793", "slug": "948c3c1bd1e37f64", "title": "Artificial sweeteners induce glucose intolerance by altering the gut microbiota", "authors": ["Jotham Suez", "Tal Korem", "David Zeevi", "Gili Zilberman-Schapira", "Christoph A. Thaiss", "Ori Maza", "David Israeli", "Niv Zmora", "Shlomit Gilad", "Adina Weinberger", "Yael Kuperman", "Alon Harmelin", "Ilana Kolodkin-Gal", "Hagit Shapiro", "Zamir Halpern", "Eran Segal", "Eran Elinav"], "year": 2014, "venue": "Nature", "url": "https://doi.org/10.1038/nature13793", "doi": "10.1038/nature13793", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2018-03", "2018-04", "2021-08", "2022-08", "2024-05"], "last_mentioned": "2024-05-05T21:06:20Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature16961", "slug": "7d677a5749b8efb1", "title": "Mastering the game of Go with deep neural networks and tree search", "authors": ["David Silver", "Aja Huang", "Chris J. Maddison", "Arthur Guez", "Laurent Sifre", "George van den Driessche", "Julian Schrittwieser", "Ioannis Antonoglou", "Veda Panneershelvam", "Marc Lanctot", "Sander Dieleman", "Dominik Grewe", "John Nham", "Nal Kalchbrenner", "Ilya Sutskever", "Timothy Lillicrap", "Madeleine Leach", "Koray Kavukcuoglu", "Thore Graepel", "Demis Hassabis"], "year": 2016, "venue": "Nature", "url": "https://doi.org/10.1038/nature16961", "doi": "10.1038/nature16961", "topics": ["Machine learning"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2017-11", "2019-10", "2020-06", "2023-05", "2024-03"], "last_mentioned": "2024-03-28T13:59:52Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/ncomms14063", "slug": "e3a1fb601fc108de", "title": "Caloric restriction improves health and survival of rhesus monkeys", "authors": ["Julie A. Mattison", "Ricki J. Colman", "T. Mark Beasley", "David B. Allison", "Joseph W. Kemnitz", "George S. Roth", "Donald K. Ingram", "Richard Weindruch", "Rafael de Cabo", "Rozalyn M. Anderson"], "year": 2017, "venue": "Nature Communications", "url": "https://doi.org/10.1038/ncomms14063", "doi": "10.1038/ncomms14063", "topics": ["Biology & medicine"], "description": "Abstract Caloric restriction (CR) without malnutrition extends lifespan and delays the onset of age-related disorders in most species but its impact in nonhuman primates has been controversial.", "description_method": "abstract-excerpt", "abstract": "Abstract Caloric restriction (CR) without malnutrition extends lifespan and delays the onset of age-related disorders in most species but its impact in nonhuman primates has been controversial. In the late 1980s two parallel studies were initiated to determine the effect of CR in rhesus monkeys. The University of Wisconsin study reported a significant positive impact of CR on survival, but the National Institute on Aging study detected no significant survival effect. Here we present a direct comparison of longitudinal data from both studies including survival, bodyweight, food intake, fasting glucose levels and age-related morbidity. We describe differences in study design that could contribute to differences in outcomes, and we report species specificity in the impact of CR in terms of optimal onset and diet. Taken together these data confirm that health benefits of CR are conserved in monkeys and suggest that CR mechanisms are likely translatable to human health.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2017-02", "2017-06", "2020-09"], "last_mentioned": "2020-09-23T04:49:27Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41430-022-01099-1", "slug": "4ac91fd0f1045a43", "title": "Ultra-processed foods: how functional is the NOVA system?", "authors": ["Véronique Braesco", "Isabelle Souchon", "Patrick Sauvant", "Typhaine Haurogné", "Matthieu Maillot", "Catherine Féart", "Nicole Darmon"], "year": 2022, "venue": "European Journal of Clinical Nutrition", "url": "https://doi.org/10.1038/s41430-022-01099-1", "doi": "10.1038/s41430-022-01099-1", "topics": ["Economics", "Distributed systems"], "description": "Abstract Background In the NOVA classification system, descriptive criteria are used to assign foods to one of four groups based on processing-related criteria.", "description_method": "abstract-excerpt", "abstract": "Abstract Background In the NOVA classification system, descriptive criteria are used to assign foods to one of four groups based on processing-related criteria. Although NOVA is widely used, its robustness and functionality remain largely unexplored. We determined whether this system leads to consistent food assignments by users. Methods French food and nutrition specialists completed an online survey in which they assigned foods to NOVA groups. The survey comprised two lists: one with 120 marketed food products with ingredient information and one with 111 generic food items without ingredient information. We quantified assignment consistency among evaluators using Fleiss’ κ (range: 0–1, where 1 = 100% agreement). Hierarchical clustering on principal components identified clusters of foods with similar distributions of NOVA assignments. Results Fleiss’ κ was 0.32 and 0.34 for the marketed foods (n = 159 evaluators) and generic foods (n = 177 evaluators), respectively. There were three clusters within the marketed foods: one contained 90 foods largely assigned to NOVA4 (91% of assignments), while the two others displayed greater assignment heterogeneity. There were four clusters within the generic foods: three clusters contained foods mostly assigned to a single NOVA group (69–79% of assignments), and the fourth cluster comprised 28 foods whose assignments were more evenly distributed across the four NOVA groups. Conclusions Although assignments were more consistent for some foods than others, overall consistency among evaluators was low, even when ingredient information was available. These results suggest current NOVA criteria do not allow for robust and functional food assignments.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-07", "2023-09", "2023-10", "2026-01", "2026-02"], "last_mentioned": "2026-02-04T01:30:24Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-021-24268-5", "slug": "9bebdacba4682edc", "title": "Changes in Atlantic major hurricane frequency since the late-19th century", "authors": ["Gabriel A. Vecchi", "Christopher Landsea", "Wei Zhang", "Gabriele Villarini", "Thomas Knutson"], "year": 2021, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-021-24268-5", "doi": "10.1038/s41467-021-24268-5", "topics": ["Science & research"], "description": "Abstract Atlantic hurricanes are a major hazard to life and property, and a topic of intense scientific interest.", "description_method": "abstract-excerpt", "abstract": "Abstract Atlantic hurricanes are a major hazard to life and property, and a topic of intense scientific interest. Historical changes in observing practices limit the utility of century-scale records of Atlantic major hurricane frequency. To evaluate past changes in frequency, we have here developed a homogenization method for Atlantic hurricane and major hurricane frequency over 1851–2019. We find that recorded century-scale increases in Atlantic hurricane and major hurricane frequency, and associated decrease in USA hurricanes strike fraction, are consistent with changes in observing practices and not likely a true climate trend. After homogenization, increases in basin-wide hurricane and major hurricane activity since the 1970s are not part of a century-scale increase, but a recovery from a deep minimum in the 1960s–1980s. We suggest internal (e.g., Atlantic multidecadal) climate variability and aerosol-induced mid-to-late-20th century major hurricane frequency reductions have probably masked century-scale greenhouse-gas warming contributions to North Atlantic major hurricane frequency.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2022-10", "2023-09", "2024-07", "2024-08"], "last_mentioned": "2024-08-27T12:14:02Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-022-28735-5", "slug": "861f460825641d8b", "title": "Associations between alcohol consumption and gray and white matter volumes in the UK Biobank", "authors": ["Remi Daviet", "Gökhan Aydogan", "Kanchana Jagannathan", "Nathaniel Spilka", "Philipp D. Koellinger", "Henry R. Kranzler", "Gideon Nave", "Reagan R. Wetherill"], "year": 2022, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-022-28735-5", "doi": "10.1038/s41467-022-28735-5", "topics": ["Neuroscience"], "description": "Abstract Heavy alcohol consumption has been associated with brain atrophy, neuronal loss, and poorer white matter fiber integrity.", "description_method": "abstract-excerpt", "abstract": "Abstract Heavy alcohol consumption has been associated with brain atrophy, neuronal loss, and poorer white matter fiber integrity. However, there is conflicting evidence on whether light-to-moderate alcohol consumption shows similar negative associations with brain structure. To address this, we examine the associations between alcohol intake and brain structure using multimodal imaging data from 36,678 generally healthy middle-aged and older adults from the UK Biobank, controlling for numerous potential confounds. Consistent with prior literature, we find negative associations between alcohol intake and brain macrostructure and microstructure. Specifically, alcohol intake is negatively associated with global brain volume measures, regional gray matter volumes, and white matter microstructure. Here, we show that the negative associations between alcohol intake and brain macrostructure and microstructure are already apparent in individuals consuming an average of only one to two daily alcohol units, and become stronger as alcohol intake increases.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2022-03", "2024-01", "2025-04"], "last_mentioned": "2025-04-22T14:55:38Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41526-019-0071-6", "slug": "170b093f9583e685", "title": "Effects of acute exposures to carbon dioxide on decision making and cognition in astronaut-like subjects", "authors": ["Robert R. Scully", "Mathias Basner", "Jad Nasrini", "Chiu-wing Lam", "Emanuel Hermosillo", "Ruben C. Gur", "Tyler Moore", "David J. Alexander", "Usha Satish", "Valerie E. Ryder"], "year": 2019, "venue": "npj Microgravity", "url": "https://doi.org/10.1038/s41526-019-0071-6", "doi": "10.1038/s41526-019-0071-6", "topics": ["Psychology", "Neuroscience"], "description": "Abstract Acute exposure to carbon dioxide (CO 2 ) concentrations below those found on the International Space Station are reported to deteriorate complex decision-making.", "description_method": "abstract-excerpt", "abstract": "Abstract Acute exposure to carbon dioxide (CO 2 ) concentrations below those found on the International Space Station are reported to deteriorate complex decision-making. Effective decision-making is critical to human spaceflight, especially during an emergency response. Therefore, effects of acutely elevated CO 2 on decision-making competency and various cognitive domains were assessed in astronaut-like subjects by the Strategic Management Simulation (SMS) and Cognition test batteries. The double-blind cross-over study included 22 participants at the Johnson Space Center randomly assigned to one of four groups. Each group was exposed to a different sequence of four concentrations of CO 2 (600, 1200, 2500, 5000 ppm). Subjects performed Cognition before entering the chamber, 15 min and 2.5 h after entering the chamber, and 15 min after exiting the chamber. The SMS was administered 30 min after subjects entered the chamber. There were no clear dose–response patterns for performance on either SMS or Cognition . Performance on most SMS measures and aggregate speed, accuracy, and efficiency scores across Cognition tests were lower at 1200 ppm than at baseline (600 ppm); however, at higher CO 2 concentrations performance was similar to or exceeded baseline for most measures. These outcomes, which conflict with those of other studies, likely indicate differing characteristics of the various subject populations and differences in the aggregation of unrecognized stressors, in addition to CO 2 , are responsible for disparate outcomes among studies. Studies with longer exposure durations are needed to verify that cognitive impairment does not develop over time in crew-like subjects.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2019-12", "2022-01", "2022-07", "2023-05", "2026-07"], "last_mentioned": "2026-07-04T20:48:10Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41533-021-00223-1", "slug": "055d77c4c58f89b2", "title": "Are smokers protected against SARS-CoV-2 infection (COVID-19)? The origins of the myth", "authors": ["Naomi A. van Westen-Lagerweij", "Eline Meijer", "Elisabeth G. Meeuwsen", "Niels H. Chavannes", "Marc C. Willemsen", "Esther A. Croes"], "year": 2021, "venue": "npj Primary Care Respiratory Medicine", "url": "https://doi.org/10.1038/s41533-021-00223-1", "doi": "10.1038/s41533-021-00223-1", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2021-04", "2021-09", "2022-01", "2025-06"], "last_mentioned": "2025-06-30T07:29:05Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41562-021-01196-4", "slug": "ede91b4d8883fbb7", "title": "The effects of remote work on collaboration among information workers", "authors": ["Longqi Yang", "David Holtz", "Sonia Jaffe", "Siddharth Suri", "Shilpi Sinha", "Jeffrey Weston", "Connor Joyce", "Neha Shah", "Kevin Sherman", "Brent Hecht", "Jaime Teevan"], "year": 2021, "venue": "Nature Human Behaviour", "url": "https://doi.org/10.1038/s41562-021-01196-4", "doi": "10.1038/s41562-021-01196-4", "topics": ["Economics"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2022-01", "2022-03", "2023-02", "2023-09", "2025-04"], "last_mentioned": "2025-04-29T14:21:08Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41579-021-00642-4", "slug": "e372403d22a65ae4", "title": "Influenza lineage extinction during the COVID-19 pandemic?", "authors": ["Marios Koutsakos", "Adam K. Wheatley", "Karen Laurie", "Stephen J. Kent", "Steve Rockman"], "year": 2021, "venue": "Nature Reviews Microbiology", "url": "https://doi.org/10.1038/s41579-021-00642-4", "doi": "10.1038/s41579-021-00642-4", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2021-12", "2022-03", "2022-04", "2023-01", "2023-08"], "last_mentioned": "2023-08-03T08:25:15Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-019-1193-4", "slug": "f0888c5510667b10", "title": "Increase in CFC-11 emissions from eastern China based on atmospheric observations", "authors": ["M. Rigby", "S. Park", "T. Saito", "L. M. Western", "A. L. Redington", "X. Fang", "S. Henne", "A. J. Manning", "R. G. Prinn", "G. S. Dutton", "P. J. Fraser", "A. L. Ganesan", "B. D. Hall", "C. M. Harth", "J. Kim", "K.-R. Kim", "P. B. Krummel", "T. Lee", "S. Li", "Q. Liang", "M. F. Lunt", "S. A. Montzka", "J. Mühle", "S. O’Doherty", "M.-K. Park", "S. Reimann", "P. K. Salameh", "P. Simmonds", "R. L. Tunnicliffe", "R. F. Weiss", "Y. Yokouchi", "D. Young"], "year": 2019, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-019-1193-4", "doi": "10.1038/s41586-019-1193-4", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2021-04", "2021-08", "2022-02"], "last_mentioned": "2022-02-08T17:11:57Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2012-7", "slug": "6db015547d5d9ef1", "title": "A pneumonia outbreak associated with a new coronavirus of probable bat origin", "authors": ["Peng Zhou", "Xing-Lou Yang", "Xian-Guang Wang", "Ben Hu", "Lei Zhang", "Wei Zhang", "Hao-Rui Si", "Yan Zhu", "Bei Li", "Chao-Lin Huang", "Hui-Dong Chen", "Jing Chen", "Yun Luo", "Hua Guo", "Ren-Di Jiang", "Mei-Qin Liu", "Ying Chen", "Xu-Rui Shen", "Xi Wang", "Xiao-Shuang Zheng", "Kai Zhao", "Quan-Jiao Chen", "Fei Deng", "Lin-Lin Liu", "Bing Yan", "Fa-Xian Zhan", "Yan-Yi Wang", "Geng-Fu Xiao", "Zheng-Li Shi"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2012-7", "doi": "10.1038/s41586-020-2012-7", "topics": ["Biology & medicine", "Economics"], "description": "Abstract Since the outbreak of severe acute respiratory syndrome (SARS) 18 years ago, a large number of SARS-related coronaviruses (SARSr-CoVs) have been discovered in their natural reservoir host, bats 1–4 .", "description_method": "abstract-excerpt", "abstract": "Abstract Since the outbreak of severe acute respiratory syndrome (SARS) 18 years ago, a large number of SARS-related coronaviruses (SARSr-CoVs) have been discovered in their natural reservoir host, bats 1–4 . Previous studies have shown that some bat SARSr-CoVs have the potential to infect humans 5–7 . Here we report the identification and characterization of a new coronavirus (2019-nCoV), which caused an epidemic of acute respiratory syndrome in humans in Wuhan, China. The epidemic, which started on 12 December 2019, had caused 2,794 laboratory-confirmed infections including 80 deaths by 26 January 2020. Full-length genome sequences were obtained from five patients at an early stage of the outbreak. The sequences are almost identical and share 79.6% sequence identity to SARS-CoV. Furthermore, we show that 2019-nCoV is 96% identical at the whole-genome level to a bat coronavirus. Pairwise protein sequence analysis of seven conserved non-structural proteins domains show that this virus belongs to the species of . In addition, 2019-nCoV virus isolated from the bronchoalveolar lavage fluid of a critically ill patient could be neutralized by sera from several patients. Notably, we confirmed that 2019-nCoV uses the same cell entry receptor—angiotensin converting enzyme II (ACE2)—as SARS-CoV.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2020-06", "2020-07", "2020-09", "2022-02"], "last_mentioned": "2022-02-10T19:01:21Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2405-7", "slug": "652f405a52342501", "title": "Estimating the effects of non-pharmaceutical interventions on COVID-19 in Europe", "authors": ["Seth Flaxman", "Swapnil Mishra", "Axel Gandy", "H. Juliette T. Unwin", "Thomas A. Mellan", "Helen Coupland", "Charles Whittaker", "Harrison Zhu", "Tresnia Berah", "Jeffrey W. Eaton", "Mélodie Monod", "Imperial College COVID-19 Response Team", "Pablo N. Perez-Guzman", "Nora Schmit", "Lucia Cilloni", "Kylie E. C. Ainslie", "Marc Baguelin", "Adhiratha Boonyasiri", "Olivia Boyd", "Lorenzo Cattarino", "Laura V. Cooper", "Zulma Cucunubá", "Gina Cuomo-Dannenburg", "Amy Dighe", "Bimandra Djaafara", "Ilaria Dorigatti", "Sabine L. van Elsland", "Richard G. FitzJohn", "Katy A. M. Gaythorpe", "Lily Geidelberg", "Nicholas C. Grassly", "William D. Green", "Timothy Hallett", "Arran Hamlet", "Wes Hinsley", "Ben Jeffrey", "Edward Knock", "Daniel J. Laydon", "Gemma Nedjati-Gilani", "Pierre Nouvellet", "Kris V. Parag", "Igor Siveroni", "Hayley A. Thompson", "Robert Verity", "Erik Volz", "Caroline E. Walters", "Haowei Wang", "Yuanrong Wang", "Oliver J. Watson", "Peter Winskill", "Xiaoyue Xi", "Patrick G. T. Walker", "Azra C. Ghani", "Christl A. Donnelly", "Steven Riley", "Michaela A. C. Vollmer", "Neil M. Ferguson", "Lucy C. Okell", "Samir Bhatt"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2405-7", "doi": "10.1038/s41586-020-2405-7", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2020-12", "2021-01", "2021-06", "2022-01", "2023-03"], "last_mentioned": "2023-03-06T12:33:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-3010-5", "slug": "6e4d5dd7f7e244c5", "title": "Global human-made mass exceeds all living biomass", "authors": ["Emily Elhacham", "Liad Ben-Uri", "Jonathan Grozovski", "Yinon M. Bar-On", "Ron Milo"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-3010-5", "doi": "10.1038/s41586-020-3010-5", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2020-12", "2021-06", "2021-10", "2022-07", "2022-08"], "last_mentioned": "2022-08-09T18:00:42Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-021-04086-x", "slug": "4b5a71bf6449cc4c", "title": "Advancing mathematics by guiding human intuition with AI", "authors": ["Alex Davies", "Petar Veličković", "Lars Buesing", "Sam Blackwell", "Daniel Zheng", "Nenad Tomašev", "Richard Tanburn", "Peter Battaglia", "Charles Blundell", "András Juhász", "Marc Lackenby", "Geordie Williamson", "Demis Hassabis", "Pushmeet Kohli"], "year": 2021, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-021-04086-x", "doi": "10.1038/s41586-021-04086-x", "topics": ["Mathematics", "Machine learning", "Algorithms"], "description": "Abstract The practice of mathematics involves discovering patterns and using these to formulate and prove conjectures, resulting in theorems.", "description_method": "abstract-excerpt", "abstract": "Abstract The practice of mathematics involves discovering patterns and using these to formulate and prove conjectures, resulting in theorems. Since the 1960s, mathematicians have used computers to assist in the discovery of patterns and formulation of conjectures 1 , most famously in the Birch and Swinnerton-Dyer conjecture 2 , a Millennium Prize Problem 3 . Here we provide examples of new fundamental results in pure mathematics that have been discovered with the assistance of machine learning—demonstrating a method by which machine learning can aid mathematicians in discovering new conjectures and theorems. We propose a process of using machine learning to discover potential patterns and relations between mathematical objects, understanding them with attribution techniques and using these observations to guide intuition and propose conjectures. We outline this machine-learning-guided framework and demonstrate its successful application to current research questions in distinct areas of pure mathematics, in each case showing how it led to meaningful mathematical contributions on important open problems: a new connection between the algebraic and geometric structure of knots, and a candidate algorithm predicted by the combinatorial invariance conjecture for symmetric groups 4 . Our work may serve as a model for collaboration between the fields of mathematics and artificial intelligence (AI) that can achieve surprising results by leveraging the respective strengths of mathematicians and machine learning.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2021-12", "2022-09", "2022-12", "2025-11"], "last_mentioned": "2025-11-17T15:29:08Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-022-05424-3", "slug": "064d984d57eefa08", "title": "Traversable wormhole dynamics on a quantum processor", "authors": ["Daniel Jafferis", "Alexander Zlokapa", "Joseph D. Lykken", "David K. Kolchmeyer", "Samantha I. Davis", "Nikolai Lauk", "Hartmut Neven", "Maria Spiropulu"], "year": 2022, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-022-05424-3", "doi": "10.1038/s41586-022-05424-3", "topics": ["Physics", "Computer systems"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2022-11", "2022-12", "2024-04", "2024-11"], "last_mentioned": "2024-11-27T02:53:24Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-024-07421-0", "slug": "8e0c2cfcb31292f5", "title": "Detecting hallucinations in large language models using semantic entropy", "authors": ["Sebastian Farquhar", "Jannik Kossen", "Lorenz Kuhn", "Yarin Gal"], "year": 2024, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-024-07421-0", "doi": "10.1038/s41586-024-07421-0", "topics": ["Statistics & probability", "Language models", "Biology & medicine"], "description": "Abstract Large language model (LLM) systems, such as ChatGPT 1 or Gemini 2 , can show impressive reasoning and question-answering capabilities but often ‘hallucinate’ false outputs and unsubstantiated answers 3,4 .", "description_method": "abstract-excerpt", "abstract": "Abstract Large language model (LLM) systems, such as ChatGPT 1 or Gemini 2 , can show impressive reasoning and question-answering capabilities but often ‘hallucinate’ false outputs and unsubstantiated answers 3,4 . Answering unreliably or without the necessary information prevents adoption in diverse fields, with problems including fabrication of legal precedents 5 or untrue facts in news articles 6 and even posing a risk to human life in medical domains such as radiology 7 . Encouraging truthfulness through supervision or reinforcement has been only partially successful 8 . Researchers need a general method for detecting hallucinations in LLMs that works even with new and unseen questions to which humans might not know the answer. Here we develop new methods grounded in statistics, proposing entropy-based uncertainty estimators for LLMs to detect a subset of hallucinations—confabulations—which are arbitrary and incorrect generations. Our method addresses the fact that one idea can be expressed in many ways by computing uncertainty at the level of meaning rather than specific sequences of words. Our method works across datasets and tasks without a priori knowledge of the task, requires no task-specific data and robustly generalizes to new tasks not seen before. By detecting when a prompt is likely to produce a confabulation, our method helps users understand when they must take extra care with LLMs and opens up new possibilities for using LLMs that are otherwise prevented by their unreliability.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2024-06", "2024-09", "2024-12", "2026-03"], "last_mentioned": "2026-03-26T21:04:24Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41588-018-0147-3", "slug": "fe02aef62c2b644d", "title": "Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals", "authors": ["23andMe Research Team", "James J. Lee", "COGENT (Cognitive Genomics Consortium)", "Social Science Genetic Association Consortium", "Robbee Wedow", "Aysu Okbay", "Edward Kong", "Omeed Maghzian", "Meghan Zacher", "Tuan Anh Nguyen-Viet", "Peter Bowers", "Julia Sidorenko", "Richard Karlsson Linnér", "Mark Alan Fontana", "Tushar Kundu", "Chanwook Lee", "Hui Li", "Ruoxi Li", "Rebecca Royer", "Pascal N. Timshel", "Raymond K. Walters", "Emily A. Willoughby", "Loïc Yengo", "Maris Alver", "Yanchun Bao", "David W. Clark", "Felix R. Day", "Nicholas A. Furlotte", "Peter K. Joshi", "Kathryn E. Kemper", "Aaron Kleinman", "Claudia Langenberg", "Reedik Mägi", "Joey W. Trampush", "Shefali Setia Verma", "Yang Wu", "Max Lam", "Jing Hua Zhao", "Zhili Zheng", "Jason D. Boardman", "Harry Campbell", "Jeremy Freese", "Kathleen Mullan Harris", "Caroline Hayward", "Pamela Herd", "Meena Kumari", "Todd Lencz", "Jian’an Luan", "Anil K. Malhotra", "Andres Metspalu", "Lili Milani", "Ken K. Ong", "John R. B. Perry", "David J. Porteous", "Marylyn D. Ritchie", "Melissa C. Smart", "Blair H. Smith", "Joyce Y. Tung", "Nicholas J. Wareham", "James F. Wilson", "Jonathan P. Beauchamp", "Dalton C. Conley", "Tõnu Esko", "Steven F. Lehrer", "Patrik K. E. Magnusson", "Sven Oskarsson", "Tune H. Pers", "Matthew R. Robinson", "Kevin Thom", "Chelsea Watson", "Christopher F. Chabris", "Michelle N. Meyer", "David I. Laibson", "Jian Yang", "Magnus Johannesson", "Philipp D. Koellinger", "Patrick Turley", "Peter M. Visscher", "Daniel J. Benjamin", "David Cesarini"], "year": 2018, "venue": "Nature Genetics", "url": "https://doi.org/10.1038/s41588-018-0147-3", "doi": "10.1038/s41588-018-0147-3", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2018-07", "2018-10", "2019-02", "2021-02", "2021-09"], "last_mentioned": "2021-09-08T18:23:14Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41590-021-01113-x", "slug": "abc685df577fa592", "title": "Immunological dysfunction persists for 8 months following initial mild-to-moderate SARS-CoV-2 infection", "authors": ["Chansavath Phetsouphanh", "David R. Darley", "Daniel B. Wilson", "Annett Howe", "C. Mee Ling Munier", "Sheila K. Patel", "Jennifer A. Juno", "Louise M. Burrell", "Stephen J. Kent", "Gregory J. Dore", "Anthony D. Kelleher", "Gail V. Matthews"], "year": 2022, "venue": "Nature Immunology", "url": "https://doi.org/10.1038/s41590-021-01113-x", "doi": "10.1038/s41590-021-01113-x", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2022-10", "2022-11", "2022-12", "2023-01", "2023-05"], "last_mentioned": "2023-05-13T01:06:47Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-020-0869-5", "slug": "9063cde6b085fb34", "title": "Temporal dynamics in viral shedding and transmissibility of COVID-19", "authors": ["Xi He", "Eric H. Y. Lau", "Peng Wu", "Xilong Deng", "Jian Wang", "Xinxin Hao", "Yiu Chung Lau", "Jessica Y. Wong", "Yujuan Guan", "Xinghua Tan", "Xiaoneng Mo", "Yanqing Chen", "Baolin Liao", "Weilie Chen", "Fengyu Hu", "Qing Zhang", "Mingqiu Zhong", "Yanrong Wu", "Lingzhai Zhao", "Fuchun Zhang", "Benjamin J. Cowling", "Fang Li", "Gabriel M. Leung"], "year": 2020, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-020-0869-5", "doi": "10.1038/s41591-020-0869-5", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2020-04", "2020-12", "2021-08", "2021-09"], "last_mentioned": "2021-09-29T19:12:20Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-022-01968-z", "slug": "0c3fc64892b1a565", "title": "Health effects associated with consumption of unprocessed red meat: a Burden of Proof study", "authors": ["Haley Lescinsky", "Ashkan Afshin", "Charlie Ashbaugh", "Catherine Bisignano", "Michael Brauer", "Giannina Ferrara", "Simon I. Hay", "Jiawei He", "Vincent Iannucci", "Laurie B. Marczak", "Susan A. McLaughlin", "Erin C. Mullany", "Marie C. Parent", "Audrey L. Serfes", "Reed J. D. Sorensen", "Aleksandr Y. Aravkin", "Peng Zheng", "Christopher J. L. Murray"], "year": 2022, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-022-01968-z", "doi": "10.1038/s41591-022-01968-z", "topics": ["Biology & medicine", "Statistics & probability"], "description": "AbstractCharacterizing the potential health effects of exposure to risk factors such as red meat consumption is essential to inform health policy and practice.", "description_method": "abstract-excerpt", "abstract": "AbstractCharacterizing the potential health effects of exposure to risk factors such as red meat consumption is essential to inform health policy and practice. Previous meta-analyses evaluating the effects of red meat intake have generated mixed findings and do not formally assess evidence strength. Here, we conducted a systematic review and implemented a meta-regression—relaxing conventional log-linearity assumptions and incorporating between-study heterogeneity—to evaluate the relationships between unprocessed red meat consumption and six potential health outcomes. We found weak evidence of association between unprocessed red meat consumption and colorectal cancer, breast cancer, type 2 diabetes and ischemic heart disease. Moreover, we found no evidence of an association between unprocessed red meat and ischemic stroke or hemorrhagic stroke. We also found that while risk for the six outcomes in our analysis combined was minimized at 0 g unprocessed red meat intake per day, the 95% uncertainty interval that incorporated between-study heterogeneity was very wide: from 0–200 g d−1. While there is some evidence that eating unprocessed red meat is associated with increased risk of disease incidence and mortality, it is weak and insufficient to make stronger or more conclusive recommendations. More rigorous, well-powered research is needed to better understand and quantify the relationship between consumption of unprocessed red meat and chronic disease.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2022-11", "2023-05", "2024-07"], "last_mentioned": "2024-07-12T02:08:11Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-023-02223-9", "slug": "d76d23f490d3bde0", "title": "The artificial sweetener erythritol and cardiovascular event risk", "authors": ["Marco Witkowski", "Ina Nemet", "Hassan Alamri", "Jennifer Wilcox", "Nilaksh Gupta", "Nisreen Nimer", "Arash Haghikia", "Xinmin S. Li", "Yuping Wu", "Prasenjit Prasad Saha", "Ilja Demuth", "Maximilian König", "Elisabeth Steinhagen-Thiessen", "Tomas Cajka", "Oliver Fiehn", "Ulf Landmesser", "W. H. Wilson Tang", "Stanley L. Hazen"], "year": 2023, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-023-02223-9", "doi": "10.1038/s41591-023-02223-9", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2023-02", "2023-10", "2024-01", "2025-06", "2025-07"], "last_mentioned": "2025-07-20T07:44:15Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-020-63657-6", "slug": "b499aa937d18616f", "title": "Deforestation and world population sustainability: a quantitative analysis", "authors": ["Mauro Bologna", "Gerardo Aquino"], "year": 2020, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-020-63657-6", "doi": "10.1038/s41598-020-63657-6", "topics": ["Statistics & probability"], "description": "Abstract In this paper we afford a quantitative analysis of the sustainability of current world population growth in relation to the parallel deforestation process adopting a statistical point of view.", "description_method": "abstract-excerpt", "abstract": "Abstract In this paper we afford a quantitative analysis of the sustainability of current world population growth in relation to the parallel deforestation process adopting a statistical point of view. We consider a simplified model based on a stochastic growth process driven by a continuous time random walk, which depicts the technological evolution of human kind, in conjunction with a deterministic generalised logistic model for humans-forest interaction and we evaluate the probability of avoiding the self-destruction of our civilisation. Based on the current resource consumption rates and best estimate of technological rate growth our study shows that we have very low probability, less than 10% in most optimistic estimate, to survive without facing a catastrophic collapse.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2020-07", "2021-12", "2023-07", "2023-08", "2023-11"], "last_mentioned": "2023-11-09T20:36:51Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-021-95699-9", "slug": "3bdb074b5f3c0384", "title": "Comparing the responses of the UK, Sweden and Denmark to COVID-19 using counterfactual modelling", "authors": ["Swapnil Mishra", "James A. Scott", "Daniel J. Laydon", "Seth Flaxman", "Axel Gandy", "Thomas A. Mellan", "H. Juliette T. Unwin", "Michaela Vollmer", "Helen Coupland", "Oliver Ratmann", "Melodie Monod", "Harrison H. Zhu", "Anne Cori", "Katy A. M. Gaythorpe", "Lilith K. Whittles", "Charles Whittaker", "Christl A. Donnelly", "Neil M. Ferguson", "Samir Bhatt"], "year": 2021, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-021-95699-9", "doi": "10.1038/s41598-021-95699-9", "topics": ["Statistics & probability"], "description": "AbstractThe UK and Sweden have among the worst per-capita COVID-19 mortality in Europe.", "description_method": "abstract-excerpt", "abstract": "AbstractThe UK and Sweden have among the worst per-capita COVID-19 mortality in Europe. Sweden stands out for its greater reliance on voluntary, rather than mandatory, control measures. We explore how the timing and effectiveness of control measures in the UK, Sweden and Denmark shaped COVID-19 mortality in each country, using a counterfactual assessment: what would the impact have been, had each country adopted the others’ policies? Using a Bayesian semi-mechanistic model without prior assumptions on the mechanism or effectiveness of interventions, we estimate the time-varying reproduction number for the UK, Sweden and Denmark from daily mortality data. We use two approaches to evaluate counterfactuals which transpose the transmission profile from one country onto another, in each country’s first wave from 13th March (when stringent interventions began) until 1st July 2020. UK mortality would have approximately doubled had Swedish policy been adopted, while Swedish mortality would have more than halved had Sweden adopted UK or Danish strategies. Danish policies were most effective, although differences between the UK and Denmark were significant for one counterfactual approach only. Our analysis shows that small changes in the timing or effectiveness of interventions have disproportionately large effects on total mortality within a rapidly growing epidemic.", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2021-09", "2021-10", "2021-12", "2022-03"], "last_mentioned": "2022-03-29T06:47:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-025-27053-2", "slug": "14801ce183e3367f", "title": "Overestimates of social media addiction are common but costly", "authors": ["Ian A. Anderson", "Wendy Wood"], "year": 2025, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-025-27053-2", "doi": "10.1038/s41598-025-27053-2", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2026-02", "2026-03", "2026-06"], "last_mentioned": "2026-06-09T22:17:57Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41893-019-0220-7", "slug": "6eaeff7b30962f58", "title": "China and India lead in greening of the world through land-use management", "authors": ["Chi Chen", "Taejin Park", "Xuhui Wang", "Shilong Piao", "Baodong Xu", "Rajiv K. Chaturvedi", "Richard Fuchs", "Victor Brovkin", "Philippe Ciais", "Rasmus Fensholt", "Hans Tømmervik", "Govindasamy Bala", "Zaichun Zhu", "Ramakrishna R. Nemani", "Ranga B. Myneni"], "year": 2019, "venue": "Nature Sustainability", "url": "https://doi.org/10.1038/s41893-019-0220-7", "doi": "10.1038/s41893-019-0220-7", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2019-02", "2019-05", "2019-11", "2023-10"], "last_mentioned": "2023-10-23T17:36:42Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/srep35070", "slug": "dbf640e8ff409bc7", "title": "The Photovoltaic Heat Island Effect: Larger solar power plants increase local temperatures", "authors": ["Greg A. Barron-Gafford", "Rebecca L. Minor", "Nathan A. Allen", "Alex D. Cronin", "Adria E. Brooks", "Mitchell A. Pavao-Zuckerman"], "year": 2016, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/srep35070", "doi": "10.1038/srep35070", "topics": ["Psychology"], "description": "Abstract While photovoltaic (PV) renewable energy production has surged, concerns remain about whether or not PV power plants induce a “heat island” (PVHI) effect, much like the increase in ambient temperatures relative to wildlands generates an Urban Heat Island effect in cities.", "description_method": "abstract-excerpt", "abstract": "Abstract While photovoltaic (PV) renewable energy production has surged, concerns remain about whether or not PV power plants induce a “heat island” (PVHI) effect, much like the increase in ambient temperatures relative to wildlands generates an Urban Heat Island effect in cities. Transitions to PV plants alter the way that incoming energy is reflected back to the atmosphere or absorbed, stored, and reradiated because PV plants change the albedo, vegetation, and structure of the terrain. Prior work on the PVHI has been mostly theoretical or based upon simulated models. Furthermore, past empirical work has been limited in scope to a single biome. Because there are still large uncertainties surrounding the potential for a PHVI effect, we examined the PVHI empirically with experiments that spanned three biomes. We found temperatures over a PV plant were regularly 3–4 °C warmer than wildlands at night, which is in direct contrast to other studies based on models that suggested that PV systems should decrease ambient temperatures. Deducing the underlying cause and scale of the PVHI effect and identifying mitigation strategies are key in supporting decision-making regarding PV development, particularly in semiarid landscapes, which are among the most likely for large-scale PV installations.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2019-02", "2021-03", "2023-03", "2023-10"], "last_mentioned": "2023-10-05T10:33:51Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.aaq0216", "slug": "7d4b3e1390613a1e", "title": "Reducing food’s environmental impacts through producers and consumers", "authors": ["J. Poore", "T. Nemecek"], "year": 2018, "venue": "Science", "url": "https://doi.org/10.1126/science.aaq0216", "doi": "10.1126/science.aaq0216", "topics": ["Other research"], "description": "The global impacts of food production Food is produced and processed by millions of farmers and intermediaries globally, with substantial associated environmental costs.", "description_method": "abstract-excerpt", "abstract": "The global impacts of food production Food is produced and processed by millions of farmers and intermediaries globally, with substantial associated environmental costs. Given the heterogeneity of producers, what is the best way to reduce food's environmental impacts? Poore and Nemecek consolidated data on the multiple environmental impacts of ∼38,000 farms producing 40 different agricultural goods around the world in a meta-analysis comparing various types of food production systems. The environmental cost of producing the same goods can be highly variable. However, this heterogeneity creates opportunities to target the small numbers of producers that have the most impact. Science , this issue p. 987", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2022-05", "2023-03", "2023-07", "2023-08"], "last_mentioned": "2023-08-03T16:34:12Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abm4454", "slug": "7b39221c2a12cd95", "title": "Dissecting the early COVID-19 cases in Wuhan", "authors": ["Michael Worobey"], "year": 2021, "venue": "Science", "url": "https://doi.org/10.1126/science.abm4454", "doi": "10.1126/science.abm4454", "topics": ["Other research"], "description": "Elucidating the origin of the pandemic requires understanding of the Wuhan outbreak", "description_method": "abstract-excerpt", "abstract": "Elucidating the origin of the pandemic requires understanding of the Wuhan outbreak", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2021-11", "2022-01", "2023-02"], "last_mentioned": "2023-02-26T17:12:33Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/1045339.1045340", "slug": "e67c196d6ecb3420", "title": "Artificial intelligence meets natural stupidity", "authors": ["Drew McDermott"], "year": 1976, "venue": "ACM SIGART Bulletin", "url": "https://doi.org/10.1145/1045339.1045340", "doi": "10.1145/1045339.1045340", "topics": ["Psychology"], "description": "As a field, artificial intelligence has always been on the border of respectability, and therefore on the border of crackpottery.", "description_method": "abstract-excerpt", "abstract": "As a field, artificial intelligence has always been on the border of respectability, and therefore on the border of crackpottery. Many critics <Dreyfus, 1972>, <Lighthill, 1973> have urged that we are over the border. We have been very defensive toward this charge, drawing ourselves up with dignity when it is made and folding the cloak of Science about us. On the other hand, in private, we have been justifiably proud of our willingness to explore weird ideas, because pursuing them is the only way to make progress.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2023-04", "2023-05", "2025-04", "2025-08", "2026-07"], "last_mentioned": "2026-07-23T12:06:49Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/2854146", "slug": "f04f9fef1361b449", "title": "Why Google stores billions of lines of code in a single repository", "authors": ["Rachel Potvin", "Josh Levenberg"], "year": 2016, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/2854146", "doi": "10.1145/2854146", "topics": ["Other research"], "description": "Google's monolithic repository provides a common source of truth for tens of thousands of developers around the world.", "description_method": "abstract-excerpt", "abstract": "Google's monolithic repository provides a common source of truth for tens of thousands of developers around the world.", "score": 0.0, "mention_count": 5, "user_count": 5, "months": ["2021-12", "2022-02", "2022-04", "2022-12", "2026-05"], "last_mentioned": "2026-05-13T23:47:17Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/321239.321249", "slug": "3e2beaccfb9461ff", "title": "Derivatives of Regular Expressions", "authors": ["Janusz A. Brzozowski"], "year": 1964, "venue": "Journal of the ACM", "url": "https://doi.org/10.1145/321239.321249", "doi": "10.1145/321239.321249", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2020-04", "2023-01", "2023-03"], "last_mentioned": "2023-03-07T17:44:44Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3317550.3321441", "slug": "8db08a3836e28448", "title": "Machine Learning Systems are Stuck in a Rut", "authors": ["Paul Barham", "Michael Isard"], "year": 2019, "venue": "Proceedings of the Workshop on Hot Topics in Operating Systems", "url": "https://doi.org/10.1145/3317550.3321441", "doi": "10.1145/3317550.3321441", "topics": ["Machine learning"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2020-05", "2020-12", "2021-07"], "last_mentioned": "2021-07-26T22:40:53Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3386324", "slug": "3d2f456ffebf2d50", "title": "Evolution of Emacs Lisp", "authors": ["Stefan Monnier", "Michael Sperber"], "year": 2020, "venue": "Proceedings of the ACM on Programming Languages", "url": "https://doi.org/10.1145/3386324", "doi": "10.1145/3386324", "topics": ["Programming languages"], "description": "While Emacs proponents largely agree that it is the world’s greatest text editor, it is almost as much a Lisp machine disguised as an editor.", "description_method": "abstract-excerpt", "abstract": "While Emacs proponents largely agree that it is the world’s greatest text editor, it is almost as much a Lisp machine disguised as an editor. Indeed, one of its chief appeals is that it is programmable via its own programming language. Emacs Lisp is a Lisp in the classic tradition. In this article, we present the history of this language over its more than 30 years of evolution. Its core has remained remarkably stable since its inception in 1985, in large part to preserve compatibility with the many third-party packages providing a multitude of extensions. Still, Emacs Lisp has evolved and continues to do so. Important aspects of Emacs Lisp have been shaped by concrete requirements of the editor it supports as well as implementation constraints. These requirements led to the choice of a Lisp dialect as Emacs’s language in the first place, specifically its simplicity and dynamic nature: Loading additional Emacs packages or changing the ones in place occurs frequently, and having to restart the editor in order to re-compile or re-link the code would be unacceptable. Fulfilling this requirement in a more static language would have been difficult at best. One of Lisp’s chief characteristics is its malleability through its uniform syntax and the use of macros. This has allowed the language to evolve much more rapidly and substantively than the evolution of its core would suggest, by letting Emacs packages provide new surface syntax alongside new functions. In particular, Emacs Lisp can be customized to look much like Common Lisp, and additional packages provide multiple-dispatch object systems, legible regular expressions, programmable pattern-matching constructs, generalized variables, and more. Still, the core has also evolved, albeit slowly. Most notably, it acquired support for lexical scoping. The timeline of Emacs Lisp development is closely tied to the projects and people who have shaped it over the years: We document Emacs Lisp history through its predecessors, Mocklisp and MacLisp, its early development up to the “Emacs schism” and the fork of Lucid Emacs, the development of XEmacs, and the subsequent rennaissance of Emacs development.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2022-11", "2022-12", "2024-05", "2024-08", "2025-11"], "last_mentioned": "2025-11-01T22:07:03Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3386332", "slug": "900f56bf25e538d2", "title": "The origins of Objective-C at PPI/Stepstone and its evolution at NeXT", "authors": ["Brad J. Cox", "Steve Naroff", "Hansen Hsu"], "year": 2020, "venue": "Proceedings of the ACM on Programming Languages", "url": "https://doi.org/10.1145/3386332", "doi": "10.1145/3386332", "topics": ["Programming languages", "Economics", "Computer systems"], "description": "The roots of Objective-C began at ITT in the early 1980s in a research group led by Tom Love investigating improving programmer productivity by an order of magnitude, a concern motivated by the perceived \"software crisis\" articulated in the late 1960s.", "description_method": "abstract-excerpt", "abstract": "The roots of Objective-C began at ITT in the early 1980s in a research group led by Tom Love investigating improving programmer productivity by an order of magnitude, a concern motivated by the perceived \"software crisis\" articulated in the late 1960s. In 1981, Brad Cox, a member of this group, began to investigate Smalltalk and object-oriented programming for this purpose, but needed a language compatible with the Unix and C environments used by ITT. That year, Cox quickly wrote up the Object-Oriented Pre-Compiler (OOPC) that would translate a Smalltalk-like syntax into C. Love felt there was a market for object-oriented solutions that could coexist with legacy languages and platforms, and after a brief stint at Schlumberger-Doll, co-founded with Cox Productivity Products International (PPI), later renamed as Stepstone, to pursue this. At PPI, Cox developed OOPC into Objective-C. Cox saw Objective-C as a crucial link in his larger vision of creating a market for \"pre-fabricated\" software components (\"software-ICs\"), which could be bought off the shelf and which, Cox believed, would unleash a \"software industrial revolution.\" Steve Naroff joined Stepstone in 1986 as Steve Jobs' NeXT Computer became an important customer for Objective-C, as it was being used in its NeXTSTEP operating system. Naroff became the primary Stepstone developer addressing NeXT's issues with Objective-C, solving a key fragility problem preventing NeXT from deploying forwards-compatible object libraries. Impressed with NeXT, Naroff left Stepstone for NeXT in 1988, and once there, added Objective-C support to Richard Stallman's GNU GCC compiler, which NeXT was using as its C compiler, removing the need to use Stepstone's ObjC to C translator. Over the next several years, Naroff and others would add significant new features to Objective-C, such as \"categories,\" \"protocols,\" and the ability to mix in C++ code. When Stepstone folded in 1994, all rights to Objective-C were acquired by NeXT. This eventually transferred to Apple when NeXT was acquired by Apple in 1997. Objective-C became the basis for Apple's Mac OS X and then iOS platforms, and Naroff and others at Apple added additional features to the language in the late 2000s as the iPhone App Store greatly expanded Objective-C's user base.", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2021-01", "2021-09", "2021-12", "2022-12", "2024-11"], "last_mentioned": "2024-11-13T22:20:02Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3505170.3511478", "slug": "ca19b801c9555d0c", "title": "Scalability and Generalization of Circuit Training for Chip Floorplanning", "authors": ["Summer Yue", "Ebrahim M. Songhori", "Joe Wenjie Jiang", "Toby Boyd", "Anna Goldie", "Azalia Mirhoseini", "Sergio Guadarrama"], "year": 2022, "venue": "Proceedings of the 2022 International Symposium on Physical Design", "url": "https://doi.org/10.1145/3505170.3511478", "doi": "10.1145/3505170.3511478", "topics": ["Machine learning"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2023-04", "2024-09", "2024-10", "2024-12"], "last_mentioned": "2024-12-01T12:45:43Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3544548.3581388", "slug": "194b0531196190e7", "title": "Why Johnny Can’t Prompt: How Non-AI Experts Try (and Fail) to Design LLM Prompts", "authors": ["J.D. Zamfirescu-Pereira", "Richmond Y. Wong", "Bjoern Hartmann", "Qian Yang"], "year": 2023, "venue": "Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems", "url": "https://doi.org/10.1145/3544548.3581388", "doi": "10.1145/3544548.3581388", "topics": ["Language models"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 3, "months": ["2023-05", "2023-06", "2024-04", "2024-05", "2025-04"], "last_mentioned": "2025-04-08T15:27:42Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/800017.800542", "slug": "961af1686502c661", "title": "Efficient implementation of the smalltalk-80 system", "authors": ["L. Peter Deutsch", "Allan M. Schiffman"], "year": 1984, "venue": "Proceedings of the 11th ACM SIGACT-SIGPLAN symposium on Principles of programming languages  - POPL '84", "url": "https://doi.org/10.1145/800017.800542", "doi": "10.1145/800017.800542", "topics": ["Programming languages"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 5, "user_count": 4, "months": ["2021-06", "2021-09", "2023-01", "2024-01", "2026-04"], "last_mentioned": "2026-04-21T12:19:21Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0708.0261", "slug": "cb547d08864b2553", "title": "An Introduction to Quantum Computing", "authors": ["Noson S. Yanofsky"], "year": 2007, "venue": "arXiv", "url": "https://arxiv.org/abs/0708.0261", "doi": null, "topics": ["Physics", "Mathematics", "Algorithms"], "description": "Quantum Computing is a new and exciting field at the intersection of mathematics, computer science and physics.", "description_method": "abstract-excerpt", "abstract": "Quantum Computing is a new and exciting field at the intersection of mathematics, computer science and physics. It concerns a utilization of quantum mechanics to improve the efficiency of computation. Here we present a gentle introduction to some of the ideas in quantum computing. The paper begins by motivating the central ideas of quantum mechanics and quantum computation with simple toy models. From there we move on to a formal presentation of the small fraction of (finite dimensional) quantum mechanics that we will need for basic quantum computation. Central notions of quantum architecture (qubits and quantum gates) are described. The paper ends with a presentation of one of the simplest quantum algorithms: Deutsch's algorithm. Our presentation demands neither advanced mathematics nor advanced physics.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2015-08", "2017-06", "2020-02"], "last_mentioned": "2020-02-25T20:04:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0910.5294", "slug": "1a2cf3f982f93543", "title": "DNA Breathing Dynamics in the Presence of a Terahertz Field", "authors": ["B. S. Alexandrov", "V. Gelev", "A. R. Bishop", "A. Usheva", "K. O. Rasmussen"], "year": 2009, "venue": "arXiv", "url": "https://arxiv.org/abs/0910.5294", "doi": "10.1016/j.physleta.2009.12.077", "topics": ["Biology & medicine", "Distributed systems"], "description": "We consider the influence of a terahertz field on the breathing dynamics of double-stranded DNA.", "description_method": "abstract-excerpt", "abstract": "We consider the influence of a terahertz field on the breathing dynamics of double-stranded DNA. We model the spontaneous formation of spatially localized openings of a damped and driven DNA chain, and find that linear instabilities lead to dynamic dimerization, while true local strand separations require a threshold amplitude mechanism. Based on our results we argue that a specific terahertz radiation exposure may significantly affect the natural dynamics of DNA, and thereby influence intricate molecular processes involved in gene expression and DNA replication.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2010-11", "2013-08", "2015-09", "2024-12"], "last_mentioned": "2024-12-12T21:20:44Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1005.3035", "slug": "4804de919e4cc438", "title": "Building up spacetime with quantum entanglement", "authors": ["Mark Van Raamsdonk"], "year": 2010, "venue": "arXiv", "url": "https://arxiv.org/abs/1005.3035", "doi": "10.1007/s10714-010-1034-0", "topics": ["Physics"], "description": "In this essay based on 0907.2939, we argue that the emergence of classically connected spacetimes is intimately related to the quantum entanglement of degrees of freedom in a non-perturbative description of quantum gravity.", "description_method": "abstract-excerpt", "abstract": "In this essay based on 0907.2939, we argue that the emergence of classically connected spacetimes is intimately related to the quantum entanglement of degrees of freedom in a non-perturbative description of quantum gravity. Disentangling the degrees of freedom associated with two regions of spacetime results in these regions pulling apart and pinching off from each other in a way that can be quantified by standard measures of entanglement.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2013-07", "2019-10", "2022-11", "2024-01"], "last_mentioned": "2024-01-22T01:29:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1109.4897", "slug": "ea3cf63f7cce2be1", "title": "Measurement of the neutrino velocity with the OPERA detector in the CNGS beam", "authors": [" The OPERA Collaboration", "T. Adam", "N. Agafonova", "A. Aleksandrov", "O. Altinok", "P. Alvarez Sanchez", "A. Anokhina", "S. Aoki", "A. Ariga", "T. Ariga", "D. Autiero", "A. Badertscher", "A. Ben Dhahbi", "A. Bertolin", "C. Bozza", "T. Brugiere", "R. Brugnera", "F. Brunet", "G. Brunetti", "S. Buontempo", "B. Carlus", "F. Cavanna", "A. Cazes", "L. Chaussard", "M. Chernyavsky", "V. Chiarella", "A. Chukanov", "G. Colosimo", "M. Crespi", "N. D'Ambrosio", "G. De Lellis", "M. De Serio", "Y. Declais", "P. del Amo Sanchez", "F. Di Capua", "A. Di Crescenzo", "D. Di Ferdinando", "N. Di Marco", "S. Dmitrievsky", "M. Dracos", "D. Duchesneau", "S. Dusini", "T. Dzhatdoev", "J. Ebert", "I. Efthymiopoulos", "O. Egorov", "A. Ereditato", "L. S. Esposito", "J. Favier", "T. Ferber", "R. A. Fini", "T. Fukuda", "A. Garfagnini", "G. Giacomelli", "M. Giorgini", "M. Giovannozzi", "C. Girerd", "J. Goldberg", "C. Gollnitz", "D. Golubkov", "L. Goncharova", "Y. Gornushkin", "G. Grella", "F. Grianti", "E. Gschwendtner", "C. Guerin", "A. M. Guler", "C. Gustavino", "C. Hagner", "K. Hamada", "T. Hara", "R. Enikeev", "M. Hierholzer", "A. Hollnagel", "M. Ieva", "H. Ishida", "K. Ishiguro", "K. Jakovcic", "C. Jollet", "M. Jones", "F. Juget", "M. Kamiscioglu", "J. Kawada", "S. H. Kim", "M. Kimura", "E. Kiritsis", "N. Kitagawa", "B. Klicek", "J. Knuesel", "K. Kodama", "M. Komatsu", "U. Kose", "I. Kreslo", "C. Lazzaro", "J. Lenkeit", "A. Ljubicic", "A. Longhin", "A. Malgin", "G. Mandrioli", "J. Marteau", "T. Matsuo", "V. Matveev", "N. Mauri", "A. Mazzoni", "E. Medinaceli", "F. Meisel", "A. Meregaglia", "P. Migliozzi", "S. Mikado", "D. Missiaen", "P. Monacelli", "K. Morishima", "U. Moser", "M. T. Muciaccia", "N. Naganawa", "T. Naka", "M. Nakamura", "T. Nakano", "Y. Nakatsuka", "D. Naumov", "V. Nikitina", "F. Nitti", "S. Ogawa", "N. Okateva", "A. Olchevsky", "O. Palamara", "A. Paoloni", "B. D. Park", "I. G. Park", "A. Pastore", "L. Patrizii", "E. Pennacchio", "H. Pessard", "C. Pistillo", "N. Polukhina", "M. Pozzato", "K. Pretzl", "F. Pupilli", "R. Rescigno", "F. Riguzzi", "T. Roganova", "H. Rokujo", "G. Rosa", "I. Rostovtseva", "A. Rubbia", "A. Russo", "V. Ryasny", "O. Ryazhskaya", "O. Sato", "Y. Sato", "Z. Sahnoun", "A. Schembri", "J. Schuler", "L. Scotto Lavina", "J. Serrano", "I. Shakiryanova", "A. Sheshukov", "H. Shibuya", "G. Shoziyoev", "S. Simone", "M. Sioli", "C. Sirignano", "G. Sirri", "J. S. Song", "M. Spinetti", "L. Stanco", "N. Starkov", "S. Stellacci", "M. Stipcevic", "T. Strauss", "S. Takahashi", "M. Tenti", "F. Terranova", "I. Tezuka", "V. Tioukov", "P. Tolun", "N. T. Tran", "S. Tufanli", "P. Vilain", "M. Vladimirov", "L. Votano", "J. -L. Vuilleumier", "G. Wilquet", "B. Wonsak", "J. Wurtz", "V. Yakushev", "C. S. Yoon", "J. Yoshida", "Y. Zaitsev", "S. Zemskova", "A. Zghiche"], "year": 2011, "venue": "arXiv", "url": "https://arxiv.org/abs/1109.4897", "doi": "10.1007/JHEP10(2012)093", "topics": ["Statistics & probability", "Economics"], "description": "The OPERA neutrino experiment at the underground Gran Sasso Laboratory has measured the velocity of neutrinos from the CERN CNGS beam over a baseline of about 730 km.", "description_method": "abstract-excerpt", "abstract": "The OPERA neutrino experiment at the underground Gran Sasso Laboratory has measured the velocity of neutrinos from the CERN CNGS beam over a baseline of about 730 km. The measurement is based on data taken by OPERA in the years 2009, 2010 and 2011. Dedicated upgrades of the CNGS timing system and of the OPERA detector, as well as a high precision geodesy campaign for the measurement of the neutrino baseline, allowed reaching comparable systematic and statistical accuracies. An arrival time of CNGS muon neutrinos with respect to the one computed assuming the speed of light in vacuum of (6.5 +/- 7.4(stat.)((+8.3)(-8.0)sys.))ns was measured corresponding to a relative difference of the muon neutrino velocity with respect to the speed of light (v-c)/c =(2.7 +/-3.1(stat.)((+3.4)(-3.3)(sys.))x10^(-6). The above result, obtained by comparing the time distributions of neutrino interactions and of protons hitting the CNGS target in 10.5 microseconds long extractions, was confirmed by a test performed at the end of 2011 using a short bunch beam allowing to measure the neutrino time of flight at the single interaction level.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2011-09", "2011-10", "2012-06", "2012-07"], "last_mentioned": "2012-07-03T10:07:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1111.3328", "slug": "c3102beee80b1394", "title": "On the reality of the quantum state", "authors": ["Matthew F. Pusey", "Jonathan Barrett", "Terry Rudolph"], "year": 2011, "venue": "arXiv", "url": "https://arxiv.org/abs/1111.3328", "doi": "10.1038/nphys2309", "topics": ["Physics", "Mathematics"], "description": "Quantum states are the key mathematical objects in quantum theory.", "description_method": "abstract-excerpt", "abstract": "Quantum states are the key mathematical objects in quantum theory. It is therefore surprising that physicists have been unable to agree on what a quantum state truly represents. One possibility is that a pure quantum state corresponds directly to reality. However, there is a long history of suggestions that a quantum state (even a pure state) represents only knowledge or information about some aspect of reality. Here we show that any model in which a quantum state represents mere information about an underlying physical state of the system, and in which systems that are prepared independently have independent physical states, must make predictions which contradict those of quantum theory.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2011-11", "2016-10", "2018-01"], "last_mentioned": "2018-01-29T10:39:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1201.0749", "slug": "aee4ec6ef8c39017", "title": "There is no 16-Clue Sudoku: Solving the Sudoku Minimum Number of Clues Problem", "authors": ["Gary McGuire", "Bastian Tugemann", "Gilles Civario"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1201.0749", "doi": null, "topics": ["Algorithms"], "description": "The sudoku minimum number of clues problem is the following question: what is the smallest number of clues that a sudoku puzzle can have?", "description_method": "abstract-excerpt", "abstract": "The sudoku minimum number of clues problem is the following question: what is the smallest number of clues that a sudoku puzzle can have? For several years it had been conjectured that the answer is 17. We have performed an exhaustive computer search for 16-clue sudoku puzzles, and did not find any, thus proving that the answer is indeed 17. In this article we describe our method and the actual search. As a part of this project we developed a novel way for enumerating hitting sets. The hitting set problem is computationally hard; it is one of Karp's 21 classic NP-complete problems. A standard backtracking algorithm for finding hitting sets would not be fast enough to search for a 16-clue sudoku puzzle exhaustively, even at today's supercomputer speeds. To make an exhaustive search possible, we designed an algorithm that allowed us to efficiently enumerate hitting sets of a suitable size.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-06", "2019-06", "2023-05"], "last_mentioned": "2023-05-12T09:10:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1209.2137", "slug": "327a7a5a25f1eebf", "title": "Decoding billions of integers per second through vectorization", "authors": ["Daniel Lemire", "Leonid Boytsov"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1209.2137", "doi": "10.1002/spe.2203", "topics": ["Databases", "Computer systems"], "description": "In many important applications -- such as search engines and relational database systems -- data is stored in the form of arrays of integers.", "description_method": "abstract-excerpt", "abstract": "In many important applications -- such as search engines and relational database systems -- data is stored in the form of arrays of integers. Encoding and, most importantly, decoding of these arrays consumes considerable CPU time. Therefore, substantial effort has been made to reduce costs associated with compression and decompression. In particular, researchers have exploited the superscalar nature of modern processors and SIMD instructions. Nevertheless, we introduce a novel vectorized scheme called SIMD-BP128 that improves over previously proposed vectorized approaches. It is nearly twice as fast as the previously fastest schemes on desktop processors (varint-G8IU and PFOR). At the same time, SIMD-BP128 saves up to 2 bits per integer. For even better compression, we propose another new vectorized scheme (SIMD-FastPFOR) that has a compression ratio within 10% of a state-of-the-art scheme (Simple-8b) while being two times faster during decoding.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2012-09", "2017-02", "2021-06", "2025-10"], "last_mentioned": "2025-10-05T14:21:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1212.0703", "slug": "92aa9fb0866966e5", "title": "The Cost of Address Translation", "authors": ["Tomasz Jurkiewicz", "Kurt Mehlhorn"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1212.0703", "doi": null, "topics": ["Algorithms", "Computer systems"], "description": "Modern computers are not random access machines (RAMs).", "description_method": "abstract-excerpt", "abstract": "Modern computers are not random access machines (RAMs). They have a memory hierarchy, multiple cores, and virtual memory. In this paper, we address the computational cost of address translation in virtual memory. Starting point for our work is the observation that the analysis of some simple algorithms (random scan of an array, binary search, heapsort) in either the RAM model or the EM model (external memory model) does not correctly predict growth rates of actual running times. We propose the VAT model (virtual address translation) to account for the cost of address translations and analyze the algorithms mentioned above and others in the model. The predictions agree with the measurements. We also analyze the VAT-cost of cache-oblivious algorithms.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2015-12", "2016-08", "2017-01", "2018-05"], "last_mentioned": "2018-05-29T10:15:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1310.8343", "slug": "95d5b46f69ca54fa", "title": "Matter-wave interference with particles selected from a molecular library with masses exceeding 10000 amu", "authors": ["Sandra Eibenberger", "Stefan Gerlich", "Markus Arndt", "Marcel Mayor", "Jens Tüxen"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1310.8343", "doi": "10.1039/C3CP51500A", "topics": ["Physics", "Biology & medicine", "Algorithms"], "description": "The quantum superposition principle, a key distinction between quantum physics and classical mechanics, is often perceived as a philosophical challenge to our concepts of reality, locality or space-time since it contrasts our intuitive expectations with experimental observations on isolated quantum systems.", "description_method": "abstract-excerpt", "abstract": "The quantum superposition principle, a key distinction between quantum physics and classical mechanics, is often perceived as a philosophical challenge to our concepts of reality, locality or space-time since it contrasts our intuitive expectations with experimental observations on isolated quantum systems. While we are used to associating the notion of localization with massive bodies, quantum physics teaches us that every individual object is associated with a wave function that may eventually delocalize by far more than the body's own extension. Numerous experiments have verified this concept at the microscopic scale but intuition wavers when it comes to delocalization experiments with complex objects. While quantum science is the uncontested ideal of a physics theory, one may ask if the superposition principle can persist on all complexity scales. This motivates matter-wave diffraction and interference studies with large compounds in a three-grating interferometer configuration which also necessitates the preparation of high-mass nanoparticle beams at low velocities. Here we demonstrate how synthetic chemistry allows us to prepare libraries of fluorous porphyrins which can be tailored to exhibit high mass, good thermal stability and relatively low polarizability, which allows us to form slow thermal beams of these high-mass compounds, which can be detected in electron ionization mass spectrometry. We present successful superposition experiments with selected species from these molecular libraries in a quantum interferometer, which utilizes the diffraction of matter waves at an optical phase grating. We observe high-contrast quantum fringe patterns with molecules exceeding a mass of 10 000 amu and 810 atoms in a single particle.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-03", "2018-09", "2021-05", "2026-01"], "last_mentioned": "2026-01-07T18:28:19Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.1431", "slug": "7afedb3c888df829", "title": "Computing in Operations Research using Julia", "authors": ["Miles Lubin", "Iain Dunning"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.1431", "doi": "10.1287/ijoc.2014.0623", "topics": ["Algorithms", "Programming languages", "Mathematics"], "description": "The state of numerical computing is currently characterized by a divide between highly efficient yet typically cumbersome low-level languages such as C, C++, and Fortran and highly expressive yet typically slow high-level languages such as Python and MATLAB.", "description_method": "abstract-excerpt", "abstract": "The state of numerical computing is currently characterized by a divide between highly efficient yet typically cumbersome low-level languages such as C, C++, and Fortran and highly expressive yet typically slow high-level languages such as Python and MATLAB. This paper explores how Julia, a modern programming language for numerical computing which claims to bridge this divide by incorporating recent advances in language and compiler design (such as just-in-time compilation), can be used for implementing software and algorithms fundamental to the field of operations research, with a focus on mathematical optimization. In particular, we demonstrate algebraic modeling for linear and nonlinear optimization and a partial implementation of a practical simplex code. Extensive cross-language benchmarks suggest that Julia is capable of obtaining state-of-the-art performance.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2014-03", "2014-06", "2015-10", "2016-11"], "last_mentioned": "2016-11-09T14:06:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.6184", "slug": "c192cb966825ee02", "title": "Do Deep Nets Really Need to be Deep?", "authors": ["Lei Jimmy Ba", "Rich Caruana"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.6184", "doi": null, "topics": ["Psychology", "Computer vision", "Machine learning"], "description": "Currently, deep neural networks are the state of the art on problems such as speech recognition and computer vision.", "description_method": "abstract-excerpt", "abstract": "Currently, deep neural networks are the state of the art on problems such as speech recognition and computer vision. In this extended abstract, we show that shallow feed-forward networks can learn the complex functions previously learned by deep nets and achieve accuracies previously only achievable with deep models. Moreover, in some cases the shallow neural nets can learn these deep functions using a total number of parameters similar to the original deep model. We evaluate our method on the TIMIT phoneme recognition task and are able to train shallow fully-connected nets that perform similarly to complex, well-engineered, deep convolutional architectures. Our success in training shallow neural nets to mimic deeper models suggests that there probably exist better algorithms for training shallow feed-forward nets than those currently available.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2014-08", "2014-09", "2014-10", "2015-01"], "last_mentioned": "2015-01-15T01:08:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1404.0703", "slug": "1d2a76bf61b68686", "title": "Joins via Geometric Resolutions: Worst-case and Beyond", "authors": ["Mahmoud Abo Khamis", "Hung Q. Ngo", "Christopher Ré", "Atri Rudra"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1404.0703", "doi": null, "topics": ["Algorithms", "Mathematics", "Databases"], "description": "We present a simple geometric framework for the relational join.", "description_method": "abstract-excerpt", "abstract": "We present a simple geometric framework for the relational join. Using this framework, we design an algorithm that achieves the fractional hypertree-width bound, which generalizes classical and recent worst-case algorithmic results on computing joins. In addition, we use our framework and the same algorithm to show a series of what are colloquially known as beyond worst-case results. The framework allows us to prove results for data stored in Btrees, multidimensional data structures, and even multiple indices per table. A key idea in our framework is formalizing the inference one does with an index as a type of geometric resolution; transforming the algorithmic problem of computing joins to a geometric problem. Our notion of geometric resolution can be viewed as a geometric analog of logical resolution. In addition to the geometry and logic connections, our algorithm can also be thought of as backtracking search with memoization.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2015-01", "2019-12", "2021-01", "2022-07"], "last_mentioned": "2022-07-30T05:56:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1405.5563", "slug": "42610482b6ef94f9", "title": "Constructor Theory of Information", "authors": ["David Deutsch", "Chiara Marletto"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1405.5563", "doi": "10.1098/rspa.2014.0540", "topics": ["Physics", "Statistics & probability", "Mathematics"], "description": "We present a theory of information expressed solely in terms of which transformations of physical systems are possible and which are impossible - i.e.", "description_method": "abstract-excerpt", "abstract": "We present a theory of information expressed solely in terms of which transformations of physical systems are possible and which are impossible - i.e. in constructor-theoretic terms. Although it includes conjectured laws of physics that are directly about information, independently of the details of particular physical instantiations, it does not regard information as an a priori mathematical or logical concept, but as something whose nature and properties are determined by the laws of physics alone. It does not suffer from the circularity at the foundations of existing information theory (namely that information and distinguishability are each defined in terms of the other). It explains the relationship between classical and quantum information, and reveals the single, constructor-theoretic property underlying the most distinctive phenomena associated with the latter, including the lack of in-principle distinguishability of some states, the impossibility of cloning, the existence of pairs of variables that cannot simultaneously have sharp values, the fact that measurement processes can be both deterministic and unpredictable, the irreducible perturbation caused by measurement, and entanglement (locally inaccessible information).", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2014-05", "2014-12", "2021-04"], "last_mentioned": "2021-04-30T13:55:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1408.5882", "slug": "8647f1161160fcbc", "title": "Convolutional Neural Networks for Sentence Classification", "authors": ["Yoon Kim"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1408.5882", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "We report on a series of experiments with convolutional neural networks (CNN) trained on top of pre-trained word vectors for sentence-level classification tasks.", "description_method": "abstract-excerpt", "abstract": "We report on a series of experiments with convolutional neural networks (CNN) trained on top of pre-trained word vectors for sentence-level classification tasks. We show that a simple CNN with little hyperparameter tuning and static vectors achieves excellent results on multiple benchmarks. Learning task-specific vectors through fine-tuning offers further gains in performance. We additionally propose a simple modification to the architecture to allow for the use of both task-specific and static vectors. The CNN models discussed herein improve upon the state of the art on 4 out of 7 tasks, which include sentiment analysis and question classification.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2015-10", "2017-11", "2018-03"], "last_mentioned": "2018-03-02T17:39:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1409.1570", "slug": "318bceed6381df75", "title": "Is the quantum state real? An extended review of $ψ$-ontology theorems", "authors": ["M. S. Leifer"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1409.1570", "doi": "10.12743/quanta.v3i1.22", "topics": ["Mathematics", "Physics", "Science & research"], "description": "Towards the end of 2011, Pusey, Barrett and Rudolph derived a theorem that aimed to show that the quantum state must be ontic (a state of reality) in a broad class of realist approaches to quantum theory.", "description_method": "abstract-excerpt", "abstract": "Towards the end of 2011, Pusey, Barrett and Rudolph derived a theorem that aimed to show that the quantum state must be ontic (a state of reality) in a broad class of realist approaches to quantum theory. This result attracted a lot of attention and controversy. The aim of this review article is to review the background to the Pusey--Barrett--Rudolph Theorem, to provide a clear presentation of the theorem itself, and to review related work that has appeared since the publication of the Pusey--Barrett--Rudolph paper. In particular, this review: Explains what it means for the quantum state to be ontic or epistemic (a state of knowledge); Reviews arguments for and against an ontic interpretation of the quantum state as they existed prior to the Pusey--Barrett--Rudolph Theorem; Explains why proving the reality of the quantum state is a very strong constraint on realist theories in that it would imply many of the known no-go theorems, such as Bell's Theorem and the need for an exponentially large ontic state space; Provides a comprehensive presentation of the Pusey--Barrett--Rudolph Theorem itself, along with subsequent improvements and criticisms of its assumptions; Reviews two other arguments for the reality of the quantum state: the first due to Hardy and the second due to Colbeck and Renner, and explains why their assumptions are less compelling than those of the Pusey--Barrett--Rudolph Theorem; Reviews subsequent work aimed at ruling out stronger notions of what it means for the quantum state to be epistemic and points out open questions in this area. The overall aim is not only to provide the background needed for the novice in this area to understand the current status, but also to discuss often overlooked subtleties that should be of interest to the experts.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2019-07", "2022-10", "2024-06", "2025-06"], "last_mentioned": "2025-06-16T16:59:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1409.4842", "slug": "ce80b938638d598c", "title": "Going Deeper with Convolutions", "authors": ["Christian Szegedy", "Wei Liu", "Yangqing Jia", "Pierre Sermanet", "Scott Reed", "Dragomir Anguelov", "Dumitru Erhan", "Vincent Vanhoucke", "Andrew Rabinovich"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1409.4842", "doi": null, "topics": ["Computer vision", "Psychology", "Machine learning"], "description": "We propose a deep convolutional neural network architecture codenamed \"Inception\", which was responsible for setting the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC 2014).", "description_method": "abstract-excerpt", "abstract": "We propose a deep convolutional neural network architecture codenamed \"Inception\", which was responsible for setting the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC 2014). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. This was achieved by a carefully crafted design that allows for increasing the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC 2014 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2015-03", "2015-07", "2017-11", "2022-04"], "last_mentioned": "2022-04-11T15:00:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1410.2895", "slug": "a1a61c7ad747dbb3", "title": "Observing Ultra-High Energy Cosmic Rays with Smartphones", "authors": ["Daniel Whiteson", "Michael Mulhearn", "Chase Shimmin", "Kyle Cranmer", "Kyle Brodie", "Dustin Burns"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1410.2895", "doi": "10.1016/j.astropartphys.2016.02.002", "topics": ["Other research"], "description": "We propose a novel approach for observing cosmic rays at ultra-high energy ($>10^{18}$~eV) by repurposing the existing network of smartphones as a ground detector array.", "description_method": "abstract-excerpt", "abstract": "We propose a novel approach for observing cosmic rays at ultra-high energy ($>10^{18}$~eV) by repurposing the existing network of smartphones as a ground detector array. Extensive air showers generated by cosmic rays produce muons and high-energy photons, which can be detected by the CMOS sensors of smartphone cameras. The small size and low efficiency of each sensor is compensated by the large number of active phones. We show that if user adoption targets are met, such a network will have significant observing power at the highest energies.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2014-10", "2017-05", "2018-11", "2021-02"], "last_mentioned": "2021-02-19T13:05:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1410.4615", "slug": "f59dfb63aea9a967", "title": "Learning to Execute", "authors": ["Wojciech Zaremba", "Ilya Sutskever"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1410.4615", "doi": null, "topics": ["Machine learning"], "description": "Recurrent Neural Networks (RNNs) with Long Short-Term Memory units (LSTM) are widely used because they are expressive and are easy to train.", "description_method": "abstract-excerpt", "abstract": "Recurrent Neural Networks (RNNs) with Long Short-Term Memory units (LSTM) are widely used because they are expressive and are easy to train. Our interest lies in empirically evaluating the expressiveness and the learnability of LSTMs in the sequence-to-sequence regime by training them to evaluate short computer programs, a domain that has traditionally been seen as too complex for neural networks. We consider a simple class of programs that can be evaluated with a single left-to-right pass using constant memory. Our main result is that LSTMs can learn to map the character-level representations of such programs to their correct outputs. Notably, it was necessary to use curriculum learning, and while conventional curriculum learning proved ineffective, we developed a new variant of curriculum learning that improved our networks' performance in all experimental conditions. The improved curriculum had a dramatic impact on an addition problem, making it possible to train an LSTM to add two 9-digit numbers with 99% accuracy.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2014-11", "2015-05", "2015-06", "2017-06"], "last_mentioned": "2017-06-03T05:24:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1411.4555", "slug": "f6d8b1d9778eebd7", "title": "Show and Tell: A Neural Image Caption Generator", "authors": ["Oriol Vinyals", "Alexander Toshev", "Samy Bengio", "Dumitru Erhan"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1411.4555", "doi": null, "topics": ["Computer vision", "Psychology", "Machine learning"], "description": "Automatically describing the content of an image is a fundamental problem in artificial intelligence that connects computer vision and natural language processing.", "description_method": "abstract-excerpt", "abstract": "Automatically describing the content of an image is a fundamental problem in artificial intelligence that connects computer vision and natural language processing. In this paper, we present a generative model based on a deep recurrent architecture that combines recent advances in computer vision and machine translation and that can be used to generate natural sentences describing an image. The model is trained to maximize the likelihood of the target description sentence given the training image. Experiments on several datasets show the accuracy of the model and the fluency of the language it learns solely from image descriptions. Our model is often quite accurate, which we verify both qualitatively and quantitatively. For instance, while the current state-of-the-art BLEU-1 score (the higher the better) on the Pascal dataset is 25, our approach yields 59, to be compared to human performance around 69. We also show BLEU-1 score improvements on Flickr30k, from 56 to 66, and on SBU, from 19 to 28. Lastly, on the newly released COCO dataset, we achieve a BLEU-4 of 27.7, which is the current state-of-the-art.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2014-11", "2015-05", "2016-06"], "last_mentioned": "2016-06-10T17:34:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1412.6544", "slug": "9b1bbab89fd33e91", "title": "Qualitatively characterizing neural network optimization problems", "authors": ["Ian J. Goodfellow", "Oriol Vinyals", "Andrew M. Saxe"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1412.6544", "doi": null, "topics": ["Machine learning", "Algorithms", "Statistics & probability"], "description": "Training neural networks involves solving large-scale non-convex optimization problems.", "description_method": "abstract-excerpt", "abstract": "Training neural networks involves solving large-scale non-convex optimization problems. This task has long been believed to be extremely difficult, with fear of local minima and other obstacles motivating a variety of schemes to improve optimization, such as unsupervised pretraining. However, modern neural networks are able to achieve negligible training error on complex tasks, using only direct training with stochastic gradient descent. We introduce a simple analysis technique to look for evidence that such networks are overcoming local optima. We find that, in fact, on a straight path from initialization to solution, a variety of state of the art neural networks never encounter any significant obstacles.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2015-08", "2016-08", "2024-05"], "last_mentioned": "2024-05-07T18:35:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1412.7149", "slug": "afcb17ef741a896a", "title": "Deep Fried Convnets", "authors": ["Zichao Yang", "Marcin Moczulski", "Misha Denil", "Nando de Freitas", "Alex Smola", "Le Song", "Ziyu Wang"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1412.7149", "doi": null, "topics": ["Computer vision", "Machine learning", "Computer systems"], "description": "The fully connected layers of a deep convolutional neural network typically contain over 90% of the network parameters, and consume the majority of the memory required to store the network parameters.", "description_method": "abstract-excerpt", "abstract": "The fully connected layers of a deep convolutional neural network typically contain over 90% of the network parameters, and consume the majority of the memory required to store the network parameters. Reducing the number of parameters while preserving essentially the same predictive performance is critically important for operating deep neural networks in memory constrained environments such as GPUs or embedded devices. In this paper we show how kernel methods, in particular a single Fastfood layer, can be used to replace all fully connected layers in a deep convolutional neural network. This novel Fastfood layer is also end-to-end trainable in conjunction with convolutional layers, allowing us to combine them into a new architecture, named deep fried convolutional networks, which substantially reduces the memory footprint of convolutional networks trained on MNIST and ImageNet with no drop in predictive performance.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2015-06", "2016-10", "2017-10", "2018-02"], "last_mentioned": "2018-02-12T15:50:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.01710", "slug": "a544179a3ca3eaae", "title": "Text Understanding from Scratch", "authors": ["Xiang Zhang", "Yann LeCun"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.01710", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "This article demontrates that we can apply deep learning to text understanding from character-level inputs all the way up to abstract text concepts, using temporal convolutional networks (ConvNets).", "description_method": "abstract-excerpt", "abstract": "This article demontrates that we can apply deep learning to text understanding from character-level inputs all the way up to abstract text concepts, using temporal convolutional networks (ConvNets). We apply ConvNets to various large-scale datasets, including ontology classification, sentiment analysis, and text categorization. We show that temporal ConvNets can achieve astonishing performance without the knowledge of words, phrases, sentences and any other syntactic or semantic structures with regards to a human language. Evidence shows that our models can work for both English and Chinese.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2015-03", "2016-06", "2018-04"], "last_mentioned": "2018-04-30T17:27:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.01852", "slug": "ed489cd82e57e766", "title": "Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification", "authors": ["Kaiming He", "Xiangyu Zhang", "Shaoqing Ren", "Jian Sun"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.01852", "doi": null, "topics": ["Machine learning", "Computer vision", "Psychology"], "description": "Rectified activation units (rectifiers) are essential for state-of-the-art neural networks.", "description_method": "abstract-excerpt", "abstract": "Rectified activation units (rectifiers) are essential for state-of-the-art neural networks. In this work, we study rectifier neural networks for image classification from two aspects. First, we propose a Parametric Rectified Linear Unit (PReLU) that generalizes the traditional rectified unit. PReLU improves model fitting with nearly zero extra computational cost and little overfitting risk. Second, we derive a robust initialization method that particularly considers the rectifier nonlinearities. This method enables us to train extremely deep rectified models directly from scratch and to investigate deeper or wider network architectures. Based on our PReLU networks (PReLU-nets), we achieve 4.94% top-5 test error on the ImageNet 2012 classification dataset. This is a 26% relative improvement over the ILSVRC 2014 winner (GoogLeNet, 6.66%). To our knowledge, our result is the first to surpass human-level performance (5.1%, Russakovsky et al.) on this visual recognition challenge.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2015-02", "2017-11", "2020-05", "2023-12"], "last_mentioned": "2023-12-22T23:48:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.02206", "slug": "969cacaad18a7f95", "title": "Learning to Search Better Than Your Teacher", "authors": ["Kai-Wei Chang", "Akshay Krishnamurthy", "Alekh Agarwal", "Hal Daumé", "John Langford"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.02206", "doi": null, "topics": ["Algorithms"], "description": "Methods for learning to search for structured prediction typically imitate a reference policy, with existing theoretical guarantees demonstrating low regret compared to that reference.", "description_method": "abstract-excerpt", "abstract": "Methods for learning to search for structured prediction typically imitate a reference policy, with existing theoretical guarantees demonstrating low regret compared to that reference. This is unsatisfactory in many applications where the reference policy is suboptimal and the goal of learning is to improve upon it. Can learning to search work even when the reference is poor? We provide a new learning to search algorithm, LOLS, which does well relative to the reference policy, but additionally guarantees low regret compared to deviations from the learned policy: a local-optimality guarantee. Consequently, LOLS can improve upon the reference policy, unlike previous algorithms. This enables us to develop structured contextual bandits, a partial information structured prediction setting with many potential applications.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2016-03", "2016-05", "2016-06"], "last_mentioned": "2016-06-15T21:22:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.03167", "slug": "8819ad4e6e250b73", "title": "Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift", "authors": ["Sergey Ioffe", "Christian Szegedy"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.03167", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "Training Deep Neural Networks is complicated by the fact that the distribution of each layer's inputs changes during training, as the parameters of the previous layers change.", "description_method": "abstract-excerpt", "abstract": "Training Deep Neural Networks is complicated by the fact that the distribution of each layer's inputs changes during training, as the parameters of the previous layers change. This slows down the training by requiring lower learning rates and careful parameter initialization, and makes it notoriously hard to train models with saturating nonlinearities. We refer to this phenomenon as internal covariate shift, and address the problem by normalizing layer inputs. Our method draws its strength from making normalization a part of the model architecture and performing the normalization for each training mini-batch. Batch Normalization allows us to use much higher learning rates and be less careful about initialization. It also acts as a regularizer, in some cases eliminating the need for Dropout. Applied to a state-of-the-art image classification model, Batch Normalization achieves the same accuracy with 14 times fewer training steps, and beats the original model by a significant margin. Using an ensemble of batch-normalized networks, we improve upon the best published result on ImageNet classification: reaching 4.9% top-5 validation error (and 4.8% test error), exceeding the accuracy of human raters.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2015-05", "2016-04", "2016-08", "2017-03"], "last_mentioned": "2017-03-02T18:15:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.04573", "slug": "b683968f8096b28b", "title": "Undecidability of the Spectral Gap (full version)", "authors": ["Toby Cubitt", "David Perez-Garcia", "Michael M. Wolf"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.04573", "doi": "10.1017/fmp.2021.15", "topics": ["Physics"], "description": "We show that the spectral gap problem is undecidable.", "description_method": "abstract-excerpt", "abstract": "We show that the spectral gap problem is undecidable. Specifically, we construct families of translationally-invariant, nearest-neighbour Hamiltonians on a 2D square lattice of d-level quantum systems (d constant), for which determining whether the system is gapped or gapless is an undecidable problem. This is true even with the promise that each Hamiltonian is either gapped or gapless in the strongest sense: it is promised to either have continuous spectrum above the ground state in the thermodynamic limit, or its spectral gap is lower-bounded by a constant in the thermodynamic limit. Moreover, this constant can be taken equal to the local interaction strength of the Hamiltonian.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2015-12", "2016-06", "2023-11", "2025-03"], "last_mentioned": "2025-03-08T13:38:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1505.05571", "slug": "3efa68315cd6c6a7", "title": "Fast exact summation using small and large superaccumulators", "authors": ["Radford M. Neal"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1505.05571", "doi": null, "topics": ["Computer systems"], "description": "I present two new methods for exactly summing a set of floating-point numbers, and then correctly rounding to the nearest floating-point number.", "description_method": "abstract-excerpt", "abstract": "I present two new methods for exactly summing a set of floating-point numbers, and then correctly rounding to the nearest floating-point number. Higher accuracy than simple summation (rounding after each addition) is important in many applications, such as finding the sample mean of data. Exact summation also guarantees identical results with parallel and serial implementations, since the exact sum is independent of order. The new methods use variations on the concept of a \"superaccumulator\" - a large fixed-point number that can exactly represent the sum of any reasonable number of floating-point values. One method uses a \"small\" superaccumulator with sixty-seven 64-bit chunks, each with 32-bit overlap with the next chunk, allowing carry propagation to be done infrequently. The small superaccumulator is used alone when summing a small number of terms. For big summations, a \"large\" superaccumulator is used as well. It consists of 4096 64-bit chunks, one for every possible combination of exponent bits and sign bit, plus counts of when each chunk needs to be transferred to the small superaccumulator. To add a term to the large superaccumulator, only a single chunk and its associated count need to be updated, which takes very few instructions if carefully implemented. On modern 64-bit processors, exactly summing a large array using this combination of large and small superaccumulators takes less than twice the time of simple, inexact, ordered summation, with a serial implementation. A parallel implementation using a small number of processor cores can be expected to perform exact summation of large arrays at a speed that reaches the limit imposed by memory bandwidth. Some common methods that attempt to improve accuracy without being exact may therefore be pointless, at least for large summations, since they are slower than computing the sum exactly.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2019-03", "2020-05", "2024-05", "2024-11"], "last_mentioned": "2024-11-04T00:14:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1506.02078", "slug": "803312daa9a07393", "title": "Visualizing and Understanding Recurrent Networks", "authors": ["Andrej Karpathy", "Justin Johnson", "Li Fei-Fei"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1506.02078", "doi": null, "topics": ["Machine learning", "Language models", "Biology & medicine"], "description": "Recurrent Neural Networks (RNNs), and specifically a variant with Long Short-Term Memory (LSTM), are enjoying renewed interest as a result of successful applications in a wide range of machine learning problems that involve sequential data.", "description_method": "abstract-excerpt", "abstract": "Recurrent Neural Networks (RNNs), and specifically a variant with Long Short-Term Memory (LSTM), are enjoying renewed interest as a result of successful applications in a wide range of machine learning problems that involve sequential data. However, while LSTMs provide exceptional results in practice, the source of their performance and their limitations remain rather poorly understood. Using character-level language models as an interpretable testbed, we aim to bridge this gap by providing an analysis of their representations, predictions and error types. In particular, our experiments reveal the existence of interpretable cells that keep track of long-range dependencies such as line lengths, quotes and brackets. Moreover, our comparative analysis with finite horizon n-gram models traces the source of the LSTM improvements to long-range structural dependencies. Finally, we provide analysis of the remaining errors and suggests areas for further study.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-07", "2019-07", "2021-01", "2024-02"], "last_mentioned": "2024-02-14T14:24:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1506.03340", "slug": "2d7c7812d849801a", "title": "Teaching Machines to Read and Comprehend", "authors": ["Karl Moritz Hermann", "Tomáš Kočiský", "Edward Grefenstette", "Lasse Espeholt", "Will Kay", "Mustafa Suleyman", "Phil Blunsom"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1506.03340", "doi": null, "topics": ["Machine learning"], "description": "Teaching machines to read natural language documents remains an elusive challenge.", "description_method": "abstract-excerpt", "abstract": "Teaching machines to read natural language documents remains an elusive challenge. Machine reading systems can be tested on their ability to answer questions posed on the contents of documents that they have seen, but until now large scale training and test datasets have been missing for this type of evaluation. In this work we define a new methodology that resolves this bottleneck and provides large scale supervised reading comprehension data. This allows us to develop a class of attention based deep neural networks that learn to read real documents and answer complex questions with minimal prior knowledge of language structure.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2015-06", "2015-12", "2016-02", "2016-05"], "last_mentioned": "2016-05-23T05:37:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1506.06726", "slug": "bb5720252bc8022f", "title": "Skip-Thought Vectors", "authors": ["Ryan Kiros", "Yukun Zhu", "Ruslan Salakhutdinov", "Richard S. Zemel", "Antonio Torralba", "Raquel Urtasun", "Sanja Fidler"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1506.06726", "doi": null, "topics": ["Machine learning", "Distributed systems"], "description": "We describe an approach for unsupervised learning of a generic, distributed sentence encoder.", "description_method": "abstract-excerpt", "abstract": "We describe an approach for unsupervised learning of a generic, distributed sentence encoder. Using the continuity of text from books, we train an encoder-decoder model that tries to reconstruct the surrounding sentences of an encoded passage. Sentences that share semantic and syntactic properties are thus mapped to similar vector representations. We next introduce a simple vocabulary expansion method to encode words that were not seen as part of training, allowing us to expand our vocabulary to a million words. After training our model, we extract and evaluate our vectors with linear models on 8 tasks: semantic relatedness, paraphrase detection, image-sentence ranking, question-type classification and 4 benchmark sentiment and subjectivity datasets. The end result is an off-the-shelf encoder that can produce highly generic sentence representations that are robust and perform well in practice. We will make our encoder publicly available.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2015-08", "2015-09", "2015-11", "2016-03"], "last_mentioned": "2016-03-23T23:12:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1507.07998", "slug": "b281f87fc0da7206", "title": "Document Embedding with Paragraph Vectors", "authors": ["Andrew M. Dai", "Christopher Olah", "Quoc V. Le"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1507.07998", "doi": null, "topics": ["Language models", "Distributed systems", "Algorithms"], "description": "Paragraph Vectors has been recently proposed as an unsupervised method for learning distributed representations for pieces of texts.", "description_method": "abstract-excerpt", "abstract": "Paragraph Vectors has been recently proposed as an unsupervised method for learning distributed representations for pieces of texts. In their work, the authors showed that the method can learn an embedding of movie review texts which can be leveraged for sentiment analysis. That proof of concept, while encouraging, was rather narrow. Here we consider tasks other than sentiment analysis, provide a more thorough comparison of Paragraph Vectors to other document modelling algorithms such as Latent Dirichlet Allocation, and evaluate performance of the method as we vary the dimensionality of the learned representation. We benchmarked the models on two document similarity data sets, one from Wikipedia, one from arXiv. We observe that the Paragraph Vector method performs significantly better than other methods, and propose a simple improvement to enhance embedding quality. Somewhat surprisingly, we also show that much like word embeddings, vector operations on Paragraph Vectors can perform useful semantic results.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2015-08", "2016-12", "2025-11"], "last_mentioned": "2025-11-02T00:33:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1508.07909", "slug": "a6c67b9777d7054d", "title": "Neural Machine Translation of Rare Words with Subword Units", "authors": ["Rico Sennrich", "Barry Haddow", "Alexandra Birch"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1508.07909", "doi": null, "topics": ["Computer vision", "Algorithms"], "description": "Neural machine translation (NMT) models typically operate with a fixed vocabulary, but translation is an open-vocabulary problem.", "description_method": "abstract-excerpt", "abstract": "Neural machine translation (NMT) models typically operate with a fixed vocabulary, but translation is an open-vocabulary problem. Previous work addresses the translation of out-of-vocabulary words by backing off to a dictionary. In this paper, we introduce a simpler and more effective approach, making the NMT model capable of open-vocabulary translation by encoding rare and unknown words as sequences of subword units. This is based on the intuition that various word classes are translatable via smaller units than words, for instance names (via character copying or transliteration), compounds (via compositional translation), and cognates and loanwords (via phonological and morphological transformations). We discuss the suitability of different word segmentation techniques, including simple character n-gram models and a segmentation based on the byte pair encoding compression algorithm, and empirically show that subword models improve over a back-off dictionary baseline for the WMT 15 translation tasks English-German and English-Russian by 1.1 and 1.3 BLEU, respectively.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-04", "2023-05", "2025-05", "2026-04"], "last_mentioned": "2026-04-20T07:42:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1509.02971", "slug": "2d21f6dd827a23ce", "title": "Continuous control with deep reinforcement learning", "authors": ["Timothy P. Lillicrap", "Jonathan J. Hunt", "Alexander Pritzel", "Nicolas Heess", "Tom Erez", "Yuval Tassa", "David Silver", "Daan Wierstra"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1509.02971", "doi": null, "topics": ["Algorithms", "Machine learning", "Robotics"], "description": "We adapt the ideas underlying the success of Deep Q-Learning to the continuous action domain.", "description_method": "abstract-excerpt", "abstract": "We adapt the ideas underlying the success of Deep Q-Learning to the continuous action domain. We present an actor-critic, model-free algorithm based on the deterministic policy gradient that can operate over continuous action spaces. Using the same learning algorithm, network architecture and hyper-parameters, our algorithm robustly solves more than 20 simulated physics tasks, including classic problems such as cartpole swing-up, dexterous manipulation, legged locomotion and car driving. Our algorithm is able to find policies whose performance is competitive with those found by a planning algorithm with full access to the dynamics of the domain and its derivatives. We further demonstrate that for many of the tasks the algorithm can learn policies end-to-end: directly from raw pixel inputs.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2015-10", "2015-12", "2016-03", "2020-12"], "last_mentioned": "2020-12-01T18:30:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1512.00567", "slug": "73e870303f81875d", "title": "Rethinking the Inception Architecture for Computer Vision", "authors": ["Christian Szegedy", "Vincent Vanhoucke", "Sergey Ioffe", "Jonathon Shlens", "Zbigniew Wojna"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1512.00567", "doi": null, "topics": ["Computer vision"], "description": "Convolutional networks are at the core of most state-of-the-art computer vision solutions for a wide variety of tasks.", "description_method": "abstract-excerpt", "abstract": "Convolutional networks are at the core of most state-of-the-art computer vision solutions for a wide variety of tasks. Since 2014 very deep convolutional networks started to become mainstream, yielding substantial gains in various benchmarks. Although increased model size and computational cost tend to translate to immediate quality gains for most tasks (as long as enough labeled data is provided for training), computational efficiency and low parameter count are still enabling factors for various use cases such as mobile vision and big-data scenarios. Here we explore ways to scale up networks in ways that aim at utilizing the added computation as efficiently as possible by suitably factorized convolutions and aggressive regularization. We benchmark our methods on the ILSVRC 2012 classification challenge validation set demonstrate substantial gains over the state of the art: 21.2% top-1 and 5.6% top-5 error for single frame evaluation using a network with a computational cost of 5 billion multiply-adds per inference and with using less than 25 million parameters. With an ensemble of 4 models and multi-crop evaluation, we report 3.5% top-5 error on the validation set (3.6% error on the test set) and 17.3% top-1 error on the validation set.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2016-01", "2016-07", "2016-08", "2017-04"], "last_mentioned": "2017-04-05T17:37:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1512.02206", "slug": "87878d78c001fdc9", "title": "What is the Computational Value of Finite Range Tunneling?", "authors": ["Vasil S. Denchev", "Sergio Boixo", "Sergei V. Isakov", "Nan Ding", "Ryan Babbush", "Vadim Smelyanskiy", "John Martinis", "Hartmut Neven"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1512.02206", "doi": "10.1103/PhysRevX.6.031015", "topics": ["Physics", "Algorithms", "Statistics & probability"], "description": "Quantum annealing (QA) has been proposed as a quantum enhanced optimization heuristic exploiting tunneling.", "description_method": "abstract-excerpt", "abstract": "Quantum annealing (QA) has been proposed as a quantum enhanced optimization heuristic exploiting tunneling. Here, we demonstrate how finite range tunneling can provide considerable computational advantage. For a crafted problem designed to have tall and narrow energy barriers separating local minima, the D-Wave 2X quantum annealer achieves significant runtime advantages relative to Simulated Annealing (SA). For instances with 945 variables, this results in a time-to-99%-success-probability that is $\\sim 10^8$ times faster than SA running on a single processor core. We also compared physical QA with Quantum Monte Carlo (QMC), an algorithm that emulates quantum tunneling on classical processors. We observe a substantial constant overhead against physical QA: D-Wave 2X again runs up to $\\sim 10^8$ times faster than an optimized implementation of QMC on a single core. We note that there exist heuristic classical algorithms that can solve most instances of Chimera structured problems in a timescale comparable to the D-Wave 2X. However, we believe that such solvers will become ineffective for the next generation of annealers currently being designed. To investigate whether finite range tunneling will also confer an advantage for problems of practical interest, we conduct numerical studies on binary optimization problems that cannot yet be represented on quantum hardware. For random instances of the number partitioning problem, we find numerically that QMC, as well as other algorithms designed to simulate QA, scale better than SA. We discuss the implications of these findings for the design of next generation quantum annealers.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2015-12", "2016-05", "2016-09"], "last_mentioned": "2016-09-24T22:48:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1512.02595", "slug": "07b0a502869cc388", "title": "Deep Speech 2: End-to-End Speech Recognition in English and Mandarin", "authors": ["Dario Amodei", "Rishita Anubhai", "Eric Battenberg", "Carl Case", "Jared Casper", "Bryan Catanzaro", "Jingdong Chen", "Mike Chrzanowski", "Adam Coates", "Greg Diamos", "Erich Elsen", "Jesse Engel", "Linxi Fan", "Christopher Fougner", "Tony Han", "Awni Hannun", "Billy Jun", "Patrick LeGresley", "Libby Lin", "Sharan Narang", "Andrew Ng", "Sherjil Ozair", "Ryan Prenger", "Jonathan Raiman", "Sanjeev Satheesh", "David Seetapun", "Shubho Sengupta", "Yi Wang", "Zhiqian Wang", "Chong Wang", "Bo Xiao", "Dani Yogatama", "Jun Zhan", "Zhenyao Zhu"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1512.02595", "doi": null, "topics": ["Psychology", "Machine learning", "Algorithms"], "description": "We show that an end-to-end deep learning approach can be used to recognize either English or Mandarin Chinese speech--two vastly different languages.", "description_method": "abstract-excerpt", "abstract": "We show that an end-to-end deep learning approach can be used to recognize either English or Mandarin Chinese speech--two vastly different languages. Because it replaces entire pipelines of hand-engineered components with neural networks, end-to-end learning allows us to handle a diverse variety of speech including noisy environments, accents and different languages. Key to our approach is our application of HPC techniques, resulting in a 7x speedup over our previous system. Because of this efficiency, experiments that previously took weeks now run in days. This enables us to iterate more quickly to identify superior architectures and algorithms. As a result, in several cases, our system is competitive with the transcription of human workers when benchmarked on standard datasets. Finally, using a technique called Batch Dispatch with GPUs in the data center, we show that our system can be inexpensively deployed in an online setting, delivering low latency when serving users at scale.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2019-12", "2025-12", "2026-07"], "last_mentioned": "2026-07-07T19:03:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1602.02410", "slug": "8dbd32d9fb7f22eb", "title": "Exploring the Limits of Language Modeling", "authors": ["Rafal Jozefowicz", "Oriol Vinyals", "Mike Schuster", "Noam Shazeer", "Yonghui Wu"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1602.02410", "doi": null, "topics": ["Language models", "Machine learning", "Computer vision"], "description": "In this work we explore recent advances in Recurrent Neural Networks for large scale Language Modeling, a task central to language understanding.", "description_method": "abstract-excerpt", "abstract": "In this work we explore recent advances in Recurrent Neural Networks for large scale Language Modeling, a task central to language understanding. We extend current models to deal with two key challenges present in this task: corpora and vocabulary sizes, and complex, long term structure of language. We perform an exhaustive study on techniques such as character Convolutional Neural Networks or Long-Short Term Memory, on the One Billion Word Benchmark. Our best single model significantly improves state-of-the-art perplexity from 51.3 down to 30.0 (whilst reducing the number of parameters by a factor of 20), while an ensemble of models sets a new record by improving perplexity from 41.0 down to 23.7. We also release these models for the NLP and ML community to study and improve upon.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2016-02", "2016-09", "2017-02", "2017-06"], "last_mentioned": "2017-06-12T07:46:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1602.02830", "slug": "1a54dcd5f7ce008f", "title": "Binarized Neural Networks: Training Deep Neural Networks with Weights and Activations Constrained to +1 or -1", "authors": ["Matthieu Courbariaux", "Itay Hubara", "Daniel Soudry", "Ran El-Yaniv", "Yoshua Bengio"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1602.02830", "doi": null, "topics": ["Machine learning", "Computer systems"], "description": "We introduce a method to train Binarized Neural Networks (BNNs) - neural networks with binary weights and activations at run-time.", "description_method": "abstract-excerpt", "abstract": "We introduce a method to train Binarized Neural Networks (BNNs) - neural networks with binary weights and activations at run-time. At training-time the binary weights and activations are used for computing the parameters gradients. During the forward pass, BNNs drastically reduce memory size and accesses, and replace most arithmetic operations with bit-wise operations, which is expected to substantially improve power-efficiency. To validate the effectiveness of BNNs we conduct two sets of experiments on the Torch7 and Theano frameworks. On both, BNNs achieved nearly state-of-the-art results over the MNIST, CIFAR-10 and SVHN datasets. Last but not least, we wrote a binary matrix multiplication GPU kernel with which it is possible to run our MNIST BNN 7 times faster than with an unoptimized GPU kernel, without suffering any loss in classification accuracy. The code for training and running our BNNs is available on-line.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2016-09", "2017-02", "2019-11", "2024-02"], "last_mentioned": "2024-02-28T19:23:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1602.05629", "slug": "d4d96c6dfad21c23", "title": "Communication-Efficient Learning of Deep Networks from Decentralized Data", "authors": ["H. Brendan McMahan", "Eider Moore", "Daniel Ramage", "Seth Hampson", "Blaise Agüera y Arcas"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1602.05629", "doi": null, "topics": ["Statistics & probability", "Security & cryptography", "Psychology"], "description": "Modern mobile devices have access to a wealth of data suitable for learning models, which in turn can greatly improve the user experience on the device.", "description_method": "abstract-excerpt", "abstract": "Modern mobile devices have access to a wealth of data suitable for learning models, which in turn can greatly improve the user experience on the device. For example, language models can improve speech recognition and text entry, and image models can automatically select good photos. However, this rich data is often privacy sensitive, large in quantity, or both, which may preclude logging to the data center and training there using conventional approaches. We advocate an alternative that leaves the training data distributed on the mobile devices, and learns a shared model by aggregating locally-computed updates. We term this decentralized approach Federated Learning. We present a practical method for the federated learning of deep networks based on iterative model averaging, and conduct an extensive empirical evaluation, considering five different model architectures and four datasets. These experiments demonstrate the approach is robust to the unbalanced and non-IID data distributions that are a defining characteristic of this setting. Communication costs are the principal constraint, and we show a reduction in required communication rounds by 10-100x as compared to synchronized stochastic gradient descent.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-04", "2020-12", "2023-04"], "last_mentioned": "2023-04-26T16:52:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1603.06560", "slug": "4d668fd0db063fc9", "title": "Hyperband: A Novel Bandit-Based Approach to Hyperparameter Optimization", "authors": ["Lisha Li", "Kevin Jamieson", "Giulia DeSalvo", "Afshin Rostamizadeh", "Ameet Talwalkar"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1603.06560", "doi": null, "topics": ["Algorithms", "Statistics & probability", "Machine learning"], "description": "Performance of machine learning algorithms depends critically on identifying a good set of hyperparameters.", "description_method": "abstract-excerpt", "abstract": "Performance of machine learning algorithms depends critically on identifying a good set of hyperparameters. While recent approaches use Bayesian optimization to adaptively select configurations, we focus on speeding up random search through adaptive resource allocation and early-stopping. We formulate hyperparameter optimization as a pure-exploration non-stochastic infinite-armed bandit problem where a predefined resource like iterations, data samples, or features is allocated to randomly sampled configurations. We introduce a novel algorithm, Hyperband, for this framework and analyze its theoretical properties, providing several desirable guarantees. Furthermore, we compare Hyperband with popular Bayesian optimization methods on a suite of hyperparameter optimization problems. We observe that Hyperband can provide over an order-of-magnitude speedup over our competitor set on a variety of deep-learning and kernel-based learning problems.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2017-11", "2018-02", "2018-07", "2023-06"], "last_mentioned": "2023-06-06T15:09:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1603.08155", "slug": "4533d5213c6b30a5", "title": "Perceptual Losses for Real-Time Style Transfer and Super-Resolution", "authors": ["Justin Johnson", "Alexandre Alahi", "Li Fei-Fei"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1603.08155", "doi": null, "topics": ["Algorithms", "Machine learning", "Computer vision"], "description": "We consider image transformation problems, where an input image is transformed into an output image.", "description_method": "abstract-excerpt", "abstract": "We consider image transformation problems, where an input image is transformed into an output image. Recent methods for such problems typically train feed-forward convolutional neural networks using a \\emph{per-pixel} loss between the output and ground-truth images. Parallel work has shown that high-quality images can be generated by defining and optimizing \\emph{perceptual} loss functions based on high-level features extracted from pretrained networks. We combine the benefits of both approaches, and propose the use of perceptual loss functions for training feed-forward networks for image transformation tasks. We show results on image style transfer, where a feed-forward network is trained to solve the optimization problem proposed by Gatys et al in real-time. Compared to the optimization-based method, our network gives similar qualitative results but is three orders of magnitude faster. We also experiment with single-image super-resolution, where replacing a per-pixel loss with a perceptual loss gives visually pleasing results.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2016-07", "2017-03", "2017-07", "2017-10"], "last_mentioned": "2017-10-02T15:59:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1604.00575", "slug": "196de904a0c1d1ac", "title": "AsicBoost - A Speedup for Bitcoin Mining", "authors": ["Timo Hanke"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1604.00575", "doi": null, "topics": ["Algorithms"], "description": "AsicBoost is a method to speed up Bitcoin mining by a factor of approximately 20%.", "description_method": "abstract-excerpt", "abstract": "AsicBoost is a method to speed up Bitcoin mining by a factor of approximately 20%. The performance gain is achieved through a high-level optimization of the Bitcoin mining algorithm which allows for drastic reduction in gate count on the mining chip. AsicBoost is applicable to all types of mining hardware and chip designs. This paper presents the idea behind the method and describes the information flow in implementations of AsicBoost.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2018-02", "2022-09", "2024-08", "2026-07"], "last_mentioned": "2026-07-28T21:00:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1605.06640", "slug": "11bfa0729d405fd6", "title": "Programming with a Differentiable Forth Interpreter", "authors": ["Matko Bošnjak", "Tim Rocktäschel", "Jason Naradowsky", "Sebastian Riedel"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1605.06640", "doi": null, "topics": ["Psychology", "Programming languages", "Machine learning"], "description": "Given that in practice training data is scarce for all but a small set of problems, a core question is how to incorporate prior knowledge into a model.", "description_method": "abstract-excerpt", "abstract": "Given that in practice training data is scarce for all but a small set of problems, a core question is how to incorporate prior knowledge into a model. In this paper, we consider the case of prior procedural knowledge for neural networks, such as knowing how a program should traverse a sequence, but not what local actions should be performed at each step. To this end, we present an end-to-end differentiable interpreter for the programming language Forth which enables programmers to write program sketches with slots that can be filled with behaviour trained from program input-output data. We can optimise this behaviour directly through gradient descent techniques on user-specified objectives, and also integrate the program into any larger neural computation graph. We show empirically that our interpreter is able to effectively leverage different levels of prior program structure and learn complex behaviours such as sequence sorting and addition. When connected to outputs of an LSTM and trained jointly, our interpreter achieves state-of-the-art accuracy for end-to-end reasoning about quantities expressed in natural language stories.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2016-07", "2018-01", "2022-08"], "last_mentioned": "2022-08-01T05:50:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1605.08803", "slug": "85279264fd7ad06b", "title": "Density estimation using Real NVP", "authors": ["Laurent Dinh", "Jascha Sohl-Dickstein", "Samy Bengio"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1605.08803", "doi": null, "topics": ["Machine learning", "Robotics", "Algorithms"], "description": "Unsupervised learning of probabilistic models is a central yet challenging problem in machine learning.", "description_method": "abstract-excerpt", "abstract": "Unsupervised learning of probabilistic models is a central yet challenging problem in machine learning. Specifically, designing models with tractable learning, sampling, inference and evaluation is crucial in solving this task. We extend the space of such models using real-valued non-volume preserving (real NVP) transformations, a set of powerful invertible and learnable transformations, resulting in an unsupervised learning algorithm with exact log-likelihood computation, exact sampling, exact inference of latent variables, and an interpretable latent space. We demonstrate its ability to model natural images on four datasets through sampling, log-likelihood evaluation and latent variable manipulations.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2016-06", "2018-10", "2019-01", "2020-04"], "last_mentioned": "2020-04-05T06:42:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1606.06160", "slug": "68e5fc8f5de6290d", "title": "DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients", "authors": ["Shuchang Zhou", "Yuxin Wu", "Zekun Ni", "Xinyu Zhou", "He Wen", "Yuheng Zou"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1606.06160", "doi": null, "topics": ["Machine learning", "Computer vision", "Statistics & probability"], "description": "We propose DoReFa-Net, a method to train convolutional neural networks that have low bitwidth weights and activations using low bitwidth parameter gradients.", "description_method": "abstract-excerpt", "abstract": "We propose DoReFa-Net, a method to train convolutional neural networks that have low bitwidth weights and activations using low bitwidth parameter gradients. In particular, during backward pass, parameter gradients are stochastically quantized to low bitwidth numbers before being propagated to convolutional layers. As convolutions during forward/backward passes can now operate on low bitwidth weights and activations/gradients respectively, DoReFa-Net can use bit convolution kernels to accelerate both training and inference. Moreover, as bit convolutions can be efficiently implemented on CPU, FPGA, ASIC and GPU, DoReFa-Net opens the way to accelerate training of low bitwidth neural network on these hardware. Our experiments on SVHN and ImageNet datasets prove that DoReFa-Net can achieve comparable prediction accuracy as 32-bit counterparts. For example, a DoReFa-Net derived from AlexNet that has 1-bit weights, 2-bit activations, can be trained from scratch using 6-bit gradients to get 46.1\\% top-1 accuracy on ImageNet validation set. The DoReFa-Net AlexNet model is released publicly.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2017-02", "2017-04", "2024-02"], "last_mentioned": "2024-02-29T08:09:33Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1608.04428", "slug": "097ed6e059cc8e41", "title": "TerpreT: A Probabilistic Programming Language for Program Induction", "authors": ["Alexander L. Gaunt", "Marc Brockschmidt", "Rishabh Singh", "Nate Kushman", "Pushmeet Kohli", "Jonathan Taylor", "Daniel Tarlow"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1608.04428", "doi": null, "topics": ["Machine learning", "Programming languages", "Algorithms"], "description": "We study machine learning formulations of inductive program synthesis; given input-output examples, we try to synthesize source code that maps inputs to corresponding outputs.", "description_method": "abstract-excerpt", "abstract": "We study machine learning formulations of inductive program synthesis; given input-output examples, we try to synthesize source code that maps inputs to corresponding outputs. Our aims are to develop new machine learning approaches based on neural networks and graphical models, and to understand the capabilities of machine learning techniques relative to traditional alternatives, such as those based on constraint solving from the programming languages community. Our key contribution is the proposal of TerpreT, a domain-specific language for expressing program synthesis problems. TerpreT is similar to a probabilistic programming language: a model is composed of a specification of a program representation (declarations of random variables) and an interpreter describing how programs map inputs to outputs (a model connecting unknowns to observations). The inference task is to observe a set of input-output examples and infer the underlying program. TerpreT has two main benefits. First, it enables rapid exploration of a range of domains, program representations, and interpreter models. Second, it separates the model specification from the inference algorithm, allowing like-to-like comparisons between different approaches to inference. From a single TerpreT specification we automatically perform inference using four different back-ends. These are based on gradient descent, linear program (LP) relaxations for graphical models, discrete satisfiability solving, and the Sketch program synthesis system. We illustrate the value of TerpreT by developing several interpreter models and performing an empirical comparison between alternative inference algorithms. Our key empirical finding is that constraint solvers dominate the gradient descent and LP-based formulations. We conclude with suggestions for the machine learning community to make progress on program synthesis.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-01", "2021-09", "2022-06", "2024-05"], "last_mentioned": "2024-05-26T19:46:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.03543", "slug": "908e7d2bdb3bb14e", "title": "Logical Induction", "authors": ["Scott Garrabrant", "Tsvi Benson-Tilsen", "Andrew Critch", "Nate Soares", "Jessica Taylor"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.03543", "doi": null, "topics": ["Statistics & probability", "Algorithms"], "description": "We present a computable algorithm that assigns probabilities to every logical statement in a given formal language, and refines those probabilities over time.", "description_method": "abstract-excerpt", "abstract": "We present a computable algorithm that assigns probabilities to every logical statement in a given formal language, and refines those probabilities over time. For instance, if the language is Peano arithmetic, it assigns probabilities to all arithmetical statements, including claims about the twin prime conjecture, the outputs of long-running computations, and its own probabilities. We show that our algorithm, an instance of what we call a logical inductor, satisfies a number of intuitive desiderata, including: (1) it learns to predict patterns of truth and falsehood in logical statements, often long before having the resources to evaluate the statements, so long as the patterns can be written down in polynomial time; (2) it learns to use appropriate statistical summaries to predict sequences of statements whose truth values appear pseudorandom; and (3) it learns to have accurate beliefs about its own current beliefs, in a manner that avoids the standard paradoxes of self-reference. For example, if a given computer program only ever produces outputs in a certain range, a logical inductor learns this fact in a timely manner; and if late digits in the decimal expansion of $π$ are difficult to predict, then a logical inductor learns to assign $\\approx 10\\%$ probability to \"the $n$th digit of $π$ is a 7\" for large $n$. Logical inductors also learn to trust their future beliefs more than their current beliefs, and their beliefs are coherent in the limit (whenever $φ\\implies ψ$, $\\mathbb{P}_\\infty(φ) \\le \\mathbb{P}_\\infty(ψ)$, and so on); and logical inductors strictly dominate the universal semimeasure in the limit. These properties and many others all follow from a single logical induction criterion, which is motivated by a series of stock trading analogies. Roughly speaking, each logical sentence $φ$ is associated with a stock that is worth \\$1 per share if [...]", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2019-01", "2023-04", "2024-11", "2026-07"], "last_mentioned": "2026-07-29T09:43:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1612.00676", "slug": "c7064667fea09f2b", "title": "Surveying the Attitudes of Physicists Concerning Foundational Issues of Quantum Mechanics", "authors": ["Sujeevan Sivasundaram", "Kristian Hvidtfelt Nielsen"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1612.00676", "doi": null, "topics": ["Physics"], "description": "Even though quantum mechanics has existed for almost 100 years, questions concerning the foundation and interpretation of the theory still remain.", "description_method": "abstract-excerpt", "abstract": "Even though quantum mechanics has existed for almost 100 years, questions concerning the foundation and interpretation of the theory still remain. These issues have gathered more attention in recent years, but does this mean that physicists are more aware of foundational issues concerning quantum mechanics? A survey was sent out to 1234 physicists affiliated to 8 different universities. 149 responded to the questions, which both concerned foundational issues related to quantum mechanics, specifically, as well as questions concerning interpretations of physical theories in general. The answers to the survey revealed that foundational concepts in quantum mechanics are still a topic that only a minority of physicists are familiar with, although a clear majority of physicists find that interpretations of physical theories are important. The various questions, as well as how the respondents answered, are presented. The survey intends to give an overview of what the opinion of the physics community, in general, is concerning issues related to quantum mechanics.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-04", "2018-09", "2019-02", "2025-02"], "last_mentioned": "2025-02-05T15:07:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1702.08719", "slug": "fa42694122318f61", "title": "Malware Guard Extension: Using SGX to Conceal Cache Attacks", "authors": ["Michael Schwarz", "Samuel Weiser", "Daniel Gruss", "Clémentine Maurice", "Stefan Mangard"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1702.08719", "doi": null, "topics": ["Security & cryptography", "Computer systems"], "description": "In modern computer systems, user processes are isolated from each other by the operating system and the hardware.", "description_method": "abstract-excerpt", "abstract": "In modern computer systems, user processes are isolated from each other by the operating system and the hardware. Additionally, in a cloud scenario it is crucial that the hypervisor isolates tenants from other tenants that are co-located on the same physical machine. However, the hypervisor does not protect tenants against the cloud provider and thus the supplied operating system and hardware. Intel SGX provides a mechanism that addresses this scenario. It aims at protecting user-level software from attacks from other processes, the operating system, and even physical attackers. In this paper, we demonstrate fine-grained software-based side-channel attacks from a malicious SGX enclave targeting co-located enclaves. Our attack is the first malware running on real SGX hardware, abusing SGX protection features to conceal itself. Furthermore, we demonstrate our attack both in a native environment and across multiple Docker containers. We perform a Prime+Probe cache side-channel attack on a co-located SGX enclave running an up-to-date RSA implementation that uses a constant-time multiplication primitive. The attack works although in SGX enclaves there are no timers, no large pages, no physical addresses, and no shared memory. In a semi-synchronous attack, we extract 96% of an RSA private key from a single trace. We extract the full RSA private key in an automated attack from 11 traces within 5 minutes.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-03", "2018-05", "2022-06"], "last_mentioned": "2022-06-15T11:28:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1705.03394", "slug": "94b85c1c42093005", "title": "That is not dead which can eternal lie: the aestivation hypothesis for resolving Fermi's paradox", "authors": ["Anders Sandberg", "Stuart Armstrong", "Milan M. Cirkovic"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1705.03394", "doi": null, "topics": ["Statistics & probability"], "description": "If a civilization wants to maximize computation it appears rational to aestivate until the far future in order to exploit the low temperature environment: this can produce a $10^{30}$ multiplier of achievable computation.", "description_method": "abstract-excerpt", "abstract": "If a civilization wants to maximize computation it appears rational to aestivate until the far future in order to exploit the low temperature environment: this can produce a $10^{30}$ multiplier of achievable computation. We hence suggest the \"aestivation hypothesis\": the reason we are not observing manifestations of alien civilizations is that they are currently (mostly) inactive, patiently waiting for future cosmic eras. This paper analyzes the assumptions going into the hypothesis and how physical law and observational evidence constrain the motivations of aliens compatible with the hypothesis.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-07", "2018-06", "2018-10"], "last_mentioned": "2018-10-10T19:09:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1705.08039", "slug": "4632f1c92af8f8a8", "title": "Poincaré Embeddings for Learning Hierarchical Representations", "authors": ["Maximilian Nickel", "Douwe Kiela"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1705.08039", "doi": null, "topics": ["Algorithms", "Mathematics", "Machine learning"], "description": "Representation learning has become an invaluable approach for learning from symbolic data such as text and graphs.", "description_method": "abstract-excerpt", "abstract": "Representation learning has become an invaluable approach for learning from symbolic data such as text and graphs. However, while complex symbolic datasets often exhibit a latent hierarchical structure, state-of-the-art methods typically learn embeddings in Euclidean vector spaces, which do not account for this property. For this purpose, we introduce a new approach for learning hierarchical representations of symbolic data by embedding them into hyperbolic space -- or more precisely into an n-dimensional Poincaré ball. Due to the underlying hyperbolic geometry, this allows us to learn parsimonious representations of symbolic data by simultaneously capturing hierarchy and similarity. We introduce an efficient algorithm to learn the embeddings based on Riemannian optimization and show experimentally that Poincaré embeddings outperform Euclidean embeddings significantly on data with latent hierarchies, both in terms of representation capacity and in terms of generalization ability.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2019-08", "2021-02", "2024-04"], "last_mentioned": "2024-04-25T12:47:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1705.08807", "slug": "e16ffa2be4d295fa", "title": "When Will AI Exceed Human Performance? Evidence from AI Experts", "authors": ["Katja Grace", "John Salvatier", "Allan Dafoe", "Baobao Zhang", "Owain Evans"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1705.08807", "doi": null, "topics": ["Psychology", "Machine learning", "Economics"], "description": "Advances in artificial intelligence (AI) will transform modern life by reshaping transportation, health, science, finance, and the military.", "description_method": "abstract-excerpt", "abstract": "Advances in artificial intelligence (AI) will transform modern life by reshaping transportation, health, science, finance, and the military. To adapt public policy, we need to better anticipate these advances. Here we report the results from a large survey of machine learning researchers on their beliefs about progress in AI. Researchers predict AI will outperform humans in many activities in the next ten years, such as translating languages (by 2024), writing high-school essays (by 2026), driving a truck (by 2027), working in retail (by 2031), writing a bestselling book (by 2049), and working as a surgeon (by 2053). Researchers believe there is a 50% chance of AI outperforming humans in all tasks in 45 years and of automating all human jobs in 120 years, with Asian respondents expecting these dates much sooner than North Americans. These results will inform discussion amongst researchers and policymakers about anticipating and managing trends in AI.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2017-05", "2017-10", "2021-03"], "last_mentioned": "2021-03-05T01:05:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1706.00526", "slug": "322eabc87dea7752", "title": "Knowledge Representation in Bicategories of Relations", "authors": ["Evan Patterson"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1706.00526", "doi": null, "topics": ["Databases", "Mathematics", "Programming languages"], "description": "We introduce the relational ontology log, or relational olog, a knowledge representation system based on the category of sets and relations.", "description_method": "abstract-excerpt", "abstract": "We introduce the relational ontology log, or relational olog, a knowledge representation system based on the category of sets and relations. It is inspired by Spivak and Kent's olog, a recent categorical framework for knowledge representation. Relational ologs interpolate between ologs and description logic, the dominant formalism for knowledge representation today. In this paper, we investigate relational ologs both for their own sake and to gain insight into the relationship between the algebraic and logical approaches to knowledge representation. On a practical level, we show by example that relational ologs have a friendly and intuitive--yet fully precise--graphical syntax, derived from the string diagrams of monoidal categories. We explain several other useful features of relational ologs not possessed by most description logics, such as a type system and a rich, flexible notion of instance data. In a more theoretical vein, we draw on categorical logic to show how relational ologs can be translated to and from logical theories in a fragment of first-order logic. Although we make extensive use of categorical language, this paper is designed to be self-contained and has considerable expository content. The only prerequisites are knowledge of first-order logic and the rudiments of category theory.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2018-05", "2018-07", "2018-12", "2019-08"], "last_mentioned": "2019-08-25T14:04:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1706.01427", "slug": "5b8c5e725f785438", "title": "A simple neural network module for relational reasoning", "authors": ["Adam Santoro", "David Raposo", "David G. T. Barrett", "Mateusz Malinowski", "Razvan Pascanu", "Peter Battaglia", "Timothy Lillicrap"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1706.01427", "doi": null, "topics": ["Databases", "Machine learning", "Psychology"], "description": "Relational reasoning is a central component of generally intelligent behavior, but has proven difficult for neural networks to learn.", "description_method": "abstract-excerpt", "abstract": "Relational reasoning is a central component of generally intelligent behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamic physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Our work shows how a deep learning architecture equipped with an RN module can implicitly discover and learn to reason about entities and their relations.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-06", "2018-12", "2023-02", "2026-07"], "last_mentioned": "2026-07-07T19:03:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1706.04599", "slug": "568ba5f3883a20fa", "title": "On Calibration of Modern Neural Networks", "authors": ["Chuan Guo", "Geoff Pleiss", "Yu Sun", "Kilian Q. Weinberger"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1706.04599", "doi": null, "topics": ["Machine learning", "Statistics & probability"], "description": "Confidence calibration -- the problem of predicting probability estimates representative of the true correctness likelihood -- is important for classification models in many applications.", "description_method": "abstract-excerpt", "abstract": "Confidence calibration -- the problem of predicting probability estimates representative of the true correctness likelihood -- is important for classification models in many applications. We discover that modern neural networks, unlike those from a decade ago, are poorly calibrated. Through extensive experiments, we observe that depth, width, weight decay, and Batch Normalization are important factors influencing calibration. We evaluate the performance of various post-processing calibration methods on state-of-the-art architectures with image and document classification datasets. Our analysis and experiments not only offer insights into neural network learning, but also provide a simple and straightforward recipe for practical settings: on most datasets, temperature scaling -- a single-parameter variant of Platt Scaling -- is surprisingly effective at calibrating predictions.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-04", "2021-04", "2023-01", "2023-09"], "last_mentioned": "2023-09-06T04:18:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1709.02753", "slug": "3137a0f2aa8ee94a", "title": "Privacy Loss in Apple's Implementation of Differential Privacy on MacOS 10.12", "authors": ["Jun Tang", "Aleksandra Korolova", "Xiaolong Bai", "Xueqiang Wang", "Xiaofeng Wang"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1709.02753", "doi": null, "topics": ["Security & cryptography", "Algorithms"], "description": "In June 2016, Apple announced that it will deploy differential privacy for some user data collection in order to ensure privacy of user data, even from Apple.", "description_method": "abstract-excerpt", "abstract": "In June 2016, Apple announced that it will deploy differential privacy for some user data collection in order to ensure privacy of user data, even from Apple. The details of Apple's approach remained sparse. Although several patents have since appeared hinting at the algorithms that may be used to achieve differential privacy, they did not include a precise explanation of the approach taken to privacy parameter choice. Such choice and the overall approach to privacy budget use and management are key questions for understanding the privacy protections provided by any deployment of differential privacy. In this work, through a combination of experiments, static and dynamic code analysis of macOS Sierra (Version 10.12) implementation, we shed light on the choices Apple made for privacy budget management. We discover and describe Apple's set-up for differentially private data processing, including the overall data pipeline, the parameters used for differentially private perturbation of each piece of data, and the frequency with which such data is sent to Apple's servers. We find that although Apple's deployment ensures that the (differential) privacy loss per each datum submitted to its servers is $1$ or $2$, the overall privacy loss permitted by the system is significantly higher, as high as $16$ per day for the four initially announced applications of Emojis, New words, Deeplinks and Lookup Hints. Furthermore, Apple renews the privacy budget available every day, which leads to a possible privacy loss of 16 times the number of days since user opt-in to differentially private data collection for those four applications. We advocate that in order to claim the full benefits of differentially private data collection, Apple must give full transparency of its implementation, enable user choice in areas related to privacy loss, and set meaningful defaults on the privacy loss permitted.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-09", "2017-11", "2018-03", "2025-04"], "last_mentioned": "2025-04-15T16:59:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1710.03740", "slug": "24ceb06b59607c3d", "title": "Mixed Precision Training", "authors": ["Paulius Micikevicius", "Sharan Narang", "Jonah Alben", "Gregory Diamos", "Erich Elsen", "David Garcia", "Boris Ginsburg", "Michael Houston", "Oleksii Kuchaiev", "Ganesh Venkatesh", "Hao Wu"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1710.03740", "doi": null, "topics": ["Machine learning", "Computer systems"], "description": "Deep neural networks have enabled progress in a wide variety of applications.", "description_method": "abstract-excerpt", "abstract": "Deep neural networks have enabled progress in a wide variety of applications. Growing the size of the neural network typically results in improved accuracy. As model sizes grow, the memory and compute requirements for training these models also increases. We introduce a technique to train deep neural networks using half precision floating point numbers. In our technique, weights, activations and gradients are stored in IEEE half-precision format. Half-precision floating numbers have limited numerical range compared to single-precision numbers. We propose two techniques to handle this loss of information. Firstly, we recommend maintaining a single-precision copy of the weights that accumulates the gradients after each optimizer step. This single-precision copy is rounded to half-precision format during training. Secondly, we propose scaling the loss appropriately to handle the loss of information with half-precision gradients. We demonstrate that this approach works for a wide variety of models including convolution neural networks, recurrent neural networks and generative adversarial networks. This technique works for large scale models with more than 100 million parameters trained on large datasets. Using this approach, we can reduce the memory consumption of deep learning models by nearly 2x. In future processors, we can also expect a significant computation speedup using half-precision hardware units.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-10", "2018-10", "2019-06", "2023-07"], "last_mentioned": "2023-07-12T20:01:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1710.10196", "slug": "fe97aded6f63bd01", "title": "Progressive Growing of GANs for Improved Quality, Stability, and Variation", "authors": ["Tero Karras", "Timo Aila", "Samuli Laine", "Jaakko Lehtinen"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1710.10196", "doi": null, "topics": ["Other research"], "description": "We describe a new training methodology for generative adversarial networks.", "description_method": "abstract-excerpt", "abstract": "We describe a new training methodology for generative adversarial networks. The key idea is to grow both the generator and discriminator progressively: starting from a low resolution, we add new layers that model increasingly fine details as training progresses. This both speeds the training up and greatly stabilizes it, allowing us to produce images of unprecedented quality, e.g., CelebA images at 1024^2. We also propose a simple way to increase the variation in generated images, and achieve a record inception score of 8.80 in unsupervised CIFAR10. Additionally, we describe several implementation details that are important for discouraging unhealthy competition between the generator and discriminator. Finally, we suggest a new metric for evaluating GAN results, both in terms of image quality and variation. As an additional contribution, we construct a higher-quality version of the CelebA dataset.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-01", "2018-04", "2018-10", "2023-11"], "last_mentioned": "2023-11-29T20:15:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1710.10377", "slug": "75188694b1cd24a3", "title": "Quantum attacks on Bitcoin, and how to protect against them", "authors": ["Divesh Aggarwal", "Gavin K. Brennen", "Troy Lee", "Miklos Santha", "Marco Tomamichel"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1710.10377", "doi": "10.5195/ledger.2018.127", "topics": ["Physics", "Security & cryptography", "Economics"], "description": "The key cryptographic protocols used to secure the internet and financial transactions of today are all susceptible to attack by the development of a sufficiently large quantum computer.", "description_method": "abstract-excerpt", "abstract": "The key cryptographic protocols used to secure the internet and financial transactions of today are all susceptible to attack by the development of a sufficiently large quantum computer. One particular area at risk are cryptocurrencies, a market currently worth over 150 billion USD. We investigate the risk of Bitcoin, and other cryptocurrencies, to attacks by quantum computers. We find that the proof-of-work used by Bitcoin is relatively resistant to substantial speedup by quantum computers in the next 10 years, mainly because specialized ASIC miners are extremely fast compared to the estimated clock speed of near-term quantum computers. On the other hand, the elliptic curve signature scheme used by Bitcoin is much more at risk, and could be completely broken by a quantum computer as early as 2027, by the most optimistic estimates. We analyze an alternative proof-of-work called Momentum, based on finding collisions in a hash function, that is even more resistant to speedup by a quantum computer. We also review the available post-quantum signature schemes to see which one would best meet the security and efficiency requirements of blockchain applications.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2017-11", "2018-01", "2018-04", "2018-06"], "last_mentioned": "2018-06-30T04:58:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1711.10455", "slug": "175a5c59e60843ba", "title": "Backprop as Functor: A compositional perspective on supervised learning", "authors": ["Brendan Fong", "David I. Spivak", "Rémy Tuyéras"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1711.10455", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "A supervised learning algorithm searches over a set of functions $A \\to B$ parametrised by a space $P$ to find the best approximation to some ideal function $f\\colon A \\to B$.", "description_method": "abstract-excerpt", "abstract": "A supervised learning algorithm searches over a set of functions $A \\to B$ parametrised by a space $P$ to find the best approximation to some ideal function $f\\colon A \\to B$. It does this by taking examples $(a,f(a)) \\in A\\times B$, and updating the parameter according to some rule. We define a category where these update rules may be composed, and show that gradient descent---with respect to a fixed step size and an error function satisfying a certain property---defines a monoidal functor from a category of parametrised functions to this category of update rules. This provides a structural perspective on backpropagation, as well as a broad generalisation of neural networks.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-02", "2018-03", "2019-04", "2021-04"], "last_mentioned": "2021-04-05T18:42:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1801.03400", "slug": "749de88067646388", "title": "Scale-free networks are rare", "authors": ["Anna D. Broido", "Aaron Clauset"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1801.03400", "doi": "10.1038/s41467-019-08746-5", "topics": ["Statistics & probability", "Biology & medicine", "Distributed systems"], "description": "A central claim in modern network science is that real-world networks are typically \"scale free,\" meaning that the fraction of nodes with degree $k$ follows a power law, decaying like $k^{-α}$, often with $2 < α< 3$.", "description_method": "abstract-excerpt", "abstract": "A central claim in modern network science is that real-world networks are typically \"scale free,\" meaning that the fraction of nodes with degree $k$ follows a power law, decaying like $k^{-α}$, often with $2 < α< 3$. However, empirical evidence for this belief derives from a relatively small number of real-world networks. We test the universality of scale-free structure by applying state-of-the-art statistical tools to a large corpus of nearly 1000 network data sets drawn from social, biological, technological, and informational sources. We fit the power-law model to each degree distribution, test its statistical plausibility, and compare it via a likelihood ratio test to alternative, non-scale-free models, e.g., the log-normal. Across domains, we find that scale-free networks are rare, with only 4% exhibiting the strongest-possible evidence of scale-free structure and 52% exhibiting the weakest-possible evidence. Furthermore, evidence of scale-free structure is not uniformly distributed across sources: social networks are at best weakly scale free, while a handful of technological and biological networks can be called strongly scale free. These results undermine the universality of scale-free networks and reveal that real-world networks exhibit a rich structural diversity that will likely require new ideas and mechanisms to explain.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-02", "2018-06", "2018-10"], "last_mentioned": "2018-10-13T14:45:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1802.05365", "slug": "2c21823edad42ee0", "title": "Deep contextualized word representations", "authors": ["Matthew E. Peters", "Mark Neumann", "Mohit Iyyer", "Matt Gardner", "Christopher Clark", "Kenton Lee", "Luke Zettlemoyer"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1802.05365", "doi": null, "topics": ["Language models"], "description": "We introduce a new type of deep contextualized word representation that models both (1) complex characteristics of word use (e.g., syntax and semantics), and (2) how these uses vary across linguistic contexts (i.e., to model polysemy).", "description_method": "abstract-excerpt", "abstract": "We introduce a new type of deep contextualized word representation that models both (1) complex characteristics of word use (e.g., syntax and semantics), and (2) how these uses vary across linguistic contexts (i.e., to model polysemy). Our word vectors are learned functions of the internal states of a deep bidirectional language model (biLM), which is pre-trained on a large text corpus. We show that these representations can be easily added to existing models and significantly improve the state of the art across six challenging NLP problems, including question answering, textual entailment and sentiment analysis. We also present an analysis showing that exposing the deep internals of the pre-trained network is crucial, allowing downstream models to mix different types of semi-supervision signals.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-09", "2018-10", "2023-02", "2023-09"], "last_mentioned": "2023-09-07T18:18:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.02796", "slug": "7e50d81435027b24", "title": "Resource Polymorphism", "authors": ["Guillaume Munch-Maccagnoni"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.02796", "doi": null, "topics": ["Programming languages", "Mathematics"], "description": "We present a resource-management model for ML-style programming languages, designed to be compatible with the OCaml philosophy and runtime model.", "description_method": "abstract-excerpt", "abstract": "We present a resource-management model for ML-style programming languages, designed to be compatible with the OCaml philosophy and runtime model. This is a proposal to extend the OCaml language with destructors, move semantics, and resource polymorphism, to improve its safety, efficiency, interoperability, and expressiveness. It builds on the ownership-and-borrowing models of systems programming languages (Cyclone, C++11, Rust) and on linear types in functional programming (Linear Lisp, Clean, Alms). It continues a synthesis of resources from systems programming and resources in linear logic initiated by Baker. It is a combination of many known and some new ideas. On the novel side, it highlights the good mathematical structure of Stroustrup's \"Resource acquisition is initialisation\" (RAII) idiom for resource management based on destructors, a notion sometimes confused with finalizers, and builds on it a notion of resource polymorphism, inspired by polarisation in proof theory, that mixes C++'s RAII and a tracing garbage collector (GC). The proposal targets a new spot in the design space, with an automatic and predictable resource-management model, at the same time based on lightweight and expressive language abstractions. It is backwards-compatible: current code is expected to run with the same performance, the new abstractions fully combine with the current ones, and it supports a resource-polymorphic extension of libraries. It does so with only a few additions to the runtime, and it integrates with the current GC implementation. It is also compatible with the upcoming multicore extension, and suggests that the Rust model for eliminating data-races applies. Interesting questions arise for a safe and practical type system, many of which have already been thoroughly investigated in the languages and prototypes Cyclone, Rust, and Alms.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2018-12", "2019-12", "2020-06", "2021-12"], "last_mentioned": "2021-12-23T08:53:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.08823", "slug": "ea060e842b6c210a", "title": "A high-bias, low-variance introduction to Machine Learning for physicists", "authors": ["Pankaj Mehta", "Marin Bukov", "Ching-Hao Wang", "Alexandre G. R. Day", "Clint Richardson", "Charles K. Fisher", "David J. Schwab"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.08823", "doi": "10.1016/j.physrep.2019.03.001", "topics": ["Machine learning", "Statistics & probability", "Psychology"], "description": "Machine Learning (ML) is one of the most exciting and dynamic areas of modern research and application.", "description_method": "abstract-excerpt", "abstract": "Machine Learning (ML) is one of the most exciting and dynamic areas of modern research and application. The purpose of this review is to provide an introduction to the core concepts and tools of machine learning in a manner easily understood and intuitive to physicists. The review begins by covering fundamental concepts in ML and modern statistics such as the bias-variance tradeoff, overfitting, regularization, generalization, and gradient descent before moving on to more advanced topics in both supervised and unsupervised learning. Topics covered in the review include ensemble models, deep learning and neural networks, clustering and data visualization, energy-based models (including MaxEnt models and Restricted Boltzmann Machines), and variational methods. Throughout, we emphasize the many natural connections between ML and statistical physics. A notable aspect of the review is the use of Python Jupyter notebooks to introduce modern ML/statistical packages to readers using physics-inspired datasets (the Ising Model and Monte-Carlo simulations of supersymmetric decays of proton-proton collisions). We conclude with an extended outlook discussing possible uses of machine learning for furthering our understanding of the physical world as well as open problems in ML where physicists may be able to contribute. (Notebooks are available at https://physics.bu.edu/~pankajm/MLnotebooks.html )", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-04", "2018-07", "2018-08", "2023-06"], "last_mentioned": "2023-06-17T23:31:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1805.10941", "slug": "5db0c785dd559600", "title": "Fast Random Integer Generation in an Interval", "authors": ["Daniel Lemire"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1805.10941", "doi": "10.1145/3230636", "topics": ["Statistics & probability", "Psychology", "Algorithms"], "description": "In simulations, probabilistic algorithms and statistical tests, we often generate random integers in an interval (e.g., [0,s)).", "description_method": "abstract-excerpt", "abstract": "In simulations, probabilistic algorithms and statistical tests, we often generate random integers in an interval (e.g., [0,s)). For example, random integers in an interval are essential to the Fisher-Yates random shuffle. Consequently, popular languages like Java, Python, C++, Swift and Go include ranged random integer generation functions as part of their runtime libraries. Pseudo-random values are usually generated in words of a fixed number of bits (e.g., 32 bits, 64 bits) using algorithms such as a linear congruential generator. We need functions to convert such random words to random integers in an interval ([0,s)) without introducing statistical biases. The standard functions in programming languages such as Java involve integer divisions. Unfortunately, division instructions are relatively expensive. We review an unbiased function to generate ranged integers from a source of random words that avoids integer divisions with high probability. To establish the practical usefulness of the approach, we show that this algorithm can multiply the speed of unbiased random shuffling on x64 processors. Our proposed approach has been adopted by the Go language for its implementation of the shuffle function.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-08", "2020-10", "2021-08"], "last_mentioned": "2021-08-18T21:29:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1806.06850", "slug": "58c1fed488c6f92e", "title": "Polynomial Regression As an Alternative to Neural Nets", "authors": ["Xi Cheng", "Bohdan Khomtchouk", "Norman Matloff", "Pete Mohanty"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1806.06850", "doi": null, "topics": ["Statistics & probability", "Machine learning"], "description": "Despite the success of neural networks (NNs), there is still a concern among many over their \"black box\" nature.", "description_method": "abstract-excerpt", "abstract": "Despite the success of neural networks (NNs), there is still a concern among many over their \"black box\" nature. Why do they work? Here we present a simple analytic argument that NNs are in fact essentially polynomial regression models. This view will have various implications for NNs, e.g. providing an explanation for why convergence problems arise in NNs, and it gives rough guidance on avoiding overfitting. In addition, we use this phenomenon to predict and confirm a multicollinearity property of NNs not previously reported in the literature. Most importantly, given this loose correspondence, one may choose to routinely use polynomial models instead of NNs, thus avoiding some major problems of the latter, such as having to set many tuning parameters and dealing with convergence issues. We present a number of empirical results; in each case, the accuracy of the polynomial approach matches or exceeds that of NN approaches. A many-featured, open-source software package, polyreg, is available.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-08", "2020-10", "2021-05", "2026-05"], "last_mentioned": "2026-05-08T15:53:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1806.09460", "slug": "dbdc02fe5c62f36b", "title": "A Tour of Reinforcement Learning: The View from Continuous Control", "authors": ["Benjamin Recht"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1806.09460", "doi": null, "topics": ["Machine learning", "Algorithms", "Psychology"], "description": "This manuscript surveys reinforcement learning from the perspective of optimization and control with a focus on continuous control applications.", "description_method": "abstract-excerpt", "abstract": "This manuscript surveys reinforcement learning from the perspective of optimization and control with a focus on continuous control applications. It surveys the general formulation, terminology, and typical experimental implementations of reinforcement learning and reviews competing solution paradigms. In order to compare the relative merits of various techniques, this survey presents a case study of the Linear Quadratic Regulator (LQR) with unknown dynamics, perhaps the simplest and best-studied problem in optimal control. The manuscript describes how merging techniques from learning theory and control can provide non-asymptotic characterizations of LQR performance and shows that these characterizations tend to match experimental behavior. In turn, when revisiting more complex applications, many of the observed phenomena in LQR persist. In particular, theory and experiment demonstrate the role and importance of models and the cost of generality in reinforcement learning algorithms. This survey concludes with a discussion of some of the challenges in designing learning systems that safely and reliably interact with complex and uncertain environments and how tools from reinforcement learning and control might be combined to approach these challenges.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-07", "2019-02", "2021-01"], "last_mentioned": "2021-01-10T22:05:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1807.03341", "slug": "89305d0766ba7d80", "title": "Troubling Trends in Machine Learning Scholarship", "authors": ["Zachary C. Lipton", "Jacob Steinhardt"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1807.03341", "doi": null, "topics": ["Machine learning", "Mathematics", "Economics"], "description": "Collectively, machine learning (ML) researchers are engaged in the creation and dissemination of knowledge about data-driven algorithms.", "description_method": "abstract-excerpt", "abstract": "Collectively, machine learning (ML) researchers are engaged in the creation and dissemination of knowledge about data-driven algorithms. In a given paper, researchers might aspire to any subset of the following goals, among others: to theoretically characterize what is learnable, to obtain understanding through empirically rigorous experiments, or to build a working system that has high predictive accuracy. While determining which knowledge warrants inquiry may be subjective, once the topic is fixed, papers are most valuable to the community when they act in service of the reader, creating foundational knowledge and communicating as clearly as possible. Recent progress in machine learning comes despite frequent departures from these ideals. In this paper, we focus on the following four patterns that appear to us to be trending in ML scholarship: (i) failure to distinguish between explanation and speculation; (ii) failure to identify the sources of empirical gains, e.g., emphasizing unnecessary modifications to neural architectures when gains actually stem from hyper-parameter tuning; (iii) mathiness: the use of mathematics that obfuscates or impresses rather than clarifies, e.g., by confusing technical and non-technical concepts; and (iv) misuse of language, e.g., by choosing terms of art with colloquial connotations or by overloading established technical terms. While the causes behind these patterns are uncertain, possibilities include the rapid expansion of the community, the consequent thinness of the reviewer pool, and the often-misaligned incentives between scholarship and short-term measures of success (e.g., bibliometrics, attention, and entrepreneurial opportunity). While each pattern offers a corresponding remedy (don't do it), we also discuss some speculative suggestions for how the community might combat these trends.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2019-09", "2020-10", "2024-09", "2025-03"], "last_mentioned": "2025-03-13T19:03:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1811.06621", "slug": "6216110b40030a2d", "title": "Streaming End-to-end Speech Recognition For Mobile Devices", "authors": ["Yanzhang He", "Tara N. Sainath", "Rohit Prabhavalkar", "Ian McGraw", "Raziel Alvarez", "Ding Zhao", "David Rybach", "Anjuli Kannan", "Yonghui Wu", "Ruoming Pang", "Qiao Liang", "Deepti Bhatia", "Yuan Shangguan", "Bo Li", "Golan Pundak", "Khe Chai Sim", "Tom Bagby", "Shuo-yiin Chang", "Kanishka Rao", "Alexander Gruenstein"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1811.06621", "doi": null, "topics": ["Psychology", "Machine learning"], "description": "End-to-end (E2E) models, which directly predict output character sequences given input speech, are good candidates for on-device speech recognition.", "description_method": "abstract-excerpt", "abstract": "End-to-end (E2E) models, which directly predict output character sequences given input speech, are good candidates for on-device speech recognition. E2E models, however, present numerous challenges: In order to be truly useful, such models must decode speech utterances in a streaming fashion, in real time; they must be robust to the long tail of use cases; they must be able to leverage user-specific context (e.g., contact lists); and above all, they must be extremely accurate. In this work, we describe our efforts at building an E2E speech recognizer using a recurrent neural network transducer. In experimental evaluations, we find that the proposed approach can outperform a conventional CTC-based model in terms of both latency and accuracy in a number of evaluation categories.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2019-05", "2019-11", "2020-07", "2020-11"], "last_mentioned": "2020-11-15T10:23:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1812.08989", "slug": "31ffe8305347f04f", "title": "The Design and Implementation of XiaoIce, an Empathetic Social Chatbot", "authors": ["Li Zhou", "Jianfeng Gao", "Di Li", "Heung-Yeung Shum"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1812.08989", "doi": null, "topics": ["Psychology"], "description": "This paper describes the development of Microsoft XiaoIce, the most popular social chatbot in the world.", "description_method": "abstract-excerpt", "abstract": "This paper describes the development of Microsoft XiaoIce, the most popular social chatbot in the world. XiaoIce is uniquely designed as an AI companion with an emotional connection to satisfy the human need for communication, affection, and social belonging. We take into account both intelligent quotient (IQ) and emotional quotient (EQ) in system design, cast human-machine social chat as decision-making over Markov Decision Processes (MDPs), and optimize XiaoIce for long-term user engagement, measured in expected Conversation-turns Per Session (CPS). We detail the system architecture and key components including dialogue manager, core chat, skills, and an empathetic computing module. We show how XiaoIce dynamically recognizes human feelings and states, understands user intent, and responds to user needs throughout long conversations. Since her launch in 2014, XiaoIce has communicated with over 660 million active users and succeeded in establishing long-term relationships with many of them. Analysis of large scale online logs shows that XiaoIce has achieved an average CPS of 23, which is significantly higher than that of other chatbots and even human conversations.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2020-12", "2023-01", "2023-03", "2023-11"], "last_mentioned": "2023-11-05T11:32:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1901.00596", "slug": "1e257f3dca198e61", "title": "A Comprehensive Survey on Graph Neural Networks", "authors": ["Zonghan Wu", "Shirui Pan", "Fengwen Chen", "Guodong Long", "Chengqi Zhang", "Philip S. Yu"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1901.00596", "doi": "10.1109/TNNLS.2020.2978386", "topics": ["Machine learning", "Computer vision", "Algorithms"], "description": "Deep learning has revolutionized many machine learning tasks in recent years, ranging from image classification and video processing to speech recognition and natural language understanding.", "description_method": "abstract-excerpt", "abstract": "Deep learning has revolutionized many machine learning tasks in recent years, ranging from image classification and video processing to speech recognition and natural language understanding. The data in these tasks are typically represented in the Euclidean space. However, there is an increasing number of applications where data are generated from non-Euclidean domains and are represented as graphs with complex relationships and interdependency between objects. The complexity of graph data has imposed significant challenges on existing machine learning algorithms. Recently, many studies on extending deep learning approaches for graph data have emerged. In this survey, we provide a comprehensive overview of graph neural networks (GNNs) in data mining and machine learning fields. We propose a new taxonomy to divide the state-of-the-art graph neural networks into four categories, namely recurrent graph neural networks, convolutional graph neural networks, graph autoencoders, and spatial-temporal graph neural networks. We further discuss the applications of graph neural networks across various domains and summarize the open source codes, benchmark data sets, and model evaluation of graph neural networks. Finally, we propose potential research directions in this rapidly growing field.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2019-01", "2020-02", "2022-05"], "last_mentioned": "2022-05-09T19:10:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1902.08318", "slug": "716d993cf4b62c23", "title": "Parsing Gigabytes of JSON per Second", "authors": ["Geoff Langdale", "Daniel Lemire"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1902.08318", "doi": "10.1007/s00778-019-00578-5", "topics": ["Science & research", "Computer systems"], "description": "JavaScript Object Notation or JSON is a ubiquitous data exchange format on the Web.", "description_method": "abstract-excerpt", "abstract": "JavaScript Object Notation or JSON is a ubiquitous data exchange format on the Web. Ingesting JSON documents can become a performance bottleneck due to the sheer volume of data. We are thus motivated to make JSON parsing as fast as possible. Despite the maturity of the problem of JSON parsing, we show that substantial speedups are possible. We present the first standard-compliant JSON parser to process gigabytes of data per second on a single core, using commodity processors. We can use a quarter or fewer instructions than a state-of-the-art reference parser like RapidJSON. Unlike other validating parsers, our software (simdjson) makes extensive use of Single Instruction, Multiple Data (SIMD) instructions. To ensure reproducibility, simdjson is freely available as open-source software under a liberal license.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2019-07", "2021-10", "2024-03", "2025-08"], "last_mentioned": "2025-08-30T15:48:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1902.11097", "slug": "6b089df946b44cba", "title": "Predictive Inequity in Object Detection", "authors": ["Benjamin Wilson", "Judy Hoffman", "Jamie Morgenstern"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1902.11097", "doi": null, "topics": ["Computer vision", "Psychology"], "description": "In this work, we investigate whether state-of-the-art object detection systems have equitable predictive performance on pedestrians with different skin tones.", "description_method": "abstract-excerpt", "abstract": "In this work, we investigate whether state-of-the-art object detection systems have equitable predictive performance on pedestrians with different skin tones. This work is motivated by many recent examples of ML and vision systems displaying higher error rates for certain demographic groups than others. We annotate an existing large scale dataset which contains pedestrians, BDD100K, with Fitzpatrick skin tones in ranges [1-3] or [4-6]. We then provide an in-depth comparative analysis of performance between these two skin tone groupings, finding that neither time of day nor occlusion explain this behavior, suggesting this disparity is not merely the result of pedestrians in the 4-6 range appearing in more difficult scenes for detection. We investigate to what extent time of day, occlusion, and reweighting the supervised loss during training affect this predictive bias.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2019-03", "2019-04", "2019-08"], "last_mentioned": "2019-08-30T16:21:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1904.04998", "slug": "a1c111ace49aa8f4", "title": "Depth from Videos in the Wild: Unsupervised Monocular Depth Learning from Unknown Cameras", "authors": ["Ariel Gordon", "Hanhan Li", "Rico Jonschkowski", "Anelia Angelova"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1904.04998", "doi": null, "topics": ["Distributed systems"], "description": "We present a novel method for simultaneous learning of depth, egomotion, object motion, and camera intrinsics from monocular videos, using only consistency across neighboring video frames as supervision signal.", "description_method": "abstract-excerpt", "abstract": "We present a novel method for simultaneous learning of depth, egomotion, object motion, and camera intrinsics from monocular videos, using only consistency across neighboring video frames as supervision signal. Similarly to prior work, our method learns by applying differentiable warping to frames and comparing the result to adjacent ones, but it provides several improvements: We address occlusions geometrically and differentiably, directly using the depth maps as predicted during training. We introduce randomized layer normalization, a novel powerful regularizer, and we account for object motion relative to the scene. To the best of our knowledge, our work is the first to learn the camera intrinsic parameters, including lens distortion, from video in an unsupervised manner, thereby allowing us to extract accurate depth and motion from arbitrary videos of unknown origin at scale. We evaluate our results on the Cityscapes, KITTI and EuRoC datasets, establishing new state of the art on depth prediction and odometry, and demonstrate qualitatively that depth prediction can be learned from a collection of YouTube videos.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2019-04", "2019-08", "2020-06", "2020-08"], "last_mentioned": "2020-08-19T08:00:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1904.09751", "slug": "97748513224fa951", "title": "The Curious Case of Neural Text Degeneration", "authors": ["Ari Holtzman", "Jan Buys", "Li Du", "Maxwell Forbes", "Yejin Choi"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1904.09751", "doi": null, "topics": ["Language models", "Statistics & probability"], "description": "Despite considerable advancements with deep neural language models, the enigma of neural text degeneration persists when these models are tested as text generators.", "description_method": "abstract-excerpt", "abstract": "Despite considerable advancements with deep neural language models, the enigma of neural text degeneration persists when these models are tested as text generators. The counter-intuitive empirical observation is that even though the use of likelihood as training objective leads to high quality models for a broad range of language understanding tasks, using likelihood as a decoding objective leads to text that is bland and strangely repetitive. In this paper, we reveal surprising distributional differences between human text and machine text. In addition, we find that decoding strategies alone can dramatically effect the quality of machine text, even when generated from exactly the same neural language model. Our findings motivate Nucleus Sampling, a simple but effective method to draw the best out of neural generation. By sampling text from the dynamic nucleus of the probability distribution, which allows for diversity while effectively truncating the less reliable tail of the distribution, the resulting text better demonstrates the quality of human text, yielding enhanced diversity without sacrificing fluency and coherence.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2019-10", "2019-11", "2020-11", "2023-02"], "last_mentioned": "2023-02-14T23:08:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1905.11946", "slug": "dafdec1213cfe784", "title": "EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks", "authors": ["Mingxing Tan", "Quoc V. Le"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1905.11946", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "Convolutional Neural Networks (ConvNets) are commonly developed at a fixed resource budget, and then scaled up for better accuracy if more resources are available.", "description_method": "abstract-excerpt", "abstract": "Convolutional Neural Networks (ConvNets) are commonly developed at a fixed resource budget, and then scaled up for better accuracy if more resources are available. In this paper, we systematically study model scaling and identify that carefully balancing network depth, width, and resolution can lead to better performance. Based on this observation, we propose a new scaling method that uniformly scales all dimensions of depth/width/resolution using a simple yet highly effective compound coefficient. We demonstrate the effectiveness of this method on scaling up MobileNets and ResNet. To go even further, we use neural architecture search to design a new baseline network and scale it up to obtain a family of models, called EfficientNets, which achieve much better accuracy and efficiency than previous ConvNets. In particular, our EfficientNet-B7 achieves state-of-the-art 84.3% top-1 accuracy on ImageNet, while being 8.4x smaller and 6.1x faster on inference than the best existing ConvNet. Our EfficientNets also transfer well and achieve state-of-the-art accuracy on CIFAR-100 (91.7%), Flowers (98.8%), and 3 other transfer learning datasets, with an order of magnitude fewer parameters. Source code is at https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2019-08", "2019-09", "2021-07", "2022-05"], "last_mentioned": "2022-05-28T16:32:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1906.07221", "slug": "7e8d0cca1e2048ba", "title": "Why and How zk-SNARK Works", "authors": ["Maksym Petkus"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1906.07221", "doi": null, "topics": ["Security & cryptography", "Algorithms"], "description": "Despite the existence of multiple great resources on zk-SNARK construction, from original papers to explainers, due to the sheer number of moving parts the subject remains a black box for many.", "description_method": "abstract-excerpt", "abstract": "Despite the existence of multiple great resources on zk-SNARK construction, from original papers to explainers, due to the sheer number of moving parts the subject remains a black box for many. While some pieces of the puzzle are given one can not see the full picture without the missing ones. Hence the focus of this work is to shed light onto the topic with a straightforward and clean approach based on examples and answering many whys along the way so that more individuals can appreciate the state of the art technology, its innovators and ultimately the beauty of math. Paper's contribution is a simplistic exposition with a sufficient and gradually increasing level of complexity, necessary to understand zk-SNARK without any prerequisite knowledge of the subject, cryptography or advanced math. The primary goal is not only to explain how it works but why it works and how it came to be this way.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2020-04", "2021-06", "2021-11", "2025-03"], "last_mentioned": "2025-03-06T17:01:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1907.06902", "slug": "f744f85fa7785afd", "title": "Are We Really Making Much Progress? A Worrying Analysis of Recent Neural Recommendation Approaches", "authors": ["Maurizio Ferrari Dacrema", "Paolo Cremonesi", "Dietmar Jannach"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1907.06902", "doi": "10.1145/3298689.3347058", "topics": ["Machine learning", "Science & research", "Algorithms"], "description": "Deep learning techniques have become the method of choice for researchers working on algorithmic aspects of recommender systems.", "description_method": "abstract-excerpt", "abstract": "Deep learning techniques have become the method of choice for researchers working on algorithmic aspects of recommender systems. With the strongly increased interest in machine learning in general, it has, as a result, become difficult to keep track of what represents the state-of-the-art at the moment, e.g., for top-n recommendation tasks. At the same time, several recent publications point out problems in today's research practice in applied machine learning, e.g., in terms of the reproducibility of the results or the choice of the baselines when proposing new models. In this work, we report the results of a systematic analysis of algorithmic proposals for top-n recommendation tasks. Specifically, we considered 18 algorithms that were presented at top-level research conferences in the last years. Only 7 of them could be reproduced with reasonable effort. For these methods, it however turned out that 6 of them can often be outperformed with comparably simple heuristic methods, e.g., based on nearest-neighbor or graph-based techniques. The remaining one clearly outperformed the baselines but did not consistently outperform a well-tuned non-neural linear ranking method. Overall, our work sheds light on a number of potential problems in today's machine learning scholarship and calls for improved scientific practices in this area. Source code of our experiments and full results are available at: https://github.com/MaurizioFD/RecSys2019_DeepLearning_Evaluation.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2019-12", "2020-08", "2021-03", "2021-07"], "last_mentioned": "2021-07-10T16:58:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1908.06841", "slug": "6f9f20753988d066", "title": "Ternary circuits: why R=3 is not the Optimal Radix for Computation", "authors": ["Daniel Etiemble"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1908.06841", "doi": null, "topics": ["Biology & medicine"], "description": "A demonstration that e=2.718 rounded to 3 is the best radix for computation is disproved.", "description_method": "abstract-excerpt", "abstract": "A demonstration that e=2.718 rounded to 3 is the best radix for computation is disproved. The MOSFET-like CNTFET technology is used to compare inverters, Nand, adders, multipliers, D Flip-Flops and SRAM cells. The transistor count ratio between ternary and binary circuits is generally greater than the log(3)/log(2) information ratio. The only exceptions concern a circuit approach that combines two circuit drawbacks (an additional power supply and a circuit conflict between transistors) and only when it implements circuits based on the ternary inverter. For arithmetic circuits such as adders and multipliers, the ternary circuits are always outperformed by the binary ones using the same technology.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2020-02", "2021-02", "2024-08"], "last_mentioned": "2024-08-09T14:32:39Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1908.11680", "slug": "2d464aa7b46deafa", "title": "Insecure Until Proven Updated: Analyzing AMD SEV's Remote Attestation", "authors": ["Robert Buhren", "Christian Werling", "Jean-Pierre Seifert"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1908.11680", "doi": "10.1145/3319535.3354216", "topics": ["Security & cryptography", "Computer systems"], "description": "Customers of cloud services have to trust the cloud providers, as they control the building blocks that form the cloud.", "description_method": "abstract-excerpt", "abstract": "Customers of cloud services have to trust the cloud providers, as they control the building blocks that form the cloud. This includes the hypervisor enabling the sharing of a single hardware platform among multiple tenants. AMD Secure Encrypted Virtualization (SEV) claims a new level of protection in cloud scenarios. AMD SEV encrypts the main memory of virtual machines with VM-specific keys, thereby denying the higher-privileged hypervisor access to a guest's memory. To enable the cloud customer to verify the correct deployment of his virtual machine, SEV additionally introduces a remote attestation protocol.This paper analyzes the firmware components that implement the SEV remote attestation protocol on the current AMD Epyc Naples CPU series. We demonstrate that it is possible to extract critical CPU-specific keys that are fundamental for the security of the remote attestation protocol.Building on the extracted keys, we propose attacks that allow a malicious cloud provider a complete circumvention of the SEV protection mechanisms. Although the underlying firmware issues were already fixed by AMD, we show that the current series of AMD Epyc CPUs, i.e., the Naples series, does not prevent the installation of previous firmware versions. We show that the severity of our proposed attacks is very high as no purely software-based mitigations are possible. This effectively renders the SEV technology on current AMD Epyc CPUs useless when confronted with an untrusted cloud provider. To overcome these issues, we also propose robust changes to the SEV design that allow future generations of the SEV technology to mitigate the proposed attacks.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-06", "2020-07", "2020-08", "2025-10"], "last_mentioned": "2025-10-10T19:25:57Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1909.11942", "slug": "0b66aa2cd6eeb3a3", "title": "ALBERT: A Lite BERT for Self-supervised Learning of Language Representations", "authors": ["Zhenzhong Lan", "Mingda Chen", "Sebastian Goodman", "Kevin Gimpel", "Piyush Sharma", "Radu Soricut"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1909.11942", "doi": null, "topics": ["Language models", "Machine learning"], "description": "Increasing model size when pretraining natural language representations often results in improved performance on downstream tasks.", "description_method": "abstract-excerpt", "abstract": "Increasing model size when pretraining natural language representations often results in improved performance on downstream tasks. However, at some point further model increases become harder due to GPU/TPU memory limitations and longer training times. To address these problems, we present two parameter-reduction techniques to lower memory consumption and increase the training speed of BERT. Comprehensive empirical evidence shows that our proposed methods lead to models that scale much better compared to the original BERT. We also use a self-supervised loss that focuses on modeling inter-sentence coherence, and show it consistently helps downstream tasks with multi-sentence inputs. As a result, our best model establishes new state-of-the-art results on the GLUE, RACE, and \\squad benchmarks while having fewer parameters compared to BERT-large. The code and the pretrained models are available at https://github.com/google-research/ALBERT.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-01", "2020-12", "2022-06"], "last_mentioned": "2022-06-04T22:45:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1910.10703", "slug": "7cfccc12c6a92e53", "title": "Metamath Zero: The Cartesian Theorem Prover", "authors": ["Mario Carneiro"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1910.10703", "doi": null, "topics": ["Mathematics"], "description": "As the usage of theorem prover technology expands, so too does the reliance on correctness of the tools.", "description_method": "abstract-excerpt", "abstract": "As the usage of theorem prover technology expands, so too does the reliance on correctness of the tools. Metamath Zero is a verification system that aims for simplicity of logic and implementation, without compromising on efficiency of verification. It is formally specified in its own language, and supports a number of translations to and from other proof languages. This paper describes the abstract logic of Metamath Zero, essentially a multi-sorted first order logic, as well as the binary proof format and the way in which it can ensure essentially linear time verification while still being concise and efficient at scale. Metamath Zero currently holds the record for fastest verification of the $\\mathtt{set .mm}$ Metamath library of proofs in ZFC (including 71 of Wiedijk's 100 formalization targets), at less than 200 ms. Ultimately, we intend to use it to verify the correctness of the implementation of the verifier down to binary executable, so it can be used as a root of trust for more complex proof systems.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-11", "2021-02", "2023-03", "2024-02"], "last_mentioned": "2024-02-16T16:16:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2002.09511", "slug": "98acc74a4852605a", "title": "Chronofold: a data structure for versioned text", "authors": ["Victor Grishchenko", "Mikhail Patrakeev"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2002.09511", "doi": null, "topics": ["Distributed systems", "Algorithms", "Economics"], "description": "Chronofold is a replicated data structure for versioned text.", "description_method": "abstract-excerpt", "abstract": "Chronofold is a replicated data structure for versioned text. It is designed for use in collaborative editors and revision control systems. Past models of this kind either retrofitted local linear orders to a distributed system (the OT approach) or employed distributed data models locally (the CRDT approach). That caused either extreme fragility in a distributed setting or egregious overheads in local use. Overall, that local/distributed impedance mismatch is cognitively taxing and causes lots of complexity. We solve that by using subjective linear orders locally at each replica, while inter-replica communication uses a distributed model. A separate translation layer insulates local data structures from the distributed environment. We modify the Lamport timestamping scheme to make that translation as trivial as possible. We believe our approach has applications beyond the domain of collaborative editing.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2020-07", "2021-07", "2026-03"], "last_mentioned": "2026-03-22T19:52:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2003.03384", "slug": "d66175f754509bdb", "title": "AutoML-Zero: Evolving Machine Learning Algorithms From Scratch", "authors": ["Esteban Real", "Chen Liang", "David R. So", "Quoc V. Le"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2003.03384", "doi": null, "topics": ["Machine learning", "Algorithms", "Psychology"], "description": "Machine learning research has advanced in multiple aspects, including model structures and learning methods.", "description_method": "abstract-excerpt", "abstract": "Machine learning research has advanced in multiple aspects, including model structures and learning methods. The effort to automate such research, known as AutoML, has also made significant progress. However, this progress has largely focused on the architecture of neural networks, where it has relied on sophisticated expert-designed layers as building blocks---or similarly restrictive search spaces. Our goal is to show that AutoML can go further: it is possible today to automatically discover complete machine learning algorithms just using basic mathematical operations as building blocks. We demonstrate this by introducing a novel framework that significantly reduces human bias through a generic search space. Despite the vastness of this space, evolutionary search can still discover two-layer neural networks trained by backpropagation. These simple neural networks can then be surpassed by evolving directly on tasks of interest, e.g. CIFAR-10 variants, where modern techniques emerge in the top algorithms, such as bilinear interactions, normalized gradients, and weight averaging. Moreover, evolution adapts algorithms to different task types: e.g., dropout-like techniques appear when little data is available. We believe these preliminary successes in discovering machine learning algorithms from scratch indicate a promising new direction for the field.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-03", "2020-04", "2020-05", "2020-07"], "last_mentioned": "2020-07-10T01:54:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2004.00107", "slug": "82116e379507b20a", "title": "Merkle-CRDTs: Merkle-DAGs meet CRDTs", "authors": ["Hector Sanjuan", "Samuli Poyhtari", "Pedro Teixeira", "Ioannis Psaras"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2004.00107", "doi": null, "topics": ["Distributed systems", "Security & cryptography", "Algorithms"], "description": "We study Merkle-DAGs as a transport and persistence layer for Conflict-Free Replicated Data Types (CRDTs), coining the term Merkle-CRDTs and providing an overview of the different concepts, properties, advantages and limitations involved.", "description_method": "abstract-excerpt", "abstract": "We study Merkle-DAGs as a transport and persistence layer for Conflict-Free Replicated Data Types (CRDTs), coining the term Merkle-CRDTs and providing an overview of the different concepts, properties, advantages and limitations involved. We show how Merkle-DAGs can act as logical clocks giving Merkle-CRDTs the potential to greatly simplify the design and implementation of convergent data types in systems with weak messaging layer guarantees and a very large number of replicas. Merkle-CRDTs can leverage highly scalable distributed technologies like DHTs and PubSub algorithms running underneath to take advantage of the security and de-duplication properties of content-addressing. Examples of such content-oriented systems could include peer-to-peer content exchange and synchronisation applications between opportunistically connected mobile devices, IoT devices or user applications running in a web browser.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2020-04", "2021-10", "2022-03", "2024-03"], "last_mentioned": "2024-03-09T23:26:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2004.03082", "slug": "49c1160b6a486e2b", "title": "egg: Fast and Extensible Equality Saturation", "authors": ["Max Willsey", "Chandrakana Nandi", "Yisu Remy Wang", "Oliver Flatt", "Zachary Tatlock", "Pavel Panchekha"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2004.03082", "doi": "10.1145/3434304", "topics": ["Robotics", "Programming languages", "Mathematics"], "description": "An e-graph efficiently represents a congruence relation over many expressions.", "description_method": "abstract-excerpt", "abstract": "An e-graph efficiently represents a congruence relation over many expressions. Although they were originally developed in the late 1970s for use in automated theorem provers, a more recent technique known as equality saturation repurposes e-graphs to implement state-of-the-art, rewrite-driven compiler optimizations and program synthesizers. However, e-graphs remain unspecialized for this newer use case. Equality saturation workloads exhibit distinct characteristics and often require ad-hoc e-graph extensions to incorporate transformations beyond purely syntactic rewrites. This work contributes two techniques that make e-graphs fast and extensible, specializing them to equality saturation. A new amortized invariant restoration technique called rebuilding takes advantage of equality saturation's distinct workload, providing asymptotic speedups over current techniques in practice. A general mechanism called e-class analyses integrates domain-specific analyses into the e-graph, reducing the need for ad hoc manipulation. We implemented these techniques in a new open-source library called egg. Our case studies on three previously published applications of equality saturation highlight how egg's performance and flexibility enable state-of-the-art results across diverse domains.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-03", "2022-07", "2023-12"], "last_mentioned": "2023-12-16T19:53:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2004.10934", "slug": "190e6e6e8c37dda0", "title": "YOLOv4: Optimal Speed and Accuracy of Object Detection", "authors": ["Alexey Bochkovskiy", "Chien-Yao Wang", "Hong-Yuan Mark Liao"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2004.10934", "doi": null, "topics": ["Computer vision", "Machine learning"], "description": "There are a huge number of features which are said to improve Convolutional Neural Network (CNN) accuracy.", "description_method": "abstract-excerpt", "abstract": "There are a huge number of features which are said to improve Convolutional Neural Network (CNN) accuracy. Practical testing of combinations of such features on large datasets, and theoretical justification of the result, is required. Some features operate on certain models exclusively and for certain problems exclusively, or only for small-scale datasets; while some features, such as batch-normalization and residual-connections, are applicable to the majority of models, tasks, and datasets. We assume that such universal features include Weighted-Residual-Connections (WRC), Cross-Stage-Partial-connections (CSP), Cross mini-Batch Normalization (CmBN), Self-adversarial-training (SAT) and Mish-activation. We use new features: WRC, CSP, CmBN, SAT, Mish activation, Mosaic data augmentation, CmBN, DropBlock regularization, and CIoU loss, and combine some of them to achieve state-of-the-art results: 43.5% AP (65.7% AP50) for the MS COCO dataset at a realtime speed of ~65 FPS on Tesla V100. Source code is at https://github.com/AlexeyAB/darknet", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2020-05", "2020-06", "2021-12", "2022-05"], "last_mentioned": "2022-05-12T21:10:33Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2006.01855", "slug": "acecf4195ee5b967", "title": "Aligning Superhuman AI with Human Behavior: Chess as a Model System", "authors": ["Reid McIlroy-Young", "Siddhartha Sen", "Jon Kleinberg", "Ashton Anderson"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2006.01855", "doi": "10.1145/3394486.3403219", "topics": ["Psychology", "Economics", "Machine learning"], "description": "As artificial intelligence becomes increasingly intelligent---in some cases, achieving superhuman performance---there is growing potential for humans to learn from and collaborate with algorithms.", "description_method": "abstract-excerpt", "abstract": "As artificial intelligence becomes increasingly intelligent---in some cases, achieving superhuman performance---there is growing potential for humans to learn from and collaborate with algorithms. However, the ways in which AI systems approach problems are often different from the ways people do, and thus may be uninterpretable and hard to learn from. A crucial step in bridging this gap between human and artificial intelligence is modeling the granular actions that constitute human behavior, rather than simply matching aggregate human performance. We pursue this goal in a model system with a long history in artificial intelligence: chess. The aggregate performance of a chess player unfolds as they make decisions over the course of a game. The hundreds of millions of games played online by players at every skill level form a rich source of data in which these decisions, and their exact context, are recorded in minute detail. Applying existing chess engines to this data, including an open-source implementation of AlphaZero, we find that they do not predict human moves well. We develop and introduce Maia, a customized version of Alpha-Zero trained on human chess games, that predicts human moves at a much higher accuracy than existing engines, and can achieve maximum accuracy when predicting decisions made by players at a specific skill level in a tuneable way. For a dual task of predicting whether a human will make a large mistake on the next move, we develop a deep neural network that significantly outperforms competitive baselines. Taken together, our results suggest that there is substantial promise in designing artificial intelligence systems with human collaboration in mind by first accurately modeling granular human decision-making.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-12", "2022-11", "2025-09"], "last_mentioned": "2025-09-02T18:52:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2006.07159", "slug": "68a53f9f7b851165", "title": "Are we done with ImageNet?", "authors": ["Lucas Beyer", "Olivier J. Hénaff", "Alexander Kolesnikov", "Xiaohua Zhai", "Aäron van den Oord"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2006.07159", "doi": null, "topics": ["Machine learning", "Psychology", "Computer vision"], "description": "Yes, and no.", "description_method": "abstract-excerpt", "abstract": "Yes, and no. We ask whether recent progress on the ImageNet classification benchmark continues to represent meaningful generalization, or whether the community has started to overfit to the idiosyncrasies of its labeling procedure. We therefore develop a significantly more robust procedure for collecting human annotations of the ImageNet validation set. Using these new labels, we reassess the accuracy of recently proposed ImageNet classifiers, and find their gains to be substantially smaller than those reported on the original labels. Furthermore, we find the original ImageNet labels to no longer be the best predictors of this independently-collected set, indicating that their usefulness in evaluating vision models may be nearing an end. Nevertheless, we find our annotation procedure to have largely remedied the errors in the original labels, reinforcing ImageNet as a powerful benchmark for future research in visual recognition.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2020-09", "2021-02", "2021-04", "2021-09"], "last_mentioned": "2021-09-25T00:33:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2007.00308", "slug": "92ed3e5550ded021", "title": "Polar Stroking: New Theory and Methods for Stroking Paths", "authors": ["Mark J. Kilgard"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2007.00308", "doi": "10.1145/3386569.3392458", "topics": ["Other research"], "description": "Stroking and filling are the two basic rendering operations on paths in vector graphics.", "description_method": "abstract-excerpt", "abstract": "Stroking and filling are the two basic rendering operations on paths in vector graphics. The theory of filling a path is well-understood in terms of contour integrals and winding numbers, but when path rendering standards specify stroking, they resort to the analogy of painting pixels with a brush that traces the outline of the path. This means important standards such as PDF, SVG, and PostScript lack a rigorous way to say what samples are inside or outside a stroked path. Our work fills this gap with a principled theory of stroking. Guided by our theory, we develop a novel polar stroking method to render stroked paths robustly with an intuitive way to bound the tessellation error without needing recursion. Because polar stroking guarantees small uniform steps in tangent angle, it provides an efficient way to accumulate arc length along a path for texturing or dashing. While this paper focuses on developing the theory of our polar stroking method, we have successfully implemented our methods on modern programmable GPUs.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2021-02", "2022-08", "2023-01", "2025-11"], "last_mentioned": "2025-11-11T00:24:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2102.12092", "slug": "8d51903893bff6a4", "title": "Zero-Shot Text-to-Image Generation", "authors": ["Aditya Ramesh", "Mikhail Pavlov", "Gabriel Goh", "Scott Gray", "Chelsea Voss", "Alec Radford", "Mark Chen", "Ilya Sutskever"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2102.12092", "doi": null, "topics": ["Computer vision", "Language models"], "description": "Text-to-image generation has traditionally focused on finding better modeling assumptions for training on a fixed dataset.", "description_method": "abstract-excerpt", "abstract": "Text-to-image generation has traditionally focused on finding better modeling assumptions for training on a fixed dataset. These assumptions might involve complex architectures, auxiliary losses, or side information such as object part labels or segmentation masks supplied during training. We describe a simple approach for this task based on a transformer that autoregressively models the text and image tokens as a single stream of data. With sufficient data and scale, our approach is competitive with previous domain-specific models when evaluated in a zero-shot fashion.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-03", "2022-04", "2022-06", "2023-02"], "last_mentioned": "2023-02-19T16:48:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2103.02515", "slug": "bb48204427293f25", "title": "Ribbon filter: practically smaller than Bloom and Xor", "authors": ["Peter C. Dillinger", "Stefan Walzer"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2103.02515", "doi": null, "topics": ["Algorithms"], "description": "Filter data structures over-approximate a set of hashable keys, i.e.", "description_method": "abstract-excerpt", "abstract": "Filter data structures over-approximate a set of hashable keys, i.e. set membership queries may incorrectly come out positive. A filter with false positive rate $f \\in (0,1]$ is known to require $\\ge \\log_2(1/f)$ bits per key. At least for larger $f \\ge 2^{-4}$, existing practical filters require a space overhead of at least 20% with respect to this information-theoretic bound. We introduce the Ribbon filter: a new filter for static sets with a broad range of configurable space overheads and false positive rates with competitive speed over that range, especially for larger $f \\ge 2^{-7}$. In many cases, Ribbon is faster than existing filters for the same space overhead, or can achieve space overhead below 10% with some additional CPU time. An experimental Ribbon design with load balancing can even achieve space overheads below 1%. A Ribbon filter resembles an Xor filter modified to maximize locality and is constructed by solving a band-like linear system over Boolean variables. In previous work, Dietzfelbinger and Walzer describe this linear system and an efficient Gaussian solver. We present and analyze a faster, more adaptable solving process we call \"Rapid Incremental Boolean Banding ON the fly,\" which resembles hash table construction. We also present and analyze an attractive Ribbon variant based on making the linear system homogeneous, and describe several more practical enhancements.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2022-09", "2022-10", "2025-05", "2026-01"], "last_mentioned": "2026-01-22T10:14:57Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2105.05233", "slug": "6ef31673fbbf3360", "title": "Diffusion Models Beat GANs on Image Synthesis", "authors": ["Prafulla Dhariwal", "Alex Nichol"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2105.05233", "doi": null, "topics": ["Machine learning", "Physics"], "description": "We show that diffusion models can achieve image sample quality superior to the current state-of-the-art generative models.", "description_method": "abstract-excerpt", "abstract": "We show that diffusion models can achieve image sample quality superior to the current state-of-the-art generative models. We achieve this on unconditional image synthesis by finding a better architecture through a series of ablations. For conditional image synthesis, we further improve sample quality with classifier guidance: a simple, compute-efficient method for trading off diversity for fidelity using gradients from a classifier. We achieve an FID of 2.97 on ImageNet 128$\\times$128, 4.59 on ImageNet 256$\\times$256, and 7.72 on ImageNet 512$\\times$512, and we match BigGAN-deep even with as few as 25 forward passes per sample, all while maintaining better coverage of the distribution. Finally, we find that classifier guidance combines well with upsampling diffusion models, further improving FID to 3.94 on ImageNet 256$\\times$256 and 3.85 on ImageNet 512$\\times$512. We release our code at https://github.com/openai/guided-diffusion", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-11", "2024-05", "2024-08", "2025-06"], "last_mentioned": "2025-06-27T23:48:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2106.03253", "slug": "faede35e6ab3c3ca", "title": "Tabular Data: Deep Learning is Not All You Need", "authors": ["Ravid Shwartz-Ziv", "Amitai Armon"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2106.03253", "doi": null, "topics": ["Machine learning", "Statistics & probability"], "description": "A key element in solving real-life data science problems is selecting the types of models to use.", "description_method": "abstract-excerpt", "abstract": "A key element in solving real-life data science problems is selecting the types of models to use. Tree ensemble models (such as XGBoost) are usually recommended for classification and regression problems with tabular data. However, several deep learning models for tabular data have recently been proposed, claiming to outperform XGBoost for some use cases. This paper explores whether these deep models should be a recommended option for tabular data by rigorously comparing the new deep models to XGBoost on various datasets. In addition to systematically comparing their performance, we consider the tuning and computation they require. Our study shows that XGBoost outperforms these deep models across the datasets, including the datasets used in the papers that proposed the deep models. We also demonstrate that XGBoost requires much less tuning. On the positive side, we show that an ensemble of deep models and XGBoost performs better on these datasets than XGBoost alone.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-06", "2022-09", "2024-07"], "last_mentioned": "2024-07-26T15:34:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2106.14204", "slug": "cbf5291b989475ab", "title": "Bitcoin, Currencies, and Fragility", "authors": ["Nassim Nicholas Taleb"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2106.14204", "doi": null, "topics": ["Economics"], "description": "This discussion applies quantitative finance methods and economic arguments to cryptocurrencies in general and bitcoin in particular -- as there are about $10,000$ cryptocurrencies, we focus (unless otherwise specified) on the most discussed crypto of those that claim to hew to the original protocol (Nakamoto 2009) and the one with, by fa", "description_method": "abstract-excerpt", "abstract": "This discussion applies quantitative finance methods and economic arguments to cryptocurrencies in general and bitcoin in particular -- as there are about $10,000$ cryptocurrencies, we focus (unless otherwise specified) on the most discussed crypto of those that claim to hew to the original protocol (Nakamoto 2009) and the one with, by far, the largest market capitalization. In its current version, in spite of the hype, bitcoin failed to satisfy the notion of \"currency without government\" (it proved to not even be a currency at all), can be neither a short nor long term store of value (its expected value is no higher than $0$), cannot operate as a reliable inflation hedge, and, worst of all, does not constitute, not even remotely, a safe haven for one's investments, a shield against government tyranny, or a tail protection vehicle for catastrophic episodes. Furthermore, bitcoin promoters appear to conflate the success of a payment mechanism (as a decentralized mode of exchange), which so far has failed, with the speculative variations in the price of a zero-sum maximally fragile asset with massive negative externalities. Going through monetary history, we show how a true numeraire must be one of minimum variance with respect to an arbitrary basket of goods and services, how gold and silver lost their inflation hedge status during the Hunt brothers squeeze in the late 1970s and what would be required from a true inflation hedged store of value.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-08", "2022-01", "2022-04", "2023-05"], "last_mentioned": "2023-05-13T22:44:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2108.04575", "slug": "a2de388640934fe7", "title": "One Glitch to Rule Them All: Fault Injection Attacks Against AMD's Secure Encrypted Virtualization", "authors": ["Robert Buhren", "Hans Niklas Jacob", "Thilo Krachenfels", "Jean-Pierre Seifert"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2108.04575", "doi": null, "topics": ["Security & cryptography", "Computer systems", "Economics"], "description": "AMD Secure Encrypted Virtualization (SEV) offers protection mechanisms for virtual machines in untrusted environments through memory and register encryption.", "description_method": "abstract-excerpt", "abstract": "AMD Secure Encrypted Virtualization (SEV) offers protection mechanisms for virtual machines in untrusted environments through memory and register encryption. To separate security-sensitive operations from software executing on the main x86 cores, SEV leverages the AMD Secure Processor (AMD-SP). This paper introduces a new approach to attack SEV-protected virtual machines (VMs) by targeting the AMD-SP. We present a voltage glitching attack that allows an attacker to execute custom payloads on the AMD-SPs of all microarchitectures that support SEV currently on the market (Zen 1, Zen 2, and Zen 3). The presented methods allow us to deploy a custom SEV firmware on the AMD-SP, which enables an adversary to decrypt a VM's memory. Furthermore, using our approach, we can extract endorsement keys of SEV-enabled CPUs, which allows us to fake attestation reports or to pose as a valid target for VM migration without requiring physical access to the target host. Moreover, we reverse-engineered the Versioned Chip Endorsement Key (VCEK) mechanism introduced with SEV Secure Nested Paging (SEV-SNP). The VCEK binds the endorsement keys to the firmware version of TCB components relevant for SEV. Building on the ability to extract the endorsement keys, we show how to derive valid VCEKs for arbitrary firmware versions. With our findings, we prove that SEV cannot adequately protect confidential data in cloud environments from insider attackers, such as rogue administrators, on currently available CPUs.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2022-04", "2023-08", "2025-05", "2025-08"], "last_mentioned": "2025-08-18T15:54:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2108.09293", "slug": "68d6a235b4ee1cbb", "title": "Asleep at the Keyboard? Assessing the Security of GitHub Copilot's Code Contributions", "authors": ["Hammond Pearce", "Baleegh Ahmad", "Benjamin Tan", "Brendan Dolan-Gavitt", "Ramesh Karri"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2108.09293", "doi": null, "topics": ["Security & cryptography", "Language models"], "description": "There is burgeoning interest in designing AI-based systems to assist humans in designing computing systems, including tools that automatically generate computer code.", "description_method": "abstract-excerpt", "abstract": "There is burgeoning interest in designing AI-based systems to assist humans in designing computing systems, including tools that automatically generate computer code. The most notable of these comes in the form of the first self-described `AI pair programmer', GitHub Copilot, a language model trained over open-source GitHub code. However, code often contains bugs - and so, given the vast quantity of unvetted code that Copilot has processed, it is certain that the language model will have learned from exploitable, buggy code. This raises concerns on the security of Copilot's code contributions. In this work, we systematically investigate the prevalence and conditions that can cause GitHub Copilot to recommend insecure code. To perform this analysis we prompt Copilot to generate code in scenarios relevant to high-risk CWEs (e.g. those from MITRE's \"Top 25\" list). We explore Copilot's performance on three distinct code generation axes -- examining how it performs given diversity of weaknesses, diversity of prompts, and diversity of domains. In total, we produce 89 different scenarios for Copilot to complete, producing 1,689 programs. Of these, we found approximately 40% to be vulnerable.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2021-10", "2022-06", "2022-12", "2023-01"], "last_mentioned": "2023-01-08T20:41:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2110.10086", "slug": "0a94fa2898f5cecc", "title": "Three Attacks on Proof-of-Stake Ethereum", "authors": ["Caspar Schwarz-Schilling", "Joachim Neu", "Barnabé Monnot", "Aditya Asgaonkar", "Ertem Nusret Tas", "David Tse"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2110.10086", "doi": null, "topics": ["Security & cryptography", "Distributed systems", "Networks"], "description": "Recently, two attacks were presented against Proof-of-Stake (PoS) Ethereum: one where short-range reorganizations of the underlying consensus chain are used to increase individual validators' profits and delay consensus decisions, and one where adversarial network delay is leveraged to stall consensus decisions indefinitely.", "description_method": "abstract-excerpt", "abstract": "Recently, two attacks were presented against Proof-of-Stake (PoS) Ethereum: one where short-range reorganizations of the underlying consensus chain are used to increase individual validators' profits and delay consensus decisions, and one where adversarial network delay is leveraged to stall consensus decisions indefinitely. We provide refined variants of these attacks, considerably relaxing the requirements on adversarial stake and network timing, and thus rendering the attacks more severe. Combining techniques from both refined attacks, we obtain a third attack which allows an adversary with vanishingly small fraction of stake and no control over network message propagation (assuming instead probabilistic message propagation) to cause even long-range consensus chain reorganizations. Honest-but-rational or ideologically motivated validators could use this attack to increase their profits or stall the protocol, threatening incentive alignment and security of PoS Ethereum. The attack can also lead to destabilization of consensus from congestion in vote processing.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-10", "2021-11", "2022-03"], "last_mentioned": "2022-03-11T00:54:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2111.09259", "slug": "fa8b8a236d1662d9", "title": "Acquisition of Chess Knowledge in AlphaZero", "authors": ["Thomas McGrath", "Andrei Kapishnikov", "Nenad Tomašev", "Adam Pearce", "Demis Hassabis", "Been Kim", "Ulrich Paquet", "Vladimir Kramnik"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2111.09259", "doi": "10.1073/pnas.2206625119", "topics": ["Machine learning", "Psychology", "Science & research"], "description": "What is learned by sophisticated neural network agents such as AlphaZero?", "description_method": "abstract-excerpt", "abstract": "What is learned by sophisticated neural network agents such as AlphaZero? This question is of both scientific and practical interest. If the representations of strong neural networks bear no resemblance to human concepts, our ability to understand faithful explanations of their decisions will be restricted, ultimately limiting what we can achieve with neural network interpretability. In this work we provide evidence that human knowledge is acquired by the AlphaZero neural network as it trains on the game of chess. By probing for a broad range of human chess concepts we show when and where these concepts are represented in the AlphaZero network. We also provide a behavioural analysis focusing on opening play, including qualitative analysis from chess Grandmaster Vladimir Kramnik. Finally, we carry out a preliminary investigation looking at the low-level details of AlphaZero's representations, and make the resulting behavioural and representational analyses available online.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-11", "2021-12", "2026-02"], "last_mentioned": "2026-02-09T15:57:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2201.04600", "slug": "be176e10a3723176", "title": "Deep Symbolic Regression for Recurrent Sequences", "authors": ["Stéphane d'Ascoli", "Pierre-Alexandre Kamienny", "Guillaume Lample", "François Charton"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2201.04600", "doi": null, "topics": ["Statistics & probability", "Mathematics", "Machine learning"], "description": "Symbolic regression, i.e.", "description_method": "abstract-excerpt", "abstract": "Symbolic regression, i.e. predicting a function from the observation of its values, is well-known to be a challenging task. In this paper, we train Transformers to infer the function or recurrence relation underlying sequences of integers or floats, a typical task in human IQ tests which has hardly been tackled in the machine learning literature. We evaluate our integer model on a subset of OEIS sequences, and show that it outperforms built-in Mathematica functions for recurrence prediction. We also demonstrate that our float model is able to yield informative approximations of out-of-vocabulary functions and constants, e.g. $\\operatorname{bessel0}(x)\\approx \\frac{\\sin(x)+\\cos(x)}{\\sqrt{πx}}$ and $1.644934\\approx π^2/6$. An interactive demonstration of our models is provided at https://symbolicregression.metademolab.com.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-01", "2022-02", "2022-03"], "last_mentioned": "2022-03-29T19:46:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2204.05149", "slug": "4bd55e909f5607f5", "title": "The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink", "authors": ["David Patterson", "Joseph Gonzalez", "Urs Hölzle", "Quoc Le", "Chen Liang", "Lluis-Miquel Munguia", "Daniel Rothchild", "David So", "Maud Texier", "Jeff Dean"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2204.05149", "doi": null, "topics": ["Machine learning"], "description": "Machine Learning (ML) workloads have rapidly grown in importance, but raised concerns about their carbon footprint.", "description_method": "abstract-excerpt", "abstract": "Machine Learning (ML) workloads have rapidly grown in importance, but raised concerns about their carbon footprint. Four best practices can reduce ML training energy by up to 100x and CO2 emissions up to 1000x. By following best practices, overall ML energy use (across research, development, and production) held steady at <15% of Google's total energy use for the past three years. If the whole ML field were to adopt best practices, total carbon emissions from training would reduce. Hence, we recommend that ML papers include emissions explicitly to foster competition on more than just model quality. Estimates of emissions in papers that omitted them have been off 100x-100,000x, so publishing emissions has the added benefit of ensuring accurate accounting. Given the importance of climate change, we must get the numbers right to make certain that we work on its biggest challenges.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2023-01", "2025-07", "2025-08"], "last_mentioned": "2025-08-21T17:26:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2204.06974", "slug": "ca4a3a4447ee249c", "title": "Planting Undetectable Backdoors in Machine Learning Models", "authors": ["Shafi Goldwasser", "Michael P. Kim", "Vinod Vaikuntanathan", "Or Zamir"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2204.06974", "doi": null, "topics": ["Machine learning"], "description": "Given the computational cost and technical expertise required to train machine learning models, users may delegate the task of learning to a service provider.", "description_method": "abstract-excerpt", "abstract": "Given the computational cost and technical expertise required to train machine learning models, users may delegate the task of learning to a service provider. We show how a malicious learner can plant an undetectable backdoor into a classifier. On the surface, such a backdoored classifier behaves normally, but in reality, the learner maintains a mechanism for changing the classification of any input, with only a slight perturbation. Importantly, without the appropriate \"backdoor key\", the mechanism is hidden and cannot be detected by any computationally-bounded observer. We demonstrate two frameworks for planting undetectable backdoors, with incomparable guarantees. First, we show how to plant a backdoor in any model, using digital signature schemes. The construction guarantees that given black-box access to the original model and the backdoored version, it is computationally infeasible to find even a single input where they differ. This property implies that the backdoored model has generalization error comparable with the original model. Second, we demonstrate how to insert undetectable backdoors in models trained using the Random Fourier Features (RFF) learning paradigm or in Random ReLU networks. In this construction, undetectability holds against powerful white-box distinguishers: given a complete description of the network and the training data, no efficient distinguisher can guess whether the model is \"clean\" or contains a backdoor. Our construction of undetectable backdoors also sheds light on the related issue of robustness to adversarial examples. In particular, our construction can produce a classifier that is indistinguishable from an \"adversarially robust\" classifier, but where every input has an adversarial example! In summary, the existence of undetectable backdoors represent a significant theoretical roadblock to certifying adversarial robustness.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2022-04", "2022-05", "2022-06", "2025-12"], "last_mentioned": "2025-12-29T08:13:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2207.14255", "slug": "4341f9fc5064200d", "title": "Efficient Training of Language Models to Fill in the Middle", "authors": ["Mohammad Bavarian", "Heewoo Jun", "Nikolas Tezak", "John Schulman", "Christine McLeavey", "Jerry Tworek", "Mark Chen"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2207.14255", "doi": null, "topics": ["Language models"], "description": "We show that autoregressive language models can learn to infill text after we apply a straightforward transformation to the dataset, which simply moves a span of text from the middle of a document to its end.", "description_method": "abstract-excerpt", "abstract": "We show that autoregressive language models can learn to infill text after we apply a straightforward transformation to the dataset, which simply moves a span of text from the middle of a document to its end. While this data augmentation has garnered much interest in recent years, we provide extensive evidence that training models with a large fraction of data transformed in this way does not harm the original left-to-right generative capability, as measured by perplexity and sampling evaluations across a wide range of scales. Given the usefulness, simplicity, and efficiency of training models to fill-in-the-middle (FIM), we suggest that future autoregressive language models be trained with FIM by default. To this end, we run a series of ablations on key hyperparameters, such as the data transformation frequency, the structure of the transformation, and the method of selecting the infill span. We use these ablations to prescribe strong default settings and best practices to train FIM models. We have released our best infilling model trained with best practices in our API, and release our infilling benchmarks to aid future research.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-08", "2022-12", "2023-03", "2025-01"], "last_mentioned": "2025-01-07T06:13:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2208.14755", "slug": "1ead9497d0d4e5fe", "title": "Python Type Hints are Turing Complete", "authors": ["Ori Roth"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2208.14755", "doi": null, "topics": ["Algorithms"], "description": "Grigore showed that Java generics are Turing complete by describing a reduction from Turing machines to Java subtyping.", "description_method": "abstract-excerpt", "abstract": "Grigore showed that Java generics are Turing complete by describing a reduction from Turing machines to Java subtyping. We apply Grigore's algorithm to Python type hints and deduce that they are Turing complete. In addition, we present an alternative reduction in which the Turing machines are simulated in real time, resulting in significantly lower compilation times. Our work is accompanied by a Python implementation of both reductions that compiles Turing machines into Python subtyping machines.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2022-09", "2023-03", "2023-09"], "last_mentioned": "2023-09-18T13:34:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2210.11416", "slug": "ec67741e42a2294a", "title": "Scaling Instruction-Finetuned Language Models", "authors": ["Hyung Won Chung", "Le Hou", "Shayne Longpre", "Barret Zoph", "Yi Tay", "William Fedus", "Yunxuan Li", "Xuezhi Wang", "Mostafa Dehghani", "Siddhartha Brahma", "Albert Webson", "Shixiang Shane Gu", "Zhuyun Dai", "Mirac Suzgun", "Xinyun Chen", "Aakanksha Chowdhery", "Alex Castro-Ros", "Marie Pellat", "Kevin Robinson", "Dasha Valter", "Sharan Narang", "Gaurav Mishra", "Adams Yu", "Vincent Zhao", "Yanping Huang", "Andrew Dai", "Hongkun Yu", "Slav Petrov", "Ed H. Chi", "Jeff Dean", "Jacob Devlin", "Adam Roberts", "Denny Zhou", "Quoc V. Le", "Jason Wei"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2210.11416", "doi": null, "topics": ["Language models", "Machine learning", "Human–computer interaction"], "description": "Finetuning language models on a collection of datasets phrased as instructions has been shown to improve model performance and generalization to unseen tasks.", "description_method": "abstract-excerpt", "abstract": "Finetuning language models on a collection of datasets phrased as instructions has been shown to improve model performance and generalization to unseen tasks. In this paper we explore instruction finetuning with a particular focus on (1) scaling the number of tasks, (2) scaling the model size, and (3) finetuning on chain-of-thought data. We find that instruction finetuning with the above aspects dramatically improves performance on a variety of model classes (PaLM, T5, U-PaLM), prompting setups (zero-shot, few-shot, CoT), and evaluation benchmarks (MMLU, BBH, TyDiQA, MGSM, open-ended generation). For instance, Flan-PaLM 540B instruction-finetuned on 1.8K tasks outperforms PALM 540B by a large margin (+9.4% on average). Flan-PaLM 540B achieves state-of-the-art performance on several benchmarks, such as 75.2% on five-shot MMLU. We also publicly release Flan-T5 checkpoints, which achieve strong few-shot performance even compared to much larger models, such as PaLM 62B. Overall, instruction finetuning is a general method for improving the performance and usability of pretrained language models.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-10", "2022-12", "2023-02", "2023-05"], "last_mentioned": "2023-05-09T00:45:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2211.00241", "slug": "7e783891a5635cc2", "title": "Adversarial Policies Beat Superhuman Go AIs", "authors": ["Tony T. Wang", "Adam Gleave", "Tom Tseng", "Kellin Pelrine", "Nora Belrose", "Joseph Miller", "Michael D. Dennis", "Yawen Duan", "Viktor Pogrebniak", "Sergey Levine", "Stuart Russell"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2211.00241", "doi": null, "topics": ["Security & cryptography", "Algorithms"], "description": "We attack the state-of-the-art Go-playing AI system KataGo by training adversarial policies against it, achieving a >97% win rate against KataGo running at superhuman settings.", "description_method": "abstract-excerpt", "abstract": "We attack the state-of-the-art Go-playing AI system KataGo by training adversarial policies against it, achieving a >97% win rate against KataGo running at superhuman settings. Our adversaries do not win by playing Go well. Instead, they trick KataGo into making serious blunders. Our attack transfers zero-shot to other superhuman Go-playing AIs, and is comprehensible to the extent that human experts can implement it without algorithmic assistance to consistently beat superhuman AIs. The core vulnerability uncovered by our attack persists even in KataGo agents adversarially trained to defend against our attack. Our results demonstrate that even superhuman AI systems may harbor surprising failure modes. Example games are available https://goattack.far.ai/.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-02", "2023-04", "2023-11"], "last_mentioned": "2023-11-17T21:14:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.09748", "slug": "f52b3a82c7d82ee2", "title": "Scalable Diffusion Models with Transformers", "authors": ["William Peebles", "Saining Xie"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.09748", "doi": null, "topics": ["Physics", "Language models", "Machine learning"], "description": "We explore a new class of diffusion models based on the transformer architecture.", "description_method": "abstract-excerpt", "abstract": "We explore a new class of diffusion models based on the transformer architecture. We train latent diffusion models of images, replacing the commonly-used U-Net backbone with a transformer that operates on latent patches. We analyze the scalability of our Diffusion Transformers (DiTs) through the lens of forward pass complexity as measured by Gflops. We find that DiTs with higher Gflops -- through increased transformer depth/width or increased number of input tokens -- consistently have lower FID. In addition to possessing good scalability properties, our largest DiT-XL/2 models outperform all prior diffusion models on the class-conditional ImageNet 512x512 and 256x256 benchmarks, achieving a state-of-the-art FID of 2.27 on the latter.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-02", "2025-05", "2025-06"], "last_mentioned": "2025-06-29T20:59:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.13138", "slug": "c2594c387f02fc39", "title": "Large Language Models Encode Clinical Knowledge", "authors": ["Karan Singhal", "Shekoofeh Azizi", "Tao Tu", "S. Sara Mahdavi", "Jason Wei", "Hyung Won Chung", "Nathan Scales", "Ajay Tanwani", "Heather Cole-Lewis", "Stephen Pfohl", "Perry Payne", "Martin Seneviratne", "Paul Gamble", "Chris Kelly", "Nathaneal Scharli", "Aakanksha Chowdhery", "Philip Mansfield", "Blaise Aguera y Arcas", "Dale Webster", "Greg S. Corrado", "Yossi Matias", "Katherine Chou", "Juraj Gottweis", "Nenad Tomasev", "Yun Liu", "Alvin Rajkomar", "Joelle Barral", "Christopher Semturs", "Alan Karthikesalingam", "Vivek Natarajan"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.13138", "doi": null, "topics": ["Biology & medicine", "Language models", "Psychology"], "description": "Large language models (LLMs) have demonstrated impressive capabilities in natural language understanding and generation, but the quality bar for medical and clinical applications is high.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLMs) have demonstrated impressive capabilities in natural language understanding and generation, but the quality bar for medical and clinical applications is high. Today, attempts to assess models' clinical knowledge typically rely on automated evaluations on limited benchmarks. There is no standard to evaluate model predictions and reasoning across a breadth of tasks. To address this, we present MultiMedQA, a benchmark combining six existing open question answering datasets spanning professional medical exams, research, and consumer queries; and HealthSearchQA, a new free-response dataset of medical questions searched online. We propose a framework for human evaluation of model answers along multiple axes including factuality, precision, possible harm, and bias. In addition, we evaluate PaLM (a 540-billion parameter LLM) and its instruction-tuned variant, Flan-PaLM, on MultiMedQA. Using a combination of prompting strategies, Flan-PaLM achieves state-of-the-art accuracy on every MultiMedQA multiple-choice dataset (MedQA, MedMCQA, PubMedQA, MMLU clinical topics), including 67.6% accuracy on MedQA (US Medical License Exam questions), surpassing prior state-of-the-art by over 17%. However, human evaluation reveals key gaps in Flan-PaLM responses. To resolve this we introduce instruction prompt tuning, a parameter-efficient approach for aligning LLMs to new domains using a few exemplars. The resulting model, Med-PaLM, performs encouragingly, but remains inferior to clinicians. We show that comprehension, recall of knowledge, and medical reasoning improve with model scale and instruction prompt tuning, suggesting the potential utility of LLMs in medicine. Our human evaluations reveal important limitations of today's models, reinforcing the importance of both evaluation frameworks and method development in creating safe, helpful LLM models for clinical applications.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2022-12", "2023-01", "2023-03"], "last_mentioned": "2023-03-31T02:04:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.10743", "slug": "fb13bdd7166a2a0b", "title": "Tighter Bounds on the Expressivity of Transformer Encoders", "authors": ["David Chiang", "Peter Cholak", "Anand Pillay"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.10743", "doi": null, "topics": ["Language models", "Machine learning"], "description": "Characterizing neural networks in terms of better-understood formal systems has the potential to yield new insights into the power and limitations of these networks.", "description_method": "abstract-excerpt", "abstract": "Characterizing neural networks in terms of better-understood formal systems has the potential to yield new insights into the power and limitations of these networks. Doing so for transformers remains an active area of research. Bhattamishra and others have shown that transformer encoders are at least as expressive as a certain kind of counter machine, while Merrill and Sabharwal have shown that fixed-precision transformer encoders recognize only languages in uniform $TC^0$. We connect and strengthen these results by identifying a variant of first-order logic with counting quantifiers that is simultaneously an upper bound for fixed-precision transformer encoders and a lower bound for transformer encoders. This brings us much closer than before to an exact characterization of the languages that transformer encoders recognize.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2023-06", "2023-12", "2024-01"], "last_mentioned": "2024-01-09T12:05:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2302.06590", "slug": "54fe24811e759b1d", "title": "The Impact of AI on Developer Productivity: Evidence from GitHub Copilot", "authors": ["Sida Peng", "Eirini Kalliamvakou", "Peter Cihon", "Mert Demirer"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2302.06590", "doi": null, "topics": ["Other research"], "description": "Generative AI tools hold promise to increase human productivity.", "description_method": "abstract-excerpt", "abstract": "Generative AI tools hold promise to increase human productivity. This paper presents results from a controlled experiment with GitHub Copilot, an AI pair programmer. Recruited software developers were asked to implement an HTTP server in JavaScript as quickly as possible. The treatment group, with access to the AI pair programmer, completed the task 55.8% faster than the control group. Observed heterogenous effects show promise for AI pair programmers to help people transition into software development careers.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-06", "2025-08", "2026-01"], "last_mentioned": "2026-01-01T22:16:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.14342", "slug": "afd7fbad6232c680", "title": "Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training", "authors": ["Hong Liu", "Zhiyuan Li", "David Hall", "Percy Liang", "Tengyu Ma"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.14342", "doi": null, "topics": ["Language models", "Statistics & probability", "Algorithms"], "description": "Given the massive cost of language model pre-training, a non-trivial improvement of the optimization algorithm would lead to a material reduction on the time and cost of training.", "description_method": "abstract-excerpt", "abstract": "Given the massive cost of language model pre-training, a non-trivial improvement of the optimization algorithm would lead to a material reduction on the time and cost of training. Adam and its variants have been state-of-the-art for years, and more sophisticated second-order (Hessian-based) optimizers often incur too much per-step overhead. In this paper, we propose Sophia, Second-order Clipped Stochastic Optimization, a simple scalable second-order optimizer that uses a light-weight estimate of the diagonal Hessian as the pre-conditioner. The update is the moving average of the gradients divided by the moving average of the estimated Hessian, followed by element-wise clipping. The clipping controls the worst-case update size and tames the negative impact of non-convexity and rapid change of Hessian along the trajectory. Sophia only estimates the diagonal Hessian every handful of iterations, which has negligible average per-step time and memory overhead. On language modeling with GPT models of sizes ranging from 125M to 1.5B, Sophia achieves a 2x speed-up compared to Adam in the number of steps, total compute, and wall-clock time, achieving the same perplexity with 50% fewer steps, less total compute, and reduced wall-clock time. Theoretically, we show that Sophia, in a much simplified setting, adapts to the heterogeneous curvatures in different parameter dimensions, and thus has a run-time bound that does not depend on the condition number of the loss.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-05", "2023-06", "2023-08"], "last_mentioned": "2023-08-14T14:40:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.18248", "slug": "668e8d5a9aa99b56", "title": "Do Language Models Know When They're Hallucinating References?", "authors": ["Ayush Agrawal", "Mirac Suzgun", "Lester Mackey", "Adam Tauman Kalai"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.18248", "doi": null, "topics": ["Language models", "Distributed systems"], "description": "State-of-the-art language models (LMs) are notoriously susceptible to generating hallucinated information.", "description_method": "abstract-excerpt", "abstract": "State-of-the-art language models (LMs) are notoriously susceptible to generating hallucinated information. Such inaccurate outputs not only undermine the reliability of these models but also limit their use and raise serious concerns about misinformation and propaganda. In this work, we focus on hallucinated book and article references and present them as the \"model organism\" of language model hallucination research, due to their frequent and easy-to-discern nature. We posit that if a language model cites a particular reference in its output, then it should ideally possess sufficient information about its authors and content, among other relevant details. Using this basic insight, we illustrate that one can identify hallucinated references without ever consulting any external resources, by asking a set of direct or indirect queries to the language model about the references. These queries can be considered as \"consistency checks.\" Our findings highlight that while LMs, including GPT-4, often produce inconsistent author lists for hallucinated references, they also often accurately recall the authors of real references. In this sense, the LM can be said to \"know\" when it is hallucinating references. Furthermore, these findings show how hallucinated references can be dissected to shed light on their nature. Replication code and results can be found at https://github.com/microsoft/hallucinated-references.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-08", "2023-09", "2023-10", "2025-03"], "last_mentioned": "2025-03-20T05:09:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2305.20050", "slug": "20505c81992a1da0", "title": "Let's Verify Step by Step", "authors": ["Hunter Lightman", "Vineet Kosaraju", "Yura Burda", "Harri Edwards", "Bowen Baker", "Teddy Lee", "Jan Leike", "John Schulman", "Ilya Sutskever", "Karl Cobbe"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2305.20050", "doi": null, "topics": ["Language models"], "description": "In recent years, large language models have greatly improved in their ability to perform complex multi-step reasoning.", "description_method": "abstract-excerpt", "abstract": "In recent years, large language models have greatly improved in their ability to perform complex multi-step reasoning. However, even state-of-the-art models still regularly produce logical mistakes. To train more reliable models, we can turn either to outcome supervision, which provides feedback for a final result, or process supervision, which provides feedback for each intermediate reasoning step. Given the importance of training reliable models, and given the high cost of human feedback, it is important to carefully compare the both methods. Recent work has already begun this comparison, but many questions still remain. We conduct our own investigation, finding that process supervision significantly outperforms outcome supervision for training models to solve problems from the challenging MATH dataset. Our process-supervised model solves 78% of problems from a representative subset of the MATH test set. Additionally, we show that active learning significantly improves the efficacy of process supervision. To support related research, we also release PRM800K, the complete dataset of 800,000 step-level human feedback labels used to train our best reward model.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-11", "2024-08", "2025-01"], "last_mentioned": "2025-01-28T17:53:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.03341", "slug": "c524fcec2ec39a11", "title": "Inference-Time Intervention: Eliciting Truthful Answers from a Language Model", "authors": ["Kenneth Li", "Oam Patel", "Fernanda Viégas", "Hanspeter Pfister", "Martin Wattenberg"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.03341", "doi": null, "topics": ["Language models"], "description": "We introduce Inference-Time Intervention (ITI), a technique designed to enhance the \"truthfulness\" of large language models (LLMs).", "description_method": "abstract-excerpt", "abstract": "We introduce Inference-Time Intervention (ITI), a technique designed to enhance the \"truthfulness\" of large language models (LLMs). ITI operates by shifting model activations during inference, following a set of directions across a limited number of attention heads. This intervention significantly improves the performance of LLaMA models on the TruthfulQA benchmark. On an instruction-finetuned LLaMA called Alpaca, ITI improves its truthfulness from 32.5% to 65.1%. We identify a tradeoff between truthfulness and helpfulness and demonstrate how to balance it by tuning the intervention strength. ITI is minimally invasive and computationally inexpensive. Moreover, the technique is data efficient: while approaches like RLHF require extensive annotations, ITI locates truthful directions using only few hundred examples. Our findings suggest that LLMs may have an internal representation of the likelihood of something being true, even as they produce falsehoods on the surface.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-06", "2024-04", "2024-07", "2025-08"], "last_mentioned": "2025-08-03T17:52:57Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.04050", "slug": "3c52f3f021751b47", "title": "LLMZip: Lossless Text Compression using Large Language Models", "authors": ["Chandra Shekhara Kaushik Valmeekam", "Krishna Narayanan", "Dileep Kalathil", "Jean-Francois Chamberland", "Srinivas Shakkottai"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.04050", "doi": null, "topics": ["Language models", "Statistics & probability", "Algorithms"], "description": "We provide new estimates of an asymptotic upper bound on the entropy of English using the large language model LLaMA-7B as a predictor for the next token given a window of past tokens.", "description_method": "abstract-excerpt", "abstract": "We provide new estimates of an asymptotic upper bound on the entropy of English using the large language model LLaMA-7B as a predictor for the next token given a window of past tokens. This estimate is significantly smaller than currently available estimates in \\cite{cover1978convergent}, \\cite{lutati2023focus}. A natural byproduct is an algorithm for lossless compression of English text which combines the prediction from the large language model with a lossless compression scheme. Preliminary results from limited experiments suggest that our scheme outperforms state-of-the-art text compression schemes such as BSC, ZPAQ, and paq8h.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-07", "2024-03", "2024-09", "2025-07"], "last_mentioned": "2025-07-21T21:46:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.05685", "slug": "3f96ad00553be20b", "title": "Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena", "authors": ["Lianmin Zheng", "Wei-Lin Chiang", "Ying Sheng", "Siyuan Zhuang", "Zhanghao Wu", "Yonghao Zhuang", "Zi Lin", "Zhuohan Li", "Dacheng Li", "Eric P. Xing", "Hao Zhang", "Joseph E. Gonzalez", "Ion Stoica"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.05685", "doi": null, "topics": ["Language models", "Psychology"], "description": "Evaluating large language model (LLM) based chat assistants is challenging due to their broad capabilities and the inadequacy of existing benchmarks in measuring human preferences.", "description_method": "abstract-excerpt", "abstract": "Evaluating large language model (LLM) based chat assistants is challenging due to their broad capabilities and the inadequacy of existing benchmarks in measuring human preferences. To address this, we explore using strong LLMs as judges to evaluate these models on more open-ended questions. We examine the usage and limitations of LLM-as-a-judge, including position, verbosity, and self-enhancement biases, as well as limited reasoning ability, and propose solutions to mitigate some of them. We then verify the agreement between LLM judges and human preferences by introducing two benchmarks: MT-bench, a multi-turn question set; and Chatbot Arena, a crowdsourced battle platform. Our results reveal that strong LLM judges like GPT-4 can match both controlled and crowdsourced human preferences well, achieving over 80% agreement, the same level of agreement between humans. Hence, LLM-as-a-judge is a scalable and explainable way to approximate human preferences, which are otherwise very expensive to obtain. Additionally, we show our benchmark and traditional benchmarks complement each other by evaluating several variants of LLaMA and Vicuna. The MT-bench questions, 3K expert votes, and 30K conversations with human preferences are publicly available at https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-07", "2023-10", "2025-04", "2025-06"], "last_mentioned": "2025-06-26T14:41:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2307.15043", "slug": "d9727f1325e9981f", "title": "Universal and Transferable Adversarial Attacks on Aligned Language Models", "authors": ["Andy Zou", "Zifan Wang", "Nicholas Carlini", "Milad Nasr", "J. Zico Kolter", "Matt Fredrikson"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2307.15043", "doi": null, "topics": ["Security & cryptography", "Language models", "Statistics & probability"], "description": "Because \"out-of-the-box\" large language models are capable of generating a great deal of objectionable content, recent work has focused on aligning these models in an attempt to prevent undesirable generation.", "description_method": "abstract-excerpt", "abstract": "Because \"out-of-the-box\" large language models are capable of generating a great deal of objectionable content, recent work has focused on aligning these models in an attempt to prevent undesirable generation. While there has been some success at circumventing these measures -- so-called \"jailbreaks\" against LLMs -- these attacks have required significant human ingenuity and are brittle in practice. In this paper, we propose a simple and effective attack method that causes aligned language models to generate objectionable behaviors. Specifically, our approach finds a suffix that, when attached to a wide range of queries for an LLM to produce objectionable content, aims to maximize the probability that the model produces an affirmative response (rather than refusing to answer). However, instead of relying on manual engineering, our approach automatically produces these adversarial suffixes by a combination of greedy and gradient-based search techniques, and also improves over past automatic prompt generation methods. Surprisingly, we find that the adversarial prompts generated by our approach are quite transferable, including to black-box, publicly released LLMs. Specifically, we train an adversarial attack suffix on multiple prompts (i.e., queries asking for many different types of objectionable content), as well as multiple models (in our case, Vicuna-7B and 13B). When doing so, the resulting attack suffix is able to induce objectionable content in the public interfaces to ChatGPT, Bard, and Claude, as well as open source LLMs such as LLaMA-2-Chat, Pythia, Falcon, and others. In total, this work significantly advances the state-of-the-art in adversarial attacks against aligned language models, raising important questions about how such systems can be prevented from producing objectionable information. Code is available at github.com/llm-attacks/llm-attacks.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2023-08", "2023-12", "2024-01", "2025-08"], "last_mentioned": "2025-08-16T06:47:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2308.03296", "slug": "ef621e6b852fbeb4", "title": "Studying Large Language Model Generalization with Influence Functions", "authors": ["Roger Grosse", "Juhan Bae", "Cem Anil", "Nelson Elhage", "Alex Tamkin", "Amirhossein Tajdini", "Benoit Steiner", "Dustin Li", "Esin Durmus", "Ethan Perez", "Evan Hubinger", "Kamilė Lukošiūtė", "Karina Nguyen", "Nicholas Joseph", "Sam McCandlish", "Jared Kaplan", "Samuel R. Bowman"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2308.03296", "doi": null, "topics": ["Machine learning", "Language models", "Psychology"], "description": "When trying to gain better visibility into a machine learning model in order to understand and mitigate the associated risks, a potentially valuable source of evidence is: which training examples most contribute to a given behavior?", "description_method": "abstract-excerpt", "abstract": "When trying to gain better visibility into a machine learning model in order to understand and mitigate the associated risks, a potentially valuable source of evidence is: which training examples most contribute to a given behavior? Influence functions aim to answer a counterfactual: how would the model's parameters (and hence its outputs) change if a given sequence were added to the training set? While influence functions have produced insights for small models, they are difficult to scale to large language models (LLMs) due to the difficulty of computing an inverse-Hessian-vector product (IHVP). We use the Eigenvalue-corrected Kronecker-Factored Approximate Curvature (EK-FAC) approximation to scale influence functions up to LLMs with up to 52 billion parameters. In our experiments, EK-FAC achieves similar accuracy to traditional influence function estimators despite the IHVP computation being orders of magnitude faster. We investigate two algorithmic techniques to reduce the cost of computing gradients of candidate training sequences: TF-IDF filtering and query batching. We use influence functions to investigate the generalization patterns of LLMs, including the sparsity of the influence patterns, increasing abstraction with scale, math and programming abilities, cross-lingual generalization, and role-playing behavior. Despite many apparently sophisticated forms of generalization, we identify a surprising limitation: influences decay to near-zero when the order of key phrases is flipped. Overall, influence functions give us a powerful new tool for studying the generalization properties of LLMs.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2023-11", "2023-12", "2026-07"], "last_mentioned": "2026-07-24T21:17:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2309.06180", "slug": "26c336c25455ffe5", "title": "Efficient Memory Management for Large Language Model Serving with PagedAttention", "authors": ["Woosuk Kwon", "Zhuohan Li", "Siyuan Zhuang", "Ying Sheng", "Lianmin Zheng", "Cody Hao Yu", "Joseph E. Gonzalez", "Hao Zhang", "Ion Stoica"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2309.06180", "doi": null, "topics": ["Language models", "Computer systems", "Algorithms"], "description": "High throughput serving of large language models (LLMs) requires batching sufficiently many requests at a time.", "description_method": "abstract-excerpt", "abstract": "High throughput serving of large language models (LLMs) requires batching sufficiently many requests at a time. However, existing systems struggle because the key-value cache (KV cache) memory for each request is huge and grows and shrinks dynamically. When managed inefficiently, this memory can be significantly wasted by fragmentation and redundant duplication, limiting the batch size. To address this problem, we propose PagedAttention, an attention algorithm inspired by the classical virtual memory and paging techniques in operating systems. On top of it, we build vLLM, an LLM serving system that achieves (1) near-zero waste in KV cache memory and (2) flexible sharing of KV cache within and across requests to further reduce memory usage. Our evaluations show that vLLM improves the throughput of popular LLMs by 2-4$\\times$ with the same level of latency compared to the state-of-the-art systems, such as FasterTransformer and Orca. The improvement is more pronounced with longer sequences, larger models, and more complex decoding algorithms. vLLM's source code is publicly available at https://github.com/vllm-project/vllm", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2024-03", "2024-05", "2025-01", "2026-02"], "last_mentioned": "2026-02-02T14:18:19Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2309.07181", "slug": "76db8ef9dcb244c1", "title": "The Grand Illusion: The Myth of Software Portability and Implications for ML Progress", "authors": ["Fraser Mince", "Dzung Dinh", "Jonas Kgomo", "Neil Thompson", "Sara Hooker"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2309.07181", "doi": null, "topics": ["Machine learning"], "description": "Pushing the boundaries of machine learning often requires exploring different hardware and software combinations.", "description_method": "abstract-excerpt", "abstract": "Pushing the boundaries of machine learning often requires exploring different hardware and software combinations. However, the freedom to experiment across different tooling stacks can be at odds with the drive for efficiency, which has produced increasingly specialized AI hardware and incentivized consolidation around a narrow set of ML frameworks. Exploratory research can be restricted if software and hardware are co-evolving, making it even harder to stray away from mainstream ideas that work well with popular tooling stacks. While this friction increasingly impacts the rate of innovation in machine learning, to our knowledge the lack of portability in tooling has not been quantified. In this work, we ask: How portable are popular ML software frameworks? We conduct a large-scale study of the portability of mainstream ML frameworks across different hardware types. Our findings paint an uncomfortable picture -- frameworks can lose more than 40% of their key functions when ported to other hardware. Worse, even when functions are portable, the slowdown in their performance can be extreme and render performance untenable. Collectively, our results reveal how costly straying from a narrow set of hardware-software combinations can be - and suggest that specialization of hardware impedes innovation in machine learning research.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2023-10", "2024-01", "2024-09"], "last_mentioned": "2024-09-24T00:34:37Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2309.08632", "slug": "deaa876a499e487f", "title": "Pretraining on the Test Set Is All You Need", "authors": ["Rylan Schaeffer"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2309.08632", "doi": null, "topics": ["Language models"], "description": "Inspired by recent work demonstrating the promise of smaller Transformer-based language models pretrained on carefully curated data, we supercharge such approaches by investing heavily in curating a novel, high quality, non-synthetic data mixture based solely on evaluation benchmarks.", "description_method": "abstract-excerpt", "abstract": "Inspired by recent work demonstrating the promise of smaller Transformer-based language models pretrained on carefully curated data, we supercharge such approaches by investing heavily in curating a novel, high quality, non-synthetic data mixture based solely on evaluation benchmarks. Using our novel dataset mixture consisting of less than 100 thousand tokens, we pretrain a 1 million parameter transformer-based LLM \\textbf{phi-CTNL} (pronounced ``fictional\") that achieves perfect results across diverse academic benchmarks, strictly outperforming all known foundation models. \\textbf{phi-CTNL} also beats power-law scaling and exhibits a never-before-seen grokking-like ability to accurately predict downstream evaluation benchmarks' canaries.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2023-10", "2023-11", "2024-03", "2025-02"], "last_mentioned": "2025-02-18T17:41:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2309.16575", "slug": "85a671149fb10e8a", "title": "A Benchmark for Learning to Translate a New Language from One Grammar Book", "authors": ["Garrett Tanzer", "Mirac Suzgun", "Eline Visser", "Dan Jurafsky", "Luke Melas-Kyriazi"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2309.16575", "doi": null, "topics": ["Language models", "Networks"], "description": "Large language models (LLMs) can perform impressive feats with in-context learning or lightweight finetuning.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLMs) can perform impressive feats with in-context learning or lightweight finetuning. It is natural to wonder how well these models adapt to genuinely new tasks, but how does one find tasks that are unseen in internet-scale training sets? We turn to a field that is explicitly motivated and bottlenecked by a scarcity of web data: low-resource languages. In this paper, we introduce MTOB (Machine Translation from One Book), a benchmark for learning to translate between English and Kalamang -- a language with less than 200 speakers and therefore virtually no presence on the web -- using several hundred pages of field linguistics reference materials. This task framing is novel in that it asks a model to learn a language from a single human-readable book of grammar explanations, rather than a large mined corpus of in-domain data, more akin to L2 learning than L1 acquisition. We demonstrate that baselines using current LLMs are promising but fall short of human performance, achieving 44.7 chrF on Kalamang to English translation and 45.8 chrF on English to Kalamang translation, compared to 51.6 and 57.0 chrF by a human who learned Kalamang from the same reference materials. We hope that MTOB will help measure LLM capabilities along a new dimension, and that the methods developed to solve it could help expand access to language technology for underserved communities by leveraging qualitatively different kinds of data than traditional machine translation.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-03", "2024-06", "2024-11", "2025-04"], "last_mentioned": "2025-04-15T00:32:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.01798", "slug": "de57da1de7b7aa98", "title": "Large Language Models Cannot Self-Correct Reasoning Yet", "authors": ["Jie Huang", "Xinyun Chen", "Swaroop Mishra", "Huaixiu Steven Zheng", "Adams Wei Yu", "Xinying Song", "Denny Zhou"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.01798", "doi": null, "topics": ["Language models"], "description": "Large Language Models (LLMs) have emerged as a groundbreaking technology with their unparalleled text generation capabilities across various applications.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) have emerged as a groundbreaking technology with their unparalleled text generation capabilities across various applications. Nevertheless, concerns persist regarding the accuracy and appropriateness of their generated content. A contemporary methodology, self-correction, has been proposed as a remedy to these issues. Building upon this premise, this paper critically examines the role and efficacy of self-correction within LLMs, shedding light on its true potential and limitations. Central to our investigation is the notion of intrinsic self-correction, whereby an LLM attempts to correct its initial responses based solely on its inherent capabilities, without the crutch of external feedback. In the context of reasoning, our research indicates that LLMs struggle to self-correct their responses without external feedback, and at times, their performance even degrades after self-correction. Drawing from these insights, we offer suggestions for future research and practical applications in this field.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-10", "2024-04", "2024-08"], "last_mentioned": "2024-08-11T20:39:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.06770", "slug": "b36ffb44e7a44c44", "title": "SWE-bench: Can Language Models Resolve Real-World GitHub Issues?", "authors": ["Carlos E. Jimenez", "John Yang", "Alexander Wettig", "Shunyu Yao", "Kexin Pei", "Ofir Press", "Karthik Narasimhan"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.06770", "doi": null, "topics": ["Language models", "Programming languages", "Robotics"], "description": "Language models have outpaced our ability to evaluate them effectively, but for their future development it is essential to study the frontier of their capabilities.", "description_method": "abstract-excerpt", "abstract": "Language models have outpaced our ability to evaluate them effectively, but for their future development it is essential to study the frontier of their capabilities. We find real-world software engineering to be a rich, sustainable, and challenging testbed for evaluating the next generation of language models. To this end, we introduce SWE-bench, an evaluation framework consisting of $2,294$ software engineering problems drawn from real GitHub issues and corresponding pull requests across $12$ popular Python repositories. Given a codebase along with a description of an issue to be resolved, a language model is tasked with editing the codebase to address the issue. Resolving issues in SWE-bench frequently requires understanding and coordinating changes across multiple functions, classes, and even files simultaneously, calling for models to interact with execution environments, process extremely long contexts and perform complex reasoning that goes far beyond traditional code generation tasks. Our evaluations show that both state-of-the-art proprietary models and our fine-tuned model SWE-Llama can resolve only the simplest issues. The best-performing model, Claude 2, is able to solve a mere $1.96$% of the issues. Advances on SWE-bench represent steps towards LMs that are more practical, intelligent, and autonomous.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-01", "2025-09", "2025-12", "2026-03"], "last_mentioned": "2026-03-12T21:43:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.20216", "slug": "e5d710e8b1980be7", "title": "Does GPT-4 pass the Turing test?", "authors": ["Cameron R. Jones", "Benjamin K. Bergen"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.20216", "doi": null, "topics": ["Language models", "Psychology"], "description": "We evaluated GPT-4 in a public online Turing test.", "description_method": "abstract-excerpt", "abstract": "We evaluated GPT-4 in a public online Turing test. The best-performing GPT-4 prompt passed in 49.7% of games, outperforming ELIZA (22%) and GPT-3.5 (20%), but falling short of the baseline set by human participants (66%). Participants' decisions were based mainly on linguistic style (35%) and socioemotional traits (27%), supporting the idea that intelligence, narrowly conceived, is not sufficient to pass the Turing test. Participant knowledge about LLMs and number of games played positively correlated with accuracy in detecting AI, suggesting learning and practice as possible strategies to mitigate deception. Despite known limitations as a test of intelligence, we argue that the Turing test continues to be relevant as an assessment of naturalistic communication and deception. AI models with the ability to masquerade as humans could have widespread societal consequences, and we analyse the effectiveness of different strategies and criteria for judging humanlikeness.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-11", "2023-12", "2024-01", "2024-06"], "last_mentioned": "2024-06-10T15:10:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2311.02206", "slug": "b387b8d9b38088fc", "title": "Optimizing Datalog for the GPU", "authors": ["Yihao Sun", "Ahmedur Rahman Shovon", "Thomas Gilray", "Kristopher Micinski", "Sidharth Kumar"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2311.02206", "doi": null, "topics": ["Algorithms", "Databases", "Mathematics"], "description": "Modern Datalog engines (e.g., LogicBlox, Soufflé, ddlog) enable their users to write declarative queries which compute recursive deductions over extensional facts, leaving high-performance operationalization (query planning, semi-naïve evaluation, and parallelization) to the engine.", "description_method": "abstract-excerpt", "abstract": "Modern Datalog engines (e.g., LogicBlox, Soufflé, ddlog) enable their users to write declarative queries which compute recursive deductions over extensional facts, leaving high-performance operationalization (query planning, semi-naïve evaluation, and parallelization) to the engine. Such engines form the backbone of modern high-throughput applications in static analysis, network monitoring, and social-media mining. In this paper, we present a methodology for implementing a modern in-memory Datalog engine on data center GPUs, allowing us to achieve significant (up to 45x) gains compared to Soufflé (a modern CPU-based engine) on context-sensitive points-to analysis of httpd. We present GPUlog, a Datalog engine backend that implements iterated relational algebra kernels over a novel range-indexed data structure we call the hash-indexed sorted array (HISA). HISA combines the algorithmic benefits of incremental range-indexed relations with the raw computation throughput of operations over dense data structures. Our experiments show that GPUlog is significantly faster than CPU-based Datalog engines while achieving a favorable memory footprint compared to contemporary GPU-based joins.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-12", "2024-10", "2024-11", "2026-04"], "last_mentioned": "2026-04-26T02:49:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2311.08105", "slug": "2031e19f5216f803", "title": "DiLoCo: Distributed Low-Communication Training of Language Models", "authors": ["Arthur Douillard", "Qixuan Feng", "Andrei A. Rusu", "Rachita Chhaparia", "Yani Donchev", "Adhiguna Kuncoro", "Marc'Aurelio Ranzato", "Arthur Szlam", "Jiajun Shen"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2311.08105", "doi": null, "topics": ["Language models", "Distributed systems", "Algorithms"], "description": "Large language models (LLM) have become a critical component in many applications of machine learning.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLM) have become a critical component in many applications of machine learning. However, standard approaches to training LLM require a large number of tightly interconnected accelerators, with devices exchanging gradients and other intermediate states at each optimization step. While it is difficult to build and maintain a single computing cluster hosting many accelerators, it might be easier to find several computing clusters each hosting a smaller number of devices. In this work, we propose a distributed optimization algorithm, Distributed Low-Communication (DiLoCo), that enables training of language models on islands of devices that are poorly connected. The approach is a variant of federated averaging, where the number of inner steps is large, the inner optimizer is AdamW, and the outer optimizer is Nesterov momentum. On the widely used C4 dataset, we show that DiLoCo on 8 workers performs as well as fully synchronous optimization while communicating 500 times less. DiLoCo exhibits great robustness to the data distribution of each worker. It is also robust to resources becoming unavailable over time, and vice versa, it can seamlessly leverage resources that become available during training.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-11", "2024-03", "2024-09", "2025-05"], "last_mentioned": "2025-05-12T15:32:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2311.12022", "slug": "399a575cd9d77c2d", "title": "GPQA: A Graduate-Level Google-Proof Q&A Benchmark", "authors": ["David Rein", "Betty Li Hou", "Asa Cooper Stickland", "Jackson Petty", "Richard Yuanzhe Pang", "Julien Dirani", "Julian Michael", "Samuel R. Bowman"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2311.12022", "doi": null, "topics": ["Science & research", "Physics", "Language models"], "description": "We present GPQA, a challenging dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry.", "description_method": "abstract-excerpt", "abstract": "We present GPQA, a challenging dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry. We ensure that the questions are high-quality and extremely difficult: experts who have or are pursuing PhDs in the corresponding domains reach 65% accuracy (74% when discounting clear mistakes the experts identified in retrospect), while highly skilled non-expert validators only reach 34% accuracy, despite spending on average over 30 minutes with unrestricted access to the web (i.e., the questions are \"Google-proof\"). The questions are also difficult for state-of-the-art AI systems, with our strongest GPT-4 based baseline achieving 39% accuracy. If we are to use future AI systems to help us answer very hard questions, for example, when developing new scientific knowledge, we need to develop scalable oversight methods that enable humans to supervise their outputs, which may be difficult even if the supervisors are themselves skilled and knowledgeable. The difficulty of GPQA both for skilled non-experts and frontier AI systems should enable realistic scalable oversight experiments, which we hope can help devise ways for human experts to reliably get truthful information from AI systems that surpass human capabilities.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-03", "2024-04", "2025-03", "2025-12"], "last_mentioned": "2025-12-08T02:09:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2311.16863", "slug": "a2908f2b4d8b5c3e", "title": "Power Hungry Processing: Watts Driving the Cost of AI Deployment?", "authors": ["Alexandra Sasha Luccioni", "Yacine Jernite", "Emma Strubell"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2311.16863", "doi": "10.1145/3630106.3658542", "topics": ["Machine learning"], "description": "Recent years have seen a surge in the popularity of commercial AI products based on generative, multi-purpose AI systems promising a unified approach to building machine learning (ML) models into technology.", "description_method": "abstract-excerpt", "abstract": "Recent years have seen a surge in the popularity of commercial AI products based on generative, multi-purpose AI systems promising a unified approach to building machine learning (ML) models into technology. However, this ambition of ``generality'' comes at a steep cost to the environment, given the amount of energy these systems require and the amount of carbon that they emit. In this work, we propose the first systematic comparison of the ongoing inference cost of various categories of ML systems, covering both task-specific (i.e. finetuned models that carry out a single task) and `general-purpose' models, (i.e. those trained for multiple tasks). We measure deployment cost as the amount of energy and carbon required to perform 1,000 inferences on representative benchmark dataset using these models. We find that multi-purpose, generative architectures are orders of magnitude more expensive than task-specific systems for a variety of tasks, even when controlling for the number of model parameters. We conclude with a discussion around the current trend of deploying multi-purpose generative ML systems, and caution that their utility should be more intentionally weighed against increased costs in terms of energy and emissions. All the data from our study can be accessed via an interactive demo to carry out further exploration and analysis.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-12", "2024-02", "2024-04", "2025-08"], "last_mentioned": "2025-08-05T19:01:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2312.00841", "slug": "ccec85fccad72a72", "title": "Do Black Holes have Singularities?", "authors": ["R. P. Kerr"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2312.00841", "doi": null, "topics": ["Physics"], "description": "There is no proof that black holes contain singularities when they are generated by real physical bodies.", "description_method": "abstract-excerpt", "abstract": "There is no proof that black holes contain singularities when they are generated by real physical bodies. Roger Penrose claimed sixty years ago that trapped surfaces inevitably lead to light rays of finite affine length (FALL's). Penrose and Stephen Hawking then asserted that these must end in actual singularities. When they could not prove this they decreed it to be self evident. It is shown that there are counterexamples through every point in the Kerr metric. These are asymptotic to at least one event horizon and do not end in singularities.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-02", "2024-05", "2024-06", "2024-07"], "last_mentioned": "2024-07-15T19:26:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2401.04088", "slug": "2be3a164d6bbe706", "title": "Mixtral of Experts", "authors": ["Albert Q. Jiang", "Alexandre Sablayrolles", "Antoine Roux", "Arthur Mensch", "Blanche Savary", "Chris Bamford", "Devendra Singh Chaplot", "Diego de las Casas", "Emma Bou Hanna", "Florian Bressand", "Gianna Lengyel", "Guillaume Bour", "Guillaume Lample", "Lélio Renard Lavaud", "Lucile Saulnier", "Marie-Anne Lachaux", "Pierre Stock", "Sandeep Subramanian", "Sophia Yang", "Szymon Antoniak", "Teven Le Scao", "Théophile Gervet", "Thibaut Lavril", "Thomas Wang", "Timothée Lacroix", "William El Sayed"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2401.04088", "doi": null, "topics": ["Language models", "Mathematics"], "description": "We introduce Mixtral 8x7B, a Sparse Mixture of Experts (SMoE) language model.", "description_method": "abstract-excerpt", "abstract": "We introduce Mixtral 8x7B, a Sparse Mixture of Experts (SMoE) language model. Mixtral has the same architecture as Mistral 7B, with the difference that each layer is composed of 8 feedforward blocks (i.e. experts). For every token, at each layer, a router network selects two experts to process the current state and combine their outputs. Even though each token only sees two experts, the selected experts can be different at each timestep. As a result, each token has access to 47B parameters, but only uses 13B active parameters during inference. Mixtral was trained with a context size of 32k tokens and it outperforms or matches Llama 2 70B and GPT-3.5 across all evaluated benchmarks. In particular, Mixtral vastly outperforms Llama 2 70B on mathematics, code generation, and multilingual benchmarks. We also provide a model fine-tuned to follow instructions, Mixtral 8x7B - Instruct, that surpasses GPT-3.5 Turbo, Claude-2.1, Gemini Pro, and Llama 2 70B - chat model on human benchmarks. Both the base and instruct models are released under the Apache 2.0 license.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-02", "2024-11", "2026-07"], "last_mentioned": "2026-07-10T04:37:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.00159", "slug": "07d21ed585c87390", "title": "Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research", "authors": ["Luca Soldaini", "Rodney Kinney", "Akshita Bhagia", "Dustin Schwenk", "David Atkinson", "Russell Authur", "Ben Bogin", "Khyathi Chandu", "Jennifer Dumas", "Yanai Elazar", "Valentin Hofmann", "Ananya Harsh Jha", "Sachin Kumar", "Li Lucy", "Xinxi Lyu", "Nathan Lambert", "Ian Magnusson", "Jacob Morrison", "Niklas Muennighoff", "Aakanksha Naik", "Crystal Nam", "Matthew E. Peters", "Abhilasha Ravichander", "Kyle Richardson", "Zejiang Shen", "Emma Strubell", "Nishant Subramani", "Oyvind Tafjord", "Pete Walsh", "Luke Zettlemoyer", "Noah A. Smith", "Hannaneh Hajishirzi", "Iz Beltagy", "Dirk Groeneveld", "Jesse Dodge", "Kyle Lo"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.00159", "doi": null, "topics": ["Language models", "Science & research"], "description": "Information about pretraining corpora used to train the current best-performing language models is seldom discussed: commercial models rarely detail their data, and even open models are often released without accompanying training data or recipes to reproduce them.", "description_method": "abstract-excerpt", "abstract": "Information about pretraining corpora used to train the current best-performing language models is seldom discussed: commercial models rarely detail their data, and even open models are often released without accompanying training data or recipes to reproduce them. As a result, it is challenging to conduct and advance scientific research on language modeling, such as understanding how training data impacts model capabilities and limitations. To facilitate scientific research on language model pretraining, we curate and release Dolma, a three-trillion-token English corpus, built from a diverse mixture of web content, scientific papers, code, public-domain books, social media, and encyclopedic materials. We extensively document Dolma, including its design principles, details about its construction, and a summary of its contents. We present analyses and experimental results on intermediate states of Dolma to share what we have learned about important data curation practices. Finally, we open-source our data curation toolkit to enable reproduction of our work as well as support further research in large-scale data curation.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-02", "2024-06", "2025-05", "2025-12"], "last_mentioned": "2025-12-01T08:20:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2403.05530", "slug": "40b7c136b55fcf55", "title": "Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context", "authors": [" Gemini Team", "Petko Georgiev", "Ving Ian Lei", "Ryan Burnell", "Libin Bai", "Anmol Gulati", "Garrett Tanzer", "Damien Vincent", "Zhufeng Pan", "Shibo Wang", "Soroosh Mariooryad", "Yifan Ding", "Xinyang Geng", "Fred Alcober", "Roy Frostig", "Mark Omernick", "Lexi Walker", "Cosmin Paduraru", "Christina Sorokin", "Andrea Tacchetti", "Colin Gaffney", "Samira Daruki", "Olcan Sercinoglu", "Zach Gleicher", "Juliette Love", "Paul Voigtlaender", "Rohan Jain", "Gabriela Surita", "Kareem Mohamed", "Rory Blevins", "Junwhan Ahn", "Tao Zhu", "Kornraphop Kawintiranon", "Orhan Firat", "Yiming Gu", "Yujing Zhang", "Matthew Rahtz", "Manaal Faruqui", "Natalie Clay", "Justin Gilmer", "JD Co-Reyes", "Ivo Penchev", "Rui Zhu", "Nobuyuki Morioka", "Kevin Hui", "Krishna Haridasan", "Victor Campos", "Mahdis Mahdieh", "Mandy Guo", "Samer Hassan", "Kevin Kilgour", "Arpi Vezer", "Heng-Tze Cheng", "Raoul de Liedekerke", "Siddharth Goyal", "Paul Barham", "DJ Strouse", "Seb Noury", "Jonas Adler", "Mukund Sundararajan", "Sharad Vikram", "Dmitry Lepikhin", "Michela Paganini", "Xavier Garcia", "Fan Yang", "Dasha Valter", "Maja Trebacz", "Kiran Vodrahalli", "Chulayuth Asawaroengchai", "Roman Ring", "Norbert Kalb", "Livio Baldini Soares", "Siddhartha Brahma", "David Steiner", "Tianhe Yu", "Fabian Mentzer", "Antoine He", "Lucas Gonzalez", "Bibo Xu", "Raphael Lopez Kaufman", "Laurent El Shafey", "Junhyuk Oh", "Tom Hennigan", "George van den Driessche", "Seth Odoom", "Mario Lucic", "Becca Roelofs", "Sid Lall", "Amit Marathe", "Betty Chan", "Santiago Ontanon", "Luheng He", "Denis Teplyashin", "Jonathan Lai", "Phil Crone", "Bogdan Damoc", "Lewis Ho", "Sebastian Riedel", "Karel Lenc", "Chih-Kuan Yeh", "Aakanksha Chowdhery", "Yang Xu", "Mehran Kazemi", "Ehsan Amid", "Anastasia Petrushkina", "Kevin Swersky", "Ali Khodaei", "Gowoon Chen", "Chris Larkin", "Mario Pinto", "Geng Yan", "Adria Puigdomenech Badia", "Piyush Patil", "Steven Hansen", "Dave Orr", "Sebastien M. R. Arnold", "Jordan Grimstad", "Andrew Dai", "Sholto Douglas", "Rishika Sinha", "Vikas Yadav", "Xi Chen", "Elena Gribovskaya", "Jacob Austin", "Jeffrey Zhao", "Kaushal Patel", "Paul Komarek", "Sophia Austin", "Sebastian Borgeaud", "Linda Friso", "Abhimanyu Goyal", "Ben Caine", "Kris Cao", "Da-Woon Chung", "Matthew Lamm", "Gabe Barth-Maron", "Thais Kagohara", "Kate Olszewska", "Mia Chen", "Kaushik Shivakumar", "Rishabh Agarwal", "Harshal Godhia", "Ravi Rajwar", "Javier Snaider", "Xerxes Dotiwalla", "Yuan Liu", "Aditya Barua", "Victor Ungureanu", "Yuan Zhang", "Bat-Orgil Batsaikhan", "Mateo Wirth", "James Qin", "Ivo Danihelka", "Tulsee Doshi", "Martin Chadwick", "Jilin Chen", "Sanil Jain", "Quoc Le", "Arjun Kar", "Madhu Gurumurthy", "Cheng Li", "Ruoxin Sang", "Fangyu Liu", "Lampros Lamprou", "Rich Munoz", "Nathan Lintz", "Harsh Mehta", "Heidi Howard", "Malcolm Reynolds", "Lora Aroyo", "Quan Wang", "Lorenzo Blanco", "Albin Cassirer", "Jordan Griffith", "Dipanjan Das", "Stephan Lee", "Jakub Sygnowski", "Zach Fisher", "James Besley", "Richard Powell", "Zafarali Ahmed", "Dominik Paulus", "David Reitter", "Zalan Borsos", "Rishabh Joshi", "Aedan Pope", "Steven Hand", "Vittorio Selo", "Vihan Jain", "Nikhil Sethi", "Megha Goel", "Takaki Makino", "Rhys May", "Zhen Yang", "Johan Schalkwyk", "Christina Butterfield", "Anja Hauth", "Alex Goldin", "Will Hawkins", "Evan Senter", "Sergey Brin", "Oliver Woodman", "Marvin Ritter", "Eric Noland", "Minh Giang", "Vijay Bolina", "Lisa Lee", "Tim Blyth", "Ian Mackinnon", "Machel Reid", "Obaid Sarvana", "David Silver", "Alexander Chen", "Lily Wang", "Loren Maggiore", "Oscar Chang", "Nithya Attaluri", "Gregory Thornton", "Chung-Cheng Chiu", "Oskar Bunyan", "Nir Levine", "Timothy Chung", "Evgenii Eltyshev", "Xiance Si", "Timothy Lillicrap", "Demetra Brady", "Vaibhav Aggarwal", "Boxi Wu", "Yuanzhong Xu", "Ross McIlroy", "Kartikeya Badola", "Paramjit Sandhu", "Erica Moreira", "Wojciech Stokowiec", "Ross Hemsley", "Dong Li", "Alex Tudor", "Pranav Shyam", "Elahe Rahimtoroghi", "Salem Haykal", "Pablo Sprechmann", "Xiang Zhou", "Diana Mincu", "Yujia Li", "Ravi Addanki", "Kalpesh Krishna", "Xiao Wu", "Alexandre Frechette", "Matan Eyal", "Allan Dafoe", "Dave Lacey", "Jay Whang", "Thi Avrahami", "Ye Zhang", "Emanuel Taropa", "Hanzhao Lin", "Daniel Toyama", "Eliza Rutherford", "Motoki Sano", "HyunJeong Choe", "Alex Tomala", "Chalence Safranek-Shrader", "Nora Kassner", "Mantas Pajarskas", "Matt Harvey", "Sean Sechrist", "Meire Fortunato", "Christina Lyu", "Gamaleldin Elsayed", "Chenkai Kuang", "James Lottes", "Eric Chu", "Chao Jia", "Chih-Wei Chen", "Peter Humphreys", "Kate Baumli", "Connie Tao", "Rajkumar Samuel", "Cicero Nogueira dos Santos", "Anders Andreassen", "Nemanja Rakićević", "Dominik Grewe", "Aviral Kumar", "Stephanie Winkler", "Jonathan Caton", "Andrew Brock", "Sid Dalmia", "Hannah Sheahan", "Iain Barr", "Yingjie Miao", "Paul Natsev", "Jacob Devlin", "Feryal Behbahani", "Flavien Prost", "Yanhua Sun", "Artiom Myaskovsky", "Thanumalayan Sankaranarayana Pillai", "Dan Hurt", "Angeliki Lazaridou", "Xi Xiong", "Ce Zheng", "Fabio Pardo", "Xiaowei Li", "Dan Horgan", "Joe Stanton", "Moran Ambar", "Fei Xia", "Alejandro Lince", "Mingqiu Wang", "Basil Mustafa", "Albert Webson", "Hyo Lee", "Rohan Anil", "Martin Wicke", "Timothy Dozat", "Abhishek Sinha", "Enrique Piqueras", "Elahe Dabir", "Shyam Upadhyay", "Anudhyan Boral", "Lisa Anne Hendricks", "Corey Fry", "Josip Djolonga", "Yi Su", "Jake Walker", "Jane Labanowski", "Ronny Huang", "Vedant Misra", "Jeremy Chen", "RJ Skerry-Ryan", "Avi Singh", "Shruti Rijhwani", "Dian Yu", "Alex Castro-Ros", "Beer Changpinyo", "Romina Datta", "Sumit Bagri", "Arnar Mar Hrafnkelsson", "Marcello Maggioni", "Daniel Zheng", "Yury Sulsky", "Shaobo Hou", "Tom Le Paine", "Antoine Yang", "Jason Riesa", "Dominika Rogozinska", "Dror Marcus", "Dalia El Badawy", "Qiao Zhang", "Luyu Wang", "Helen Miller", "Jeremy Greer", "Lars Lowe Sjos", "Azade Nova", "Heiga Zen", "Rahma Chaabouni", "Mihaela Rosca", "Jiepu Jiang", "Charlie Chen", "Ruibo Liu", "Tara Sainath", "Maxim Krikun", "Alex Polozov", "Jean-Baptiste Lespiau", "Josh Newlan", "Zeyncep Cankara", "Soo Kwak", "Yunhan Xu", "Phil Chen", "Andy Coenen", "Clemens Meyer", "Katerina Tsihlas", "Ada Ma", "Juraj Gottweis", "Jinwei Xing", "Chenjie Gu", "Jin Miao", "Christian Frank", "Zeynep Cankara", "Sanjay Ganapathy", "Ishita Dasgupta", "Steph Hughes-Fitt", "Heng Chen", "David Reid", "Keran Rong", "Hongmin Fan", "Joost van Amersfoort", "Vincent Zhuang", "Aaron Cohen", "Shixiang Shane Gu", "Anhad Mohananey", "Anastasija Ilic", "Taylor Tobin", "John Wieting", "Anna Bortsova", "Phoebe Thacker", "Emma Wang", "Emily Caveness", "Justin Chiu", "Eren Sezener", "Alex Kaskasoli", "Steven Baker", "Katie Millican", "Mohamed Elhawaty", "Kostas Aisopos", "Carl Lebsack", "Nathan Byrd", "Hanjun Dai", "Wenhao Jia", "Matthew Wiethoff", "Elnaz Davoodi", "Albert Weston", "Lakshman Yagati", "Arun Ahuja", "Isabel Gao", "Golan Pundak", "Susan Zhang", "Michael Azzam", "Khe Chai Sim", "Sergi Caelles", "James Keeling", "Abhanshu Sharma", "Andy Swing", "YaGuang Li", "Chenxi Liu", "Carrie Grimes Bostock", "Yamini Bansal", "Zachary Nado", "Ankesh Anand", "Josh Lipschultz", "Abhijit Karmarkar", "Lev Proleev", "Abe Ittycheriah", "Soheil Hassas Yeganeh", "George Polovets", "Aleksandra Faust", "Jiao Sun", "Alban Rrustemi", "Pen Li", "Rakesh Shivanna", "Jeremiah Liu", "Chris Welty", "Federico Lebron", "Anirudh Baddepudi", "Sebastian Krause", "Emilio Parisotto", "Radu Soricut", "Zheng Xu", "Dawn Bloxwich", "Melvin Johnson", "Behnam Neyshabur", "Justin Mao-Jones", "Renshen Wang", "Vinay Ramasesh", "Zaheer Abbas", "Arthur Guez", "Constant Segal", "Duc Dung Nguyen", "James Svensson", "Le Hou", "Sarah York", "Kieran Milan", "Sophie Bridgers", "Wiktor Gworek", "Marco Tagliasacchi", "James Lee-Thorp", "Michael Chang", "Alexey Guseynov", "Ale Jakse Hartman", "Michael Kwong", "Ruizhe Zhao", "Sheleem Kashem", "Elizabeth Cole", "Antoine Miech", "Richard Tanburn", "Mary Phuong", "Filip Pavetic", "Sebastien Cevey", "Ramona Comanescu", "Richard Ives", "Sherry Yang", "Cosmo Du", "Bo Li", "Zizhao Zhang", "Mariko Iinuma", "Clara Huiyi Hu", "Aurko Roy", "Shaan Bijwadia", "Zhenkai Zhu", "Danilo Martins", "Rachel Saputro", "Anita Gergely", "Steven Zheng", "Dawei Jia", "Ioannis Antonoglou", "Adam Sadovsky", "Shane Gu", "Yingying Bi", "Alek Andreev", "Sina Samangooei", "Mina Khan", "Tomas Kocisky", "Angelos Filos", "Chintu Kumar", "Colton Bishop", "Adams Yu", "Sarah Hodkinson", "Sid Mittal", "Premal Shah", "Alexandre Moufarek", "Yong Cheng", "Adam Bloniarz", "Jaehoon Lee", "Pedram Pejman", "Paul Michel", "Stephen Spencer", "Vladimir Feinberg", "Xuehan Xiong", "Nikolay Savinov", "Charlotte Smith", "Siamak Shakeri", "Dustin Tran", "Mary Chesus", "Bernd Bohnet", "George Tucker", "Tamara von Glehn", "Carrie Muir", "Yiran Mao", "Hideto Kazawa", "Ambrose Slone", "Kedar Soparkar", "Disha Shrivastava", "James Cobon-Kerr", "Michael Sharman", "Jay Pavagadhi", "Carlos Araya", "Karolis Misiunas", "Nimesh Ghelani", "Michael Laskin", "David Barker", "Qiujia Li", "Anton Briukhov", "Neil Houlsby", "Mia Glaese", "Balaji Lakshminarayanan", "Nathan Schucher", "Yunhao Tang", "Eli Collins", "Hyeontaek Lim", "Fangxiaoyu Feng", "Adria Recasens", "Guangda Lai", "Alberto Magni", "Nicola De Cao", "Aditya Siddhant", "Zoe Ashwood", "Jordi Orbay", "Mostafa Dehghani", "Jenny Brennan", "Yifan He", "Kelvin Xu", "Yang Gao", "Carl Saroufim", "James Molloy", "Xinyi Wu", "Seb Arnold", "Solomon Chang", "Julian Schrittwieser", "Elena Buchatskaya", "Soroush Radpour", "Martin Polacek", "Skye Giordano", "Ankur Bapna", "Simon Tokumine", "Vincent Hellendoorn", "Thibault Sottiaux", "Sarah Cogan", "Aliaksei Severyn", "Mohammad Saleh", "Shantanu Thakoor", "Laurent Shefey", "Siyuan Qiao", "Meenu Gaba", "Shuo-yiin Chang", "Craig Swanson", "Biao Zhang", "Benjamin Lee", "Paul Kishan Rubenstein", "Gan Song", "Tom Kwiatkowski", "Anna Koop", "Ajay Kannan", "David Kao", "Parker Schuh", "Axel Stjerngren", "Golnaz Ghiasi", "Gena Gibson", "Luke Vilnis", "Ye Yuan", "Felipe Tiengo Ferreira", "Aishwarya Kamath", "Ted Klimenko", "Ken Franko", "Kefan Xiao", "Indro Bhattacharya", "Miteyan Patel", "Rui Wang", "Alex Morris", "Robin Strudel", "Vivek Sharma", "Peter Choy", "Sayed Hadi Hashemi", "Jessica Landon", "Mara Finkelstein", "Priya Jhakra", "Justin Frye", "Megan Barnes", "Matthew Mauger", "Dennis Daun", "Khuslen Baatarsukh", "Matthew Tung", "Wael Farhan", "Henryk Michalewski", "Fabio Viola", "Felix de Chaumont Quitry", "Charline Le Lan", "Tom Hudson", "Qingze Wang", "Felix Fischer", "Ivy Zheng", "Elspeth White", "Anca Dragan", "Jean-baptiste Alayrac", "Eric Ni", "Alexander Pritzel", "Adam Iwanicki", "Michael Isard", "Anna Bulanova", "Lukas Zilka", "Ethan Dyer", "Devendra Sachan", "Srivatsan Srinivasan", "Hannah Muckenhirn", "Honglong Cai", "Amol Mandhane", "Mukarram Tariq", "Jack W. Rae", "Gary Wang", "Kareem Ayoub", "Nicholas FitzGerald", "Yao Zhao", "Woohyun Han", "Chris Alberti", "Dan Garrette", "Kashyap Krishnakumar", "Mai Gimenez", "Anselm Levskaya", "Daniel Sohn", "Josip Matak", "Inaki Iturrate", "Michael B. Chang", "Jackie Xiang", "Yuan Cao", "Nishant Ranka", "Geoff Brown", "Adrian Hutter", "Vahab Mirrokni", "Nanxin Chen", "Kaisheng Yao", "Zoltan Egyed", "Francois Galilee", "Tyler Liechty", "Praveen Kallakuri", "Evan Palmer", "Sanjay Ghemawat", "Jasmine Liu", "David Tao", "Chloe Thornton", "Tim Green", "Mimi Jasarevic", "Sharon Lin", "Victor Cotruta", "Yi-Xuan Tan", "Noah Fiedel", "Hongkun Yu", "Ed Chi", "Alexander Neitz", "Jens Heitkaemper", "Anu Sinha", "Denny Zhou", "Yi Sun", "Charbel Kaed", "Brice Hulse", "Swaroop Mishra", "Maria Georgaki", "Sneha Kudugunta", "Clement Farabet", "Izhak Shafran", "Daniel Vlasic", "Anton Tsitsulin", "Rajagopal Ananthanarayanan", "Alen Carin", "Guolong Su", "Pei Sun", "Shashank V", "Gabriel Carvajal", "Josef Broder", "Iulia Comsa", "Alena Repina", "William Wong", "Warren Weilun Chen", "Peter Hawkins", "Egor Filonov", "Lucia Loher", "Christoph Hirnschall", "Weiyi Wang", "Jingchen Ye", "Andrea Burns", "Hardie Cate", "Diana Gage Wright", "Federico Piccinini", "Lei Zhang", "Chu-Cheng Lin", "Ionel Gog", "Yana Kulizhskaya", "Ashwin Sreevatsa", "Shuang Song", "Luis C. Cobo", "Anand Iyer", "Chetan Tekur", "Guillermo Garrido", "Zhuyun Xiao", "Rupert Kemp", "Huaixiu Steven Zheng", "Hui Li", "Ananth Agarwal", "Christel Ngani", "Kati Goshvadi", "Rebeca Santamaria-Fernandez", "Wojciech Fica", "Xinyun Chen", "Chris Gorgolewski", "Sean Sun", "Roopal Garg", "Xinyu Ye", "S. M. Ali Eslami", "Nan Hua", "Jon Simon", "Pratik Joshi", "Yelin Kim", "Ian Tenney", "Sahitya Potluri", "Lam Nguyen Thiet", "Quan Yuan", "Florian Luisier", "Alexandra Chronopoulou", "Salvatore Scellato", "Praveen Srinivasan", "Minmin Chen", "Vinod Koverkathu", "Valentin Dalibard", "Yaming Xu", "Brennan Saeta", "Keith Anderson", "Thibault Sellam", "Nick Fernando", "Fantine Huot", "Junehyuk Jung", "Mani Varadarajan", "Michael Quinn", "Amit Raul", "Maigo Le", "Ruslan Habalov", "Jon Clark", "Komal Jalan", "Kalesha Bullard", "Achintya Singhal", "Thang Luong", "Boyu Wang", "Sujeevan Rajayogam", "Julian Eisenschlos", "Johnson Jia", "Daniel Finchelstein", "Alex Yakubovich", "Daniel Balle", "Michael Fink", "Sameer Agarwal", "Jing Li", "Dj Dvijotham", "Shalini Pal", "Kai Kang", "Jaclyn Konzelmann", "Jennifer Beattie", "Olivier Dousse", "Diane Wu", "Remi Crocker", "Chen Elkind", "Siddhartha Reddy Jonnalagadda", "Jong Lee", "Dan Holtmann-Rice", "Krystal Kallarackal", "Rosanne Liu", "Denis Vnukov", "Neera Vats", "Luca Invernizzi", "Mohsen Jafari", "Huanjie Zhou", "Lilly Taylor", "Jennifer Prendki", "Marcus Wu", "Tom Eccles", "Tianqi Liu", "Kavya Kopparapu", "Francoise Beaufays", "Christof Angermueller", "Andreea Marzoca", "Shourya Sarcar", "Hilal Dib", "Jeff Stanway", "Frank Perbet", "Nejc Trdin", "Rachel Sterneck", "Andrey Khorlin", "Dinghua Li", "Xihui Wu", "Sonam Goenka", "David Madras", "Sasha Goldshtein", "Willi Gierke", "Tong Zhou", "Yaxin Liu", "Yannie Liang", "Anais White", "Yunjie Li", "Shreya Singh", "Sanaz Bahargam", "Mark Epstein", "Sujoy Basu", "Li Lao", "Adnan Ozturel", "Carl Crous", "Alex Zhai", "Han Lu", "Zora Tung", "Neeraj Gaur", "Alanna Walton", "Lucas Dixon", "Ming Zhang", "Amir Globerson", "Grant Uy", "Andrew Bolt", "Olivia Wiles", "Milad Nasr", "Ilia Shumailov", "Marco Selvi", "Francesco Piccinno", "Ricardo Aguilar", "Sara McCarthy", "Misha Khalman", "Mrinal Shukla", "Vlado Galic", "John Carpenter", "Kevin Villela", "Haibin Zhang", "Harry Richardson", "James Martens", "Matko Bosnjak", "Shreyas Rammohan Belle", "Jeff Seibert", "Mahmoud Alnahlawi", "Brian McWilliams", "Sankalp Singh", "Annie Louis", "Wen Ding", "Dan Popovici", "Lenin Simicich", "Laura Knight", "Pulkit Mehta", "Nishesh Gupta", "Chongyang Shi", "Saaber Fatehi", "Jovana Mitrovic", "Alex Grills", "Joseph Pagadora", "Tsendsuren Munkhdalai", "Dessie Petrova", "Danielle Eisenbud", "Zhishuai Zhang", "Damion Yates", "Bhavishya Mittal", "Nilesh Tripuraneni", "Yannis Assael", "Thomas Brovelli", "Prateek Jain", "Mihajlo Velimirovic", "Canfer Akbulut", "Jiaqi Mu", "Wolfgang Macherey", "Ravin Kumar", "Jun Xu", "Haroon Qureshi", "Gheorghe Comanici", "Jeremy Wiesner", "Zhitao Gong", "Anton Ruddock", "Matthias Bauer", "Nick Felt", "Anirudh GP", "Anurag Arnab", "Dustin Zelle", "Jonas Rothfuss", "Bill Rosgen", "Ashish Shenoy", "Bryan Seybold", "Xinjian Li", "Jayaram Mudigonda", "Goker Erdogan", "Jiawei Xia", "Jiri Simsa", "Andrea Michi", "Yi Yao", "Christopher Yew", "Steven Kan", "Isaac Caswell", "Carey Radebaugh", "Andre Elisseeff", "Pedro Valenzuela", "Kay McKinney", "Kim Paterson", "Albert Cui", "Eri Latorre-Chimoto", "Solomon Kim", "William Zeng", "Ken Durden", "Priya Ponnapalli", "Tiberiu Sosea", "Christopher A. Choquette-Choo", "James Manyika", "Brona Robenek", "Harsha Vashisht", "Sebastien Pereira", "Hoi Lam", "Marko Velic", "Denese Owusu-Afriyie", "Katherine Lee", "Tolga Bolukbasi", "Alicia Parrish", "Shawn Lu", "Jane Park", "Balaji Venkatraman", "Alice Talbert", "Lambert Rosique", "Yuchung Cheng", "Andrei Sozanschi", "Adam Paszke", "Praveen Kumar", "Jessica Austin", "Lu Li", "Khalid Salama", "Bartek Perz", "Wooyeol Kim", "Nandita Dukkipati", "Anthony Baryshnikov", "Christos Kaplanis", "XiangHai Sheng", "Yuri Chervonyi", "Caglar Unlu", "Diego de Las Casas", "Harry Askham", "Kathryn Tunyasuvunakool", "Felix Gimeno", "Siim Poder", "Chester Kwak", "Matt Miecnikowski", "Vahab Mirrokni", "Alek Dimitriev", "Aaron Parisi", "Dangyi Liu", "Tomy Tsai", "Toby Shevlane", "Christina Kouridi", "Drew Garmon", "Adrian Goedeckemeyer", "Adam R. Brown", "Anitha Vijayakumar", "Ali Elqursh", "Sadegh Jazayeri", "Jin Huang", "Sara Mc Carthy", "Jay Hoover", "Lucy Kim", "Sandeep Kumar", "Wei Chen", "Courtney Biles", "Garrett Bingham", "Evan Rosen", "Lisa Wang", "Qijun Tan", "David Engel", "Francesco Pongetti", "Dario de Cesare", "Dongseong Hwang", "Lily Yu", "Jennifer Pullman", "Srini Narayanan", "Kyle Levin", "Siddharth Gopal", "Megan Li", "Asaf Aharoni", "Trieu Trinh", "Jessica Lo", "Norman Casagrande", "Roopali Vij", "Loic Matthey", "Bramandia Ramadhana", "Austin Matthews", "CJ Carey", "Matthew Johnson", "Kremena Goranova", "Rohin Shah", "Shereen Ashraf", "Kingshuk Dasgupta", "Rasmus Larsen", "Yicheng Wang", "Manish Reddy Vuyyuru", "Chong Jiang", "Joana Ijazi", "Kazuki Osawa", "Celine Smith", "Ramya Sree Boppana", "Taylan Bilal", "Yuma Koizumi", "Ying Xu", "Yasemin Altun", "Nir Shabat", "Ben Bariach", "Alex Korchemniy", "Kiam Choo", "Olaf Ronneberger", "Chimezie Iwuanyanwu", "Shubin Zhao", "David Soergel", "Cho-Jui Hsieh", "Irene Cai", "Shariq Iqbal", "Martin Sundermeyer", "Zhe Chen", "Elie Bursztein", "Chaitanya Malaviya", "Fadi Biadsy", "Prakash Shroff", "Inderjit Dhillon", "Tejasi Latkar", "Chris Dyer", "Hannah Forbes", "Massimo Nicosia", "Vitaly Nikolaev", "Somer Greene", "Marin Georgiev", "Pidong Wang", "Nina Martin", "Hanie Sedghi", "John Zhang", "Praseem Banzal", "Doug Fritz", "Vikram Rao", "Xuezhi Wang", "Jiageng Zhang", "Viorica Patraucean", "Dayou Du", "Igor Mordatch", "Ivan Jurin", "Lewis Liu", "Ayush Dubey", "Abhi Mohan", "Janek Nowakowski", "Vlad-Doru Ion", "Nan Wei", "Reiko Tojo", "Maria Abi Raad", "Drew A. Hudson", "Vaishakh Keshava", "Shubham Agrawal", "Kevin Ramirez", "Zhichun Wu", "Hoang Nguyen", "Ji Liu", "Madhavi Sewak", "Bryce Petrini", "DongHyun Choi", "Ivan Philips", "Ziyue Wang", "Ioana Bica", "Ankush Garg", "Jarek Wilkiewicz", "Priyanka Agrawal", "Xiaowei Li", "Danhao Guo", "Emily Xue", "Naseer Shaik", "Andrew Leach", "Sadh MNM Khan", "Julia Wiesinger", "Sammy Jerome", "Abhishek Chakladar", "Alek Wenjiao Wang", "Tina Ornduff", "Folake Abu", "Alireza Ghaffarkhah", "Marcus Wainwright", "Mario Cortes", "Frederick Liu", "Joshua Maynez", "Andreas Terzis", "Pouya Samangouei", "Riham Mansour", "Tomasz Kępa", "François-Xavier Aubet", "Anton Algymr", "Dan Banica", "Agoston Weisz", "Andras Orban", "Alexandre Senges", "Ewa Andrejczuk", "Mark Geller", "Niccolo Dal Santo", "Valentin Anklin", "Majd Al Merey", "Martin Baeuml", "Trevor Strohman", "Junwen Bai", "Slav Petrov", "Yonghui Wu", "Demis Hassabis", "Koray Kavukcuoglu", "Jeff Dean", "Oriol Vinyals"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2403.05530", "doi": null, "topics": ["Language models", "Statistics & probability", "Economics"], "description": "In this report, we introduce the Gemini 1.5 family of models, representing the next generation of highly compute-efficient multimodal models capable of recalling and reasoning over fine-grained information from millions of tokens of context, including multiple long documents and hours of video and audio.", "description_method": "abstract-excerpt", "abstract": "In this report, we introduce the Gemini 1.5 family of models, representing the next generation of highly compute-efficient multimodal models capable of recalling and reasoning over fine-grained information from millions of tokens of context, including multiple long documents and hours of video and audio. The family includes two new models: (1) an updated Gemini 1.5 Pro, which exceeds the February version on the great majority of capabilities and benchmarks; (2) Gemini 1.5 Flash, a more lightweight variant designed for efficiency with minimal regression in quality. Gemini 1.5 models achieve near-perfect recall on long-context retrieval tasks across modalities, improve the state-of-the-art in long-document QA, long-video QA and long-context ASR, and match or surpass Gemini 1.0 Ultra's state-of-the-art performance across a broad set of benchmarks. Studying the limits of Gemini 1.5's long-context ability, we find continued improvement in next-token prediction and near-perfect retrieval (>99%) up to at least 10M tokens, a generational leap over existing models such as Claude 3.0 (200k) and GPT-4 Turbo (128k). Finally, we highlight real-world use cases, such as Gemini 1.5 collaborating with professionals on completing their tasks achieving 26 to 75% time savings across 10 different job categories, as well as surprising new capabilities of large language models at the frontier; when given a grammar manual for Kalamang, a language with fewer than 200 speakers worldwide, the model learns to translate English to Kalamang at a similar level to a person who learned from the same content.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-04", "2025-01", "2026-03"], "last_mentioned": "2026-03-11T22:18:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2403.07183", "slug": "f2f7eea8d36a2dd1", "title": "Monitoring AI-Modified Content at Scale: A Case Study on the Impact of ChatGPT on AI Conference Peer Reviews", "authors": ["Weixin Liang", "Zachary Izzo", "Yaohui Zhang", "Haley Lepp", "Hancheng Cao", "Xuandong Zhao", "Lingjiao Chen", "Haotian Ye", "Sheng Liu", "Zhi Huang", "Daniel A. McFarland", "James Y. Zou"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2403.07183", "doi": null, "topics": ["Science & research", "Language models", "Psychology"], "description": "We present an approach for estimating the fraction of text in a large corpus which is likely to be substantially modified or produced by a large language model (LLM).", "description_method": "abstract-excerpt", "abstract": "We present an approach for estimating the fraction of text in a large corpus which is likely to be substantially modified or produced by a large language model (LLM). Our maximum likelihood model leverages expert-written and AI-generated reference texts to accurately and efficiently examine real-world LLM-use at the corpus level. We apply this approach to a case study of scientific peer review in AI conferences that took place after the release of ChatGPT: ICLR 2024, NeurIPS 2023, CoRL 2023 and EMNLP 2023. Our results suggest that between 6.5% and 16.9% of text submitted as peer reviews to these conferences could have been substantially modified by LLMs, i.e. beyond spell-checking or minor writing updates. The circumstances in which generated text occurs offer insight into user behavior: the estimated fraction of LLM-generated text is higher in reviews which report lower confidence, were submitted close to the deadline, and from reviewers who are less likely to respond to author rebuttals. We also observe corpus-level trends in generated text which may be too subtle to detect at the individual level, and discuss the implications of such trends on peer review. We call for future interdisciplinary work to examine how LLM use is changing our information and knowledge practices.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-03", "2024-04", "2024-06"], "last_mentioned": "2024-06-23T00:25:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2404.05405", "slug": "541683d3678177bb", "title": "Physics of Language Models: Part 3.3, Knowledge Capacity Scaling Laws", "authors": ["Zeyuan Allen-Zhu", "Yuanzhi Li"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2404.05405", "doi": null, "topics": ["Language models", "Physics", "Robotics"], "description": "Scaling laws describe the relationship between the size of language models and their capabilities.", "description_method": "abstract-excerpt", "abstract": "Scaling laws describe the relationship between the size of language models and their capabilities. Unlike prior studies that evaluate a model's capability via loss or benchmarks, we estimate the number of knowledge bits a model stores. We focus on factual knowledge represented as tuples, such as (USA, capital, Washington D.C.) from a Wikipedia page. Through multiple controlled datasets, we establish that language models can and only can store 2 bits of knowledge per parameter, even when quantized to int8, and such knowledge can be flexibly extracted for downstream applications. Consequently, a 7B model can store 14B bits of knowledge, surpassing the English Wikipedia and textbooks combined based on our estimation. More broadly, we present 12 results on how (1) training duration, (2) model architecture, (3) quantization, (4) sparsity constraints such as MoE, and (5) data signal-to-noise ratio affect a model's knowledge storage capacity. Notable insights include: * The GPT-2 architecture, with rotary embedding, matches or even surpasses LLaMA/Mistral architectures in knowledge storage, particularly over shorter training durations. This arises because LLaMA/Mistral uses GatedMLP, which is less stable and harder to train. * Prepending training data with domain names (e.g., wikipedia.org) significantly increases a model's knowledge capacity. Language models can autonomously identify and prioritize domains rich in knowledge, optimizing their storage capacity.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-06", "2025-01", "2025-06", "2025-10"], "last_mentioned": "2025-10-04T20:46:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2404.08819", "slug": "16b9b4bb66853842", "title": "The Illusion of State in State-Space Models", "authors": ["William Merrill", "Jackson Petty", "Ashish Sabharwal"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2404.08819", "doi": null, "topics": ["Language models", "Machine learning", "Algorithms"], "description": "State-space models (SSMs) have emerged as a potential alternative architecture for building large language models (LLMs) compared to the previously ubiquitous transformer architecture.", "description_method": "abstract-excerpt", "abstract": "State-space models (SSMs) have emerged as a potential alternative architecture for building large language models (LLMs) compared to the previously ubiquitous transformer architecture. One theoretical weakness of transformers is that they cannot express certain kinds of sequential computation and state tracking (Merrill & Sabharwal, 2023), which SSMs are explicitly designed to address via their close architectural similarity to recurrent neural networks (RNNs). But do SSMs truly have an advantage (over transformers) in expressive power for state tracking? Surprisingly, the answer is no. Our analysis reveals that the expressive power of SSMs is limited very similarly to transformers: SSMs cannot express computation outside the complexity class $\\mathsf{TC}^0$. In particular, this means they cannot solve simple state-tracking problems like permutation composition. It follows that SSMs are provably unable to accurately track chess moves with certain notation, evaluate code, or track entities in a long narrative. To supplement our formal analysis, we report experiments showing that Mamba-style SSMs indeed struggle with state tracking. Thus, despite its recurrent formulation, the \"state\" in an SSM is an illusion: SSMs have similar expressiveness limitations to non-recurrent models like transformers, which may fundamentally limit their ability to solve real-world state-tracking problems.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2024-05", "2024-06", "2024-10", "2025-01"], "last_mentioned": "2025-01-02T19:04:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2405.02709", "slug": "3275c97161cb244b", "title": "Constant Velocity Physical Warp Drive Solution", "authors": ["Jared Fuchs", "Christopher Helmerich", "Alexey Bobrick", "Luke Sellers", "Brandon Melcher", "Gianni Martire"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2405.02709", "doi": "10.1088/1361-6382/ad26aa", "topics": ["Physics"], "description": "Warp drives are exotic solutions of general relativity that offer novel means of transportation.", "description_method": "abstract-excerpt", "abstract": "Warp drives are exotic solutions of general relativity that offer novel means of transportation. In this study, we present a solution for a constant-velocity subluminal warp drive that satisfies all of the energy conditions. The solution involves combining a stable matter shell with a shift vector distribution that closely matches well-known warp drive solutions such as the Alcubierre metric. We generate the spacetime metric numerically, evaluate the energy conditions, and confirm that the shift vector distribution cannot be reduced to a coordinate transformation. This study demonstrates that classic warp drive spacetimes can be made to satisfy the energy conditions by adding a regular matter shell with a positive ADM mass.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-05", "2024-07", "2024-11"], "last_mentioned": "2024-11-15T13:03:31Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2405.04517", "slug": "40fd3e03894caeb9", "title": "xLSTM: Extended Long Short-Term Memory", "authors": ["Maximilian Beck", "Korbinian Pöppel", "Markus Spanring", "Andreas Auer", "Oleksandra Prudnikova", "Michael Kopp", "Günter Klambauer", "Johannes Brandstetter", "Sepp Hochreiter"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2405.04517", "doi": null, "topics": ["Language models", "Machine learning"], "description": "In the 1990s, the constant error carousel and gating were introduced as the central ideas of the Long Short-Term Memory (LSTM).", "description_method": "abstract-excerpt", "abstract": "In the 1990s, the constant error carousel and gating were introduced as the central ideas of the Long Short-Term Memory (LSTM). Since then, LSTMs have stood the test of time and contributed to numerous deep learning success stories, in particular they constituted the first Large Language Models (LLMs). However, the advent of the Transformer technology with parallelizable self-attention at its core marked the dawn of a new era, outpacing LSTMs at scale. We now raise a simple question: How far do we get in language modeling when scaling LSTMs to billions of parameters, leveraging the latest techniques from modern LLMs, but mitigating known limitations of LSTMs? Firstly, we introduce exponential gating with appropriate normalization and stabilization techniques. Secondly, we modify the LSTM memory structure, obtaining: (i) sLSTM with a scalar memory, a scalar update, and new memory mixing, (ii) mLSTM that is fully parallelizable with a matrix memory and a covariance update rule. Integrating these LSTM extensions into residual block backbones yields xLSTM blocks that are then residually stacked into xLSTM architectures. Exponential gating and modified memory structures boost xLSTM capabilities to perform favorably when compared to state-of-the-art Transformers and State Space Models, both in performance and scaling.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-06", "2024-10", "2025-02"], "last_mentioned": "2025-02-15T03:04:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2405.10387", "slug": "1f04be0977585ba5", "title": "Grothendieck's use of equality", "authors": ["Kevin Buzzard"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2405.10387", "doi": null, "topics": ["Mathematics"], "description": "We discuss how the concept of equality is used by mathematicians (including Grothendieck), and what effect this has when trying to formalise mathematics.", "description_method": "abstract-excerpt", "abstract": "We discuss how the concept of equality is used by mathematicians (including Grothendieck), and what effect this has when trying to formalise mathematics. We challenge various reasonable-sounding slogans about equality.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-07", "2024-08", "2025-06", "2026-05"], "last_mentioned": "2026-05-24T22:02:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2408.02442", "slug": "0489f598dc51a88a", "title": "Let Me Speak Freely? A Study on the Impact of Format Restrictions on Performance of Large Language Models", "authors": ["Zhi Rui Tam", "Cheng-Kuang Wu", "Yi-Lin Tsai", "Chieh-Yen Lin", "Hung-yi Lee", "Yun-Nung Chen"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2408.02442", "doi": null, "topics": ["Language models"], "description": "Structured generation, the process of producing content in standardized formats like JSON and XML, is widely utilized in real-world applications to extract key output information from large language models (LLMs).", "description_method": "abstract-excerpt", "abstract": "Structured generation, the process of producing content in standardized formats like JSON and XML, is widely utilized in real-world applications to extract key output information from large language models (LLMs). This study investigates whether such constraints on generation space impact LLMs abilities, including reasoning and domain knowledge comprehension. Specifically, we evaluate LLMs performance when restricted to adhere to structured formats versus generating free-form responses across various common tasks. Surprisingly, we observe a significant decline in LLMs reasoning abilities under format restrictions. Furthermore, we find that stricter format constraints generally lead to greater performance degradation in reasoning tasks.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-08", "2024-09", "2025-09", "2026-08"], "last_mentioned": "2026-08-14T14:48:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2408.03314", "slug": "0f80938d9800c718", "title": "Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters", "authors": ["Charlie Snell", "Jaehoon Lee", "Kelvin Xu", "Aviral Kumar"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2408.03314", "doi": null, "topics": ["Language models", "Psychology"], "description": "Enabling LLMs to improve their outputs by using more test-time computation is a critical step towards building generally self-improving agents that can operate on open-ended natural language.", "description_method": "abstract-excerpt", "abstract": "Enabling LLMs to improve their outputs by using more test-time computation is a critical step towards building generally self-improving agents that can operate on open-ended natural language. In this paper, we study the scaling of inference-time computation in LLMs, with a focus on answering the question: if an LLM is allowed to use a fixed but non-trivial amount of inference-time compute, how much can it improve its performance on a challenging prompt? Answering this question has implications not only on the achievable performance of LLMs, but also on the future of LLM pretraining and how one should tradeoff inference-time and pre-training compute. Despite its importance, little research attempted to understand the scaling behaviors of various test-time inference methods. Moreover, current work largely provides negative results for a number of these strategies. In this work, we analyze two primary mechanisms to scale test-time computation: (1) searching against dense, process-based verifier reward models; and (2) updating the model's distribution over a response adaptively, given the prompt at test time. We find that in both cases, the effectiveness of different approaches to scaling test-time compute critically varies depending on the difficulty of the prompt. This observation motivates applying a \"compute-optimal\" scaling strategy, which acts to most effectively allocate test-time compute adaptively per prompt. Using this compute-optimal strategy, we can improve the efficiency of test-time compute scaling by more than 4x compared to a best-of-N baseline. Additionally, in a FLOPs-matched evaluation, we find that on problems where a smaller base model attains somewhat non-trivial success rates, test-time compute can be used to outperform a 14x larger model.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-08", "2024-09", "2024-12"], "last_mentioned": "2024-12-20T21:32:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2409.06203", "slug": "5938659bbc0dad9f", "title": "Watching TV with the Second-Party: A First Look at Automatic Content Recognition Tracking in Smart TVs", "authors": ["Gianluca Anselmi", "Yash Vekaria", "Alexander D'Souza", "Patricia Callejo", "Anna Maria Mandalari", "Zubair Shafiq"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2409.06203", "doi": null, "topics": ["Psychology", "Security & cryptography"], "description": "Smart TVs implement a unique tracking approach called Automatic Content Recognition (ACR) to profile viewing activity of their users.", "description_method": "abstract-excerpt", "abstract": "Smart TVs implement a unique tracking approach called Automatic Content Recognition (ACR) to profile viewing activity of their users. ACR is a Shazam-like technology that works by periodically capturing the content displayed on a TV's screen and matching it against a content library to detect what content is being displayed at any given point in time. While prior research has investigated third-party tracking in the smart TV ecosystem, it has not looked into second-party ACR tracking that is directly conducted by the smart TV platform. In this work, we conduct a black-box audit of ACR network traffic between ACR clients on the smart TV and ACR servers. We use our auditing approach to systematically investigate whether (1) ACR tracking is agnostic to how a user watches TV (e.g., linear vs. streaming vs. HDMI), (2) privacy controls offered by smart TVs have an impact on ACR tracking, and (3) there are any differences in ACR tracking between the UK and the US. We perform a series of experiments on two major smart TV platforms: Samsung and LG. Our results show that ACR works even when the smart TV is used as a \"dumb\" external display, opting-out stops network traffic to ACR servers, and there are differences in how ACR works across the UK and the US.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-12", "2025-04", "2025-12", "2026-04"], "last_mentioned": "2026-04-21T21:31:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2501.17186", "slug": "8e4423340a53a11f", "title": "Complete Chess Games Enable LLM Become A Chess Master", "authors": ["Yinqi Zhang", "Xintian Han", "Haolong Li", "Kedi Chen", "Shaohui Lin"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2501.17186", "doi": null, "topics": ["Language models"], "description": "Large language models (LLM) have shown remarkable abilities in text generation, question answering, language translation, reasoning and many other tasks.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLM) have shown remarkable abilities in text generation, question answering, language translation, reasoning and many other tasks. It continues to advance rapidly and is becoming increasingly influential in various fields, from technology and business to education and entertainment. Despite LLM's success in multiple areas, its ability to play abstract games, such as chess, is underexplored. Chess-playing requires the language models to output legal and reasonable moves from textual inputs. Here, we propose the Large language model ChessLLM to play full chess games. We transform the game into a textual format with the best move represented in the Forsyth-Edwards Notation. We show that by simply supervised fine-tuning, our model has achieved a professional-level Elo rating of 1788 in matches against the standard Elo-rated Stockfish when permitted to sample 10 times. We further show that data quality is important. Long-round data supervision enjoys a 350 Elo rating improvement over short-round data.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2025-08", "2026-02", "2026-06"], "last_mentioned": "2026-06-29T06:54:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2502.12115", "slug": "c8d821e882481297", "title": "SWE-Lancer: Can Frontier LLMs Earn $1 Million from Real-World Freelance Software Engineering?", "authors": ["Samuel Miserendino", "Michele Wang", "Tejal Patwardhan", "Johannes Heidecke"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2502.12115", "doi": null, "topics": ["Programming languages", "Economics"], "description": "We introduce SWE-Lancer, a benchmark of over 1,400 freelance software engineering tasks from Upwork, valued at \\$1 million USD total in real-world payouts.", "description_method": "abstract-excerpt", "abstract": "We introduce SWE-Lancer, a benchmark of over 1,400 freelance software engineering tasks from Upwork, valued at \\$1 million USD total in real-world payouts. SWE-Lancer encompasses both independent engineering tasks--ranging from \\$50 bug fixes to \\$32,000 feature implementations--and managerial tasks, where models choose between technical implementation proposals. Independent tasks are graded with end-to-end tests triple-verified by experienced software engineers, while managerial decisions are assessed against the choices of the original hired engineering managers. We evaluate model performance and find that frontier models are still unable to solve the majority of tasks. To facilitate future research, we open-source a unified Docker image and a public evaluation split, SWE-Lancer Diamond (https://github.com/openai/SWELancer-Benchmark). By mapping model performance to monetary value, we hope SWE-Lancer enables greater research into the economic impact of AI model development.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2025-02", "2025-03", "2025-05"], "last_mentioned": "2025-05-05T08:17:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2503.03961", "slug": "ed2cad7da4d5cb81", "title": "A Little Depth Goes a Long Way: The Expressive Power of Log-Depth Transformers", "authors": ["William Merrill", "Ashish Sabharwal"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2503.03961", "doi": null, "topics": ["Language models", "Psychology", "Algorithms"], "description": "Recent theoretical results show transformers cannot express sequential reasoning problems over long inputs, intuitively because their computational depth is bounded.", "description_method": "abstract-excerpt", "abstract": "Recent theoretical results show transformers cannot express sequential reasoning problems over long inputs, intuitively because their computational depth is bounded. However, prior work treats the depth as a constant, leaving it unclear to what degree bounded depth may suffice for solving problems over short inputs, or how increasing the transformer's depth affects its expressive power. We address these questions by analyzing transformers whose depth can grow minimally with context length $n$. We show even highly uniform transformers with depth $Θ(\\log n)$ can express two important problems: recognizing regular languages, which captures state tracking abilities and was known to be expressible only by an unconventional, non-uniform model of transformers, and graph connectivity, which underlies multi-step reasoning. Notably, both of these problems cannot be expressed by fixed-depth transformers under standard complexity conjectures, demonstrating the expressivity benefit of growing depth. Moreover, our theory quantitatively predicts how depth must grow with input length to express these problems, showing that depth scaling is more efficient than scaling width or chain-of-thought steps. Empirically, our detailed experiments designed to bridge the expressivity vs. learnability gap reveal that our theoretical depth requirements for regular language recognition closely match the practical depth requirements for successfully training transformers. Thus, our results clarify how depth affects a transformer's reasoning capabilities, and provide practical guidance for effective depth selection for sequential reasoning.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2025-06", "2025-08", "2025-10"], "last_mentioned": "2025-10-08T04:23:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2503.11698", "slug": "54cf2d052c2035a4", "title": "A Comparison of the Cerebras Wafer-Scale Integration Technology with Nvidia GPU-based Systems for Artificial Intelligence", "authors": ["Yudhishthira Kundu", "Manroop Kaur", "Tripty Wig", "Kriti Kumar", "Pushpanjali Kumari", "Vivek Puri", "Manish Arora"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2503.11698", "doi": null, "topics": ["Psychology"], "description": "Cerebras' wafer-scale engine (WSE) technology merges multiple dies on a single wafer.", "description_method": "abstract-excerpt", "abstract": "Cerebras' wafer-scale engine (WSE) technology merges multiple dies on a single wafer. It addresses the challenges of memory bandwidth, latency, and scalability, making it suitable for artificial intelligence. This work evaluates the WSE-3 architecture and compares it with leading GPU-based AI accelerators, notably Nvidia's H100 and B200. The work highlights the advantages of WSE-3 in performance per watt and memory scalability and provides insights into the challenges in manufacturing, thermal management, and reliability. The results suggest that wafer-scale integration can surpass conventional architectures in several metrics, though work is required to address cost-effectiveness and long-term viability.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2025-05", "2025-08", "2025-10"], "last_mentioned": "2025-10-08T18:33:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2503.14499", "slug": "61ebcf7b5dede3dc", "title": "Measuring AI Ability to Complete Long Software Tasks", "authors": ["Thomas Kwa", "Ben West", "Joel Becker", "Amy Deng", "Katharyn Garcia", "Max Hasin", "Sami Jawhar", "Megan Kinniment", "Nate Rush", "Sydney Von Arx", "Ryan Bloom", "Thomas Broadley", "Haoxing Du", "Brian Goodrich", "Nikola Jurkovic", "Luke Harold Miles", "Seraphina Nix", "Tao Lin", "Chris Painter", "Neev Parikh", "David Rein", "Lucas Jun Koba Sato", "Hjalmar Wijk", "Daniel M. Ziegler", "Elizabeth Barnes", "Lawrence Chan"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2503.14499", "doi": null, "topics": ["Other research"], "description": "Despite rapid progress on AI benchmarks, the real-world meaning of benchmark performance remains unclear.", "description_method": "abstract-excerpt", "abstract": "Despite rapid progress on AI benchmarks, the real-world meaning of benchmark performance remains unclear. To quantify the capabilities of AI systems in terms of human capabilities, we propose a new metric: 50%-task-completion time horizon. This is the time humans typically take to complete tasks that AI models can complete with 50% success rate. We first timed humans with relevant domain expertise on a combination of RE-Bench, HCAST, and 66 novel shorter tasks. On these tasks, current frontier AI models such as Claude 3.7 Sonnet have a 50% time horizon of around 50 minutes. Furthermore, frontier AI time horizon has been doubling approximately every seven months since 2019, though the trend may have accelerated in 2024. The increase in AI models' time horizons seems to be primarily driven by greater reliability and ability to adapt to mistakes, combined with better logical reasoning and tool use capabilities. We discuss the limitations of our results -- including their degree of external validity -- and the implications of increased autonomy for dangerous capabilities. If these results generalize to real-world software tasks, extrapolation of this trend predicts that within 5 years, AI systems will be capable of automating many software tasks that currently take humans a month.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2025-04", "2025-05", "2025-10", "2026-05"], "last_mentioned": "2026-05-05T17:43:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2504.09762", "slug": "f9229912b4e38793", "title": "Position: Stop Anthropomorphizing Intermediate Tokens as Reasoning/Thinking Traces!", "authors": ["Subbarao Kambhampati", "Karthik Valmeekam", "Siddhant Bhambri", "Vardhan Palod", "Lucas Saldyt", "Kaya Stechly", "Soumya Rani Samineni", "Durgesh Kalwar", "Upasana Biswas"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2504.09762", "doi": null, "topics": ["Language models"], "description": "Intermediate token generation (ITG), where a model produces output before the solution, has become a standard method to improve the performance of language models on reasoning tasks.", "description_method": "abstract-excerpt", "abstract": "Intermediate token generation (ITG), where a model produces output before the solution, has become a standard method to improve the performance of language models on reasoning tasks. These intermediate tokens have been called \\say{reasoning traces} or even \\say{thinking traces} -- implicitly anthropomorphizing the traces, and implying that these traces resemble steps a human might take when solving a challenging problem, and as such can provide an interpretable window into the operation of the model's thinking process to the end user. In this position paper, we present evidence that this anthropomorphization isn't a harmless metaphor, and instead is quite dangerous -- it confuses the nature of these models and how to use them effectively, and leads to questionable research. We call on the community to avoid such anthropomorphization of intermediate tokens.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2025-10", "2025-11", "2026-06", "2026-08"], "last_mentioned": "2026-08-19T18:46:39Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2504.13171", "slug": "2b3604406707ce40", "title": "Sleep-time Compute: Beyond Inference Scaling at Test-time", "authors": ["Kevin Lin", "Charlie Snell", "Yu Wang", "Charles Packer", "Sarah Wooders", "Ion Stoica", "Joseph E. Gonzalez"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2504.13171", "doi": null, "topics": ["Databases", "Language models"], "description": "Scaling test-time compute has emerged as a key ingredient for enabling large language models (LLMs) to solve difficult problems, but comes with high latency and inference cost.", "description_method": "abstract-excerpt", "abstract": "Scaling test-time compute has emerged as a key ingredient for enabling large language models (LLMs) to solve difficult problems, but comes with high latency and inference cost. We introduce sleep-time compute, which allows models to \"think\" offline about contexts before queries are presented: by anticipating what queries users might ask and pre-computing useful quantities, we can significantly reduce the compute requirements at test-time. To demonstrate the efficacy of our method, we create modified versions of two reasoning tasks - Stateful GSM-Symbolic and Stateful AIME. We find that sleep-time compute can reduce the amount of test-time compute needed to achieve the same accuracy by ~ 5x on Stateful GSM-Symbolic and Stateful AIME and that by scaling sleep-time compute we can further increase accuracy by up to 13% on Stateful GSM-Symbolic and 18% on Stateful AIME. Furthermore, we introduce Multi-Query GSM-Symbolic, which extends GSM-Symbolic by including multiple related queries per context. By amortizing sleep-time compute across related queries about the same context using Multi-Query GSM-Symbolic, we can decrease the average cost per query by 2.5x. We then conduct additional analysis to understand when sleep-time compute is most effective, finding the predictability of the user query to be well correlated with the efficacy of sleep-time compute. Finally, we conduct a case-study of applying sleep-time compute to a realistic agentic SWE task.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2025-05", "2025-07", "2026-05"], "last_mentioned": "2026-05-26T16:37:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2505.11581", "slug": "ca4b8f3c2bd4753c", "title": "Questioning Representational Optimism in Deep Learning: The Fractured Entangled Representation Hypothesis", "authors": ["Akarsh Kumar", "Jeff Clune", "Joel Lehman", "Kenneth O. Stanley"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2505.11581", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Psychology"], "description": "Much of the excitement in modern AI is driven by the observation that scaling up existing systems leads to better performance.", "description_method": "abstract-excerpt", "abstract": "Much of the excitement in modern AI is driven by the observation that scaling up existing systems leads to better performance. But does better performance necessarily imply better internal representations? While the representational optimist assumes it must, this position paper challenges that view. We compare neural networks evolved through an open-ended search process to networks trained via conventional stochastic gradient descent (SGD) on the simple task of generating a single image. This minimal setup offers a unique advantage: each hidden neuron's full functional behavior can be easily visualized as an image, thus revealing how the network's output behavior is internally constructed neuron by neuron. The result is striking: while both networks produce the same output behavior, their internal representations differ dramatically. The SGD-trained networks exhibit a form of disorganization that we term fractured entangled representation (FER). Interestingly, the evolved networks largely lack FER, even approaching a unified factored representation (UFR). In large models, FER may be degrading core model capacities like generalization, creativity, and (continual) learning. Therefore, understanding and mitigating FER could be critical to the future of representation learning.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2025-05", "2025-06", "2025-11", "2026-02"], "last_mentioned": "2026-02-26T12:23:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2505.12540", "slug": "31c2d503a6fe6138", "title": "Harnessing the Universal Geometry of Embeddings", "authors": ["Rishi Jha", "Collin Zhang", "Vitaly Shmatikov", "John X. Morris"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2505.12540", "doi": null, "topics": ["Mathematics", "Statistics & probability", "Security & cryptography"], "description": "We introduce the first method for translating text embeddings from one vector space to another without any paired data, encoders, or predefined sets of matches.", "description_method": "abstract-excerpt", "abstract": "We introduce the first method for translating text embeddings from one vector space to another without any paired data, encoders, or predefined sets of matches. Our unsupervised approach translates any embedding to and from a universal latent representation (i.e., a universal semantic structure conjectured by the Platonic Representation Hypothesis). Our translations achieve high cosine similarity across model pairs with different architectures, parameter counts, and training datasets. The ability to translate unknown embeddings into a different space while preserving their geometry has serious implications for the security of vector databases. An adversary with access only to embedding vectors can extract sensitive information about the underlying documents, sufficient for classification and attribute inference.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2025-12", "2026-01", "2026-03", "2026-07"], "last_mentioned": "2026-07-07T20:34:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2507.06952", "slug": "5afb4d9c1c899daf", "title": "What Has a Foundation Model Found? Using Inductive Bias to Probe for World Models", "authors": ["Keyon Vafa", "Peter G. Chang", "Ashesh Rambachan", "Sendhil Mullainathan"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2507.06952", "doi": null, "topics": ["Psychology", "Physics"], "description": "Foundation models are premised on the idea that sequence prediction can uncover deeper domain understanding, much like how Kepler's predictions of planetary motion later led to the discovery of Newtonian mechanics.", "description_method": "abstract-excerpt", "abstract": "Foundation models are premised on the idea that sequence prediction can uncover deeper domain understanding, much like how Kepler's predictions of planetary motion later led to the discovery of Newtonian mechanics. However, evaluating whether these models truly capture deeper structure remains a challenge. We develop a technique for evaluating foundation models that examines how they adapt to synthetic datasets generated from some postulated world model. Our technique measures whether the foundation model's inductive bias aligns with the world model, and so we refer to it as an inductive bias probe. Across multiple domains, we find that foundation models can excel at their training tasks yet fail to develop inductive biases towards the underlying world model when adapted to new tasks. We particularly find that foundation models trained on orbital trajectories consistently fail to apply Newtonian mechanics when adapted to new physics tasks. Further analysis reveals that these models behave as if they develop task-specific heuristics that fail to generalize.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2025-07", "2025-08", "2025-10", "2026-06"], "last_mentioned": "2026-06-11T18:40:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2508.03385", "slug": "fd6dd8ddb806bef2", "title": "Can We Fix Social Media? Testing Prosocial Interventions using Generative Social Simulation", "authors": ["Maik Larooij", "Petter Törnberg"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2508.03385", "doi": null, "topics": ["Language models", "Algorithms"], "description": "Social media platforms have been widely linked to societal harms, including rising polarization and the erosion of constructive debate.", "description_method": "abstract-excerpt", "abstract": "Social media platforms have been widely linked to societal harms, including rising polarization and the erosion of constructive debate. Can these problems be mitigated through prosocial interventions? We address this question using a novel method - generative social simulation - that embeds Large Language Models within Agent-Based Models to create socially rich synthetic platforms. We create a minimal platform where agents can post, repost, and follow others. We find that the resulting following-networks reproduce three well-documented dysfunctions: (1) partisan echo chambers; (2) concentrated influence among a small elite; and (3) the amplification of polarized voices - creating a 'social media prism' that distorts political discourse. We test six proposed interventions, from chronological feeds to bridging algorithms, finding only modest improvements - and in some cases, worsened outcomes. These results suggest that core dysfunctions may be rooted in the feedback between reactive engagement and network growth, raising the possibility that meaningful reform will require rethinking the foundational dynamics of platform architecture.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2025-08", "2025-10", "2026-03"], "last_mentioned": "2026-03-31T01:40:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2510.07192", "slug": "3175815a362bc3a3", "title": "Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples", "authors": ["Alexandra Souly", "Javier Rando", "Ed Chapman", "Xander Davies", "Burak Hasircioglu", "Ezzeldin Shereen", "Carlos Mougan", "Vasilios Mavroudis", "Erik Jones", "Chris Hicks", "Nicholas Carlini", "Yarin Gal", "Robert Kirk"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2510.07192", "doi": null, "topics": ["Security & cryptography", "Language models"], "description": "Poisoning attacks can compromise the safety of large language models (LLMs) by injecting malicious documents into their training data.", "description_method": "abstract-excerpt", "abstract": "Poisoning attacks can compromise the safety of large language models (LLMs) by injecting malicious documents into their training data. Existing work has studied pretraining poisoning assuming adversaries control a percentage of the training corpus. However, for large models, even small percentages translate to impractically large amounts of data. This work demonstrates for the first time that poisoning attacks instead require a near-constant number of documents regardless of dataset size. We conduct the largest pretraining poisoning experiments to date, pretraining models from 600M to 13B parameters on chinchilla-optimal datasets (6B to 260B tokens). We find that 250 poisoned documents similarly compromise models across all model and dataset sizes, despite the largest models training on more than 20 times more clean data. We also run smaller-scale experiments to ablate factors that could influence attack success, including broader ratios of poisoned to clean data and non-random distributions of poisoned samples. Finally, we demonstrate the same dynamics for poisoning during fine-tuning. Altogether, our results suggest that injecting backdoors through data poisoning may be easier for large models than previously believed as the number of poisons required does not scale up with model size, highlighting the need for more research on defences to mitigate this risk in future models.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2025-10", "2025-11", "2026-01"], "last_mentioned": "2026-01-11T19:01:33Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2602.16800", "slug": "78ebd2f87b4ed8cd", "title": "Large-scale online deanonymization with LLMs", "authors": ["Simon Lermen", "Daniel Paleka", "Joshua Swanson", "Michael Aerni", "Nicholas Carlini", "Florian Tramèr"], "year": 2026, "venue": "arXiv", "url": "https://arxiv.org/abs/2602.16800", "doi": null, "topics": ["Security & cryptography", "Language models", "Networks"], "description": "We show that large language models can be used to perform at-scale deanonymization.", "description_method": "abstract-excerpt", "abstract": "We show that large language models can be used to perform at-scale deanonymization. With full Internet access, our agent can re-identify Hacker News users and Anthropic Interviewer participants at high precision, given pseudonymous online profiles and conversations alone, matching what would take hours for a dedicated human investigator. We then design attacks for the closed-world setting. Given two databases of pseudonymous individuals, each containing unstructured text written by or about that individual, we implement a scalable attack pipeline that uses LLMs to: (1) extract identity-relevant features, (2) search for candidate matches via semantic embeddings, and (3) reason over top candidates to verify matches and reduce false positives. Compared to classical deanonymization work (e.g., on the Netflix prize) that required structured data, our approach works directly on raw user content across arbitrary platforms. We construct three datasets with known ground-truth data to evaluate our attacks. The first links Hacker News to LinkedIn profiles, using cross-platform references that appear in the profiles. Our second dataset matches users across Reddit movie discussion communities; and the third splits a single user's Reddit history in time to create two pseudonymous profiles to be matched. In each setting, LLM-based methods substantially outperform classical baselines, achieving up to 68% recall at 90% precision compared to near 0% for the best non-LLM method. Our results show that the practical obscurity protecting pseudonymous users online no longer holds and that threat models for online privacy need to be reconsidered.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2026-02", "2026-04", "2026-05"], "last_mentioned": "2026-05-01T17:06:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2603.12277", "slug": "d0980ba43accb1d3", "title": "Prompt Injection as Role Confusion", "authors": ["Charles Ye", "Jasmine Cui", "Dylan Hadfield-Menell"], "year": 2026, "venue": "arXiv", "url": "https://arxiv.org/abs/2603.12277", "doi": null, "topics": ["Physics", "Security & cryptography", "Psychology"], "description": "LLMs see the world as a single stream of text, partitioned into roles like or .", "description_method": "abstract-excerpt", "abstract": "LLMs see the world as a single stream of text, partitioned into roles like or . We trace prompt injection to role confusion: models perceive the source of text from how it sounds, not its labeled role. A command hidden in a webpage hijacks an agent simply because it sounds like text, despite its label. We design role probes to measure how LLMs internally perceive \"who is speaking,\" and find that injected text occupies the same representational space as the trusted role it imitates. We demonstrate this with CoT Forgery, a zero-shot attack that injects fabricated reasoning into user prompts and tool outputs. Models mistake the forgery for their own thoughts, yielding 60% attack success against frontier models with near-zero baselines. Strikingly, the degree of role confusion predicts attack success before a single token is generated. This mechanism generalizes beyond CoT Forgery to standard agent prompt injections, revealing prompt injection as a measurable consequence of role perception. To the model, sounding like a role is indistinguishable from being one. Project page and writeup: https://role-confusion.github.io", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2026-04", "2026-07", "2026-08"], "last_mentioned": "2026-08-09T00:24:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2603.20957", "slug": "15189d269b228768", "title": "Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models", "authors": ["Xinyue Liu", "Niloofar Mireshghallah", "Jane C. Ginsburg", "Tuhin Chakrabarty"], "year": 2026, "venue": "arXiv", "url": "https://arxiv.org/abs/2603.20957", "doi": null, "topics": ["Language models", "Security & cryptography"], "description": "Frontier LLM companies have repeatedly assured courts and regulators that their models do not store copies of training data.", "description_method": "abstract-excerpt", "abstract": "Frontier LLM companies have repeatedly assured courts and regulators that their models do not store copies of training data. They further rely on safety alignment strategies via RLHF, system prompts, and output filters to block verbatim regurgitation of copyrighted works, and have cited the efficacy of these measures in their legal defenses against copyright infringement claims. We show that finetuning bypasses these protections: by training models to expand plot summaries into full text, a task naturally suited for commercial writing assistants, we cause GPT-4o, Gemini-2.5-Pro, and DeepSeek-V3.1 to reproduce up to 85-90% of held-out copyrighted books, with single verbatim spans exceeding 460 words, using only semantic descriptions as prompts and no actual book text. This extraction generalizes across authors: finetuning exclusively on Haruki Murakami's novels unlocks verbatim recall of copyrighted books from over 30 unrelated authors. The effect is not specific to any training author or corpus: random author pairs and public-domain finetuning data produce comparable extraction, while finetuning on synthetic text yields near-zero extraction, indicating that finetuning on individual authors' works reactivates latent memorization from pretraining. Three models from different providers memorize the same books in the same regions ($r \\ge 0.90$), pointing to an industry-wide vulnerability. Our findings offer compelling evidence that model weights store copies of copyrighted works and that the security failures that manifest after finetuning on individual authors' works undermine a key premise of recent fair use rulings, where courts have conditioned favorable outcomes on the adequacy of measures preventing reproduction of protected expression.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2026-03", "2026-04", "2026-06", "2026-07"], "last_mentioned": "2026-07-01T01:24:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:astro-ph/0102126", "slug": "67036f72b24060de", "title": "Astronomical engineering: a strategy for modifying planetary orbits", "authors": ["D. G. Korycansky", "Gregory Laughlin", "Fred C. Adams"], "year": 2001, "venue": "arXiv", "url": "https://arxiv.org/abs/astro-ph/0102126", "doi": "10.1023/A:1002790227314", "topics": ["Physics"], "description": "The Sun's gradual brightening will seriously compromise the Earth's biosphere within ~ 1E9 years.", "description_method": "abstract-excerpt", "abstract": "The Sun's gradual brightening will seriously compromise the Earth's biosphere within ~ 1E9 years. If Earth's orbit migrates outward, however, the biosphere could remain intact over the entire main-sequence lifetime of the Sun. In this paper, we explore the feasibility of engineering such a migration over a long time period. The basic mechanism uses gravitational assists to (in effect) transfer orbital energy from Jupiter to the Earth, and thereby enlarges the orbital radius of Earth. This transfer is accomplished by a suitable intermediate body, either a Kuiper Belt object or a main belt asteroid. The object first encounters Earth during an inward pass on its initial highly elliptical orbit of large (~ 300 AU) semimajor axis. The encounter transfers energy from the object to the Earth in standard gravity-assist fashion by passing close to the leading limb of the planet. The resulting outbound trajectory of the object must cross the orbit of Jupiter; with proper timing, the outbound object encounters Jupiter and picks up the energy it lost to Earth. With small corrections to the trajectory, or additional planetary encounters (e.g., with Saturn), the object can repeat this process over many encounters. To maintain its present flux of solar energy, the Earth must experience roughly one encounter every 6000 years (for an object mass of 1E22 g). We develop the details of this scheme and discuss its ramifications.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2019-11", "2021-03", "2021-08", "2025-03"], "last_mentioned": "2025-03-20T17:42:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:astro-ph/9912202", "slug": "9649bdb796e921a5", "title": "The Effects of Moore's Law and Slacking on Large Computations", "authors": ["C Gottbrath", "J Bailin", "C Meakin", "T Thompson", "J. J. Charfman"], "year": 1999, "venue": "arXiv", "url": "https://arxiv.org/abs/astro-ph/9912202", "doi": null, "topics": ["Other research"], "description": "We show that, in the context of Moore's Law, overall productivity can be increased for large enough computations by `slacking' or waiting for some period of time before purchasing a computer and beginning the calculation.", "description_method": "abstract-excerpt", "abstract": "We show that, in the context of Moore's Law, overall productivity can be increased for large enough computations by `slacking' or waiting for some period of time before purchasing a computer and beginning the calculation.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2009-10", "2019-11", "2022-04", "2026-01"], "last_mentioned": "2026-01-31T14:35:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:cs/0111050", "slug": "35c0ba8de69cbc59", "title": "Smoothed Analysis of Algorithms: Why the Simplex Algorithm Usually Takes Polynomial Time", "authors": ["Daniel A. Spielman", "Shang-Hua Teng"], "year": 2001, "venue": "arXiv", "url": "https://arxiv.org/abs/cs/0111050", "doi": null, "topics": ["Algorithms"], "description": "We introduce the smoothed analysis of algorithms, which is a hybrid of the worst-case and average-case analysis of algorithms.", "description_method": "abstract-excerpt", "abstract": "We introduce the smoothed analysis of algorithms, which is a hybrid of the worst-case and average-case analysis of algorithms. In smoothed analysis, we measure the maximum over inputs of the expected performance of an algorithm under small random perturbations of that input. We measure this performance in terms of both the input size and the magnitude of the perturbations. We show that the simplex algorithm has polynomial smoothed complexity.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2011-01", "2016-06", "2022-06", "2024-11"], "last_mentioned": "2024-11-02T13:34:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:math/9405205", "slug": "3c019a15374add19", "title": "Seven Trees in One", "authors": ["Andreas Blass"], "year": 1994, "venue": "arXiv", "url": "https://arxiv.org/abs/math/9405205", "doi": null, "topics": ["Programming languages", "Mathematics"], "description": "Following a remark of Lawvere, we explicitly exhibit a particularly elementary bijection between the set T of finite binary trees and the set T^7 of seven-tuples of such trees.", "description_method": "abstract-excerpt", "abstract": "Following a remark of Lawvere, we explicitly exhibit a particularly elementary bijection between the set T of finite binary trees and the set T^7 of seven-tuples of such trees. \"Particularly elementary\" means that the application of the bijection to a seven-tuple of trees involves case distinctions only down to a fixed depth (namely four) in the given seven-tuple. We clarify how this and similar bijections are related to the free commutative semiring on one generator X subject to X=1+X^2. Finally, our main theorem is that the existence of particularly elementary bijections can be deduced from the provable existence, in intuitionistic type theory, of any bijections at all.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2015-06", "2022-10", "2023-01", "2024-05"], "last_mentioned": "2024-05-20T18:57:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:quant-ph/0205095", "slug": "fd549bed2247f7a1", "title": "Circuit for Shor's algorithm using 2n+3 qubits", "authors": ["Stephane Beauregard"], "year": 2002, "venue": "arXiv", "url": "https://arxiv.org/abs/quant-ph/0205095", "doi": null, "topics": ["Algorithms", "Physics"], "description": "We try to minimize the number of qubits needed to factor an integer of n bits using Shor's algorithm on a quantum computer.", "description_method": "abstract-excerpt", "abstract": "We try to minimize the number of qubits needed to factor an integer of n bits using Shor's algorithm on a quantum computer. We introduce a circuit which uses 2n+3 qubits and O(n^3 lg(n)) elementary quantum gates in a depth of O(n^3) to implement the factorization algorithm. The circuit is computable in polynomial time on a classical computer and is completely general as it does not rely on any property of the number to be factored. Keywords: Factorization, quantum circuits, modular arithmetics", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-12", "2022-07", "2024-03", "2025-04"], "last_mentioned": "2025-04-22T17:06:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/463608a", "slug": "c35698d0ba6cfd85", "title": "Political instability may be a contributor in the coming decade", "authors": ["Peter Turchin"], "year": 2010, "venue": "Nature", "url": "https://doi.org/10.1038/463608a", "doi": "10.1038/463608a", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-06", "2020-10", "2020-11", "2022-07"], "last_mentioned": "2022-07-20T21:10:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/483260a", "slug": "834ea72087261af2", "title": "The split brain: A tale of two halves", "authors": ["David Wolman"], "year": 2012, "venue": "Nature", "url": "https://doi.org/10.1038/483260a", "doi": "10.1038/483260a", "topics": ["Neuroscience"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2022-02", "2023-11", "2024-10", "2025-03"], "last_mentioned": "2025-03-16T22:58:31Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature.2012.11555", "slug": "a07748548f293d67", "title": "DNA has a 521-year half-life", "authors": ["Matt Kaplan"], "year": 2012, "venue": "Nature", "url": "https://doi.org/10.1038/nature.2012.11555", "doi": "10.1038/nature.2012.11555", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-10", "2023-02", "2023-10", "2023-12"], "last_mentioned": "2023-12-11T02:28:07Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature02842", "slug": "079e6ae6d7553e18", "title": "Modelling the recent common ancestry of all living humans", "authors": ["Douglas L. T. Rohde", "Steve Olson", "Joseph T. Chang"], "year": 2004, "venue": "Nature", "url": "https://doi.org/10.1038/nature02842", "doi": "10.1038/nature02842", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2019-03", "2019-06", "2020-06", "2023-04"], "last_mentioned": "2023-04-27T10:21:10Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature12914", "slug": "dd68caae831eebd3", "title": "Rate of tree carbon accumulation increases continuously with tree size", "authors": ["N. L. Stephenson", "A. J. Das", "R. Condit", "S. E. Russo", "P. J. Baker", "N. G. Beckman", "D. A. Coomes", "E. R. Lines", "W. K. Morris", "N. Rüger", "E. Álvarez", "C. Blundo", "S. Bunyavejchewin", "G. Chuyong", "S. J. Davies", "Á. Duque", "C. N. Ewango", "O. Flores", "J. F. Franklin", "H. R. Grau", "Z. Hao", "M. E. Harmon", "S. P. Hubbell", "D. Kenfack", "Y. Lin", "J.-R. Makana", "A. Malizia", "L. R. Malizia", "R. J. Pabst", "N. Pongpattananurak", "S.-H. Su", "I-F. Sun", "S. Tan", "D. Thomas", "P. J. van Mantgem", "X. Wang", "S. K. Wiser", "M. A. Zavala"], "year": 2014, "venue": "Nature", "url": "https://doi.org/10.1038/nature12914", "doi": "10.1038/nature12914", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-08", "2019-02", "2024-03", "2024-06"], "last_mentioned": "2024-06-23T20:37:14Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nclimate3322", "slug": "5a71b93e9cf8bc85", "title": "Global risk of deadly heat", "authors": ["Camilo Mora", "Bénédicte Dousset", "Iain R. Caldwell", "Farrah E. Powell", "Rollan C. Geronimo", "Coral R. Bielecki", "Chelsie W. W. Counsell", "Bonnie S. Dietrich", "Emily T. Johnston", "Leo V. Louis", "Matthew P. Lucas", "Marie M. McKenzie", "Alessandra G. Shea", "Han Tseng", "Thomas W. Giambelluca", "Lisa R. Leon", "Ed Hawkins", "Clay Trauernicht"], "year": 2017, "venue": "Nature Climate Change", "url": "https://doi.org/10.1038/nclimate3322", "doi": "10.1038/nclimate3322", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2018-04", "2021-06", "2021-07"], "last_mentioned": "2021-07-08T09:40:29Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/ncomms12808", "slug": "44a007ddc3c3f237", "title": "Extremotolerant tardigrade genome and improved radiotolerance of human cultured cells by tardigrade-unique protein", "authors": ["Takuma Hashimoto", "Daiki D. Horikawa", "Yuki Saito", "Hirokazu Kuwahara", "Hiroko Kozuka-Hata", "Tadasu Shin-I", "Yohei Minakuchi", "Kazuko Ohishi", "Ayuko Motoyama", "Tomoyuki Aizu", "Atsushi Enomoto", "Koyuki Kondo", "Sae Tanaka", "Yuichiro Hara", "Shigeyuki Koshikawa", "Hiroshi Sagara", "Toru Miura", "Shin-ichi Yokobori", "Kiyoshi Miyagawa", "Yutaka Suzuki", "Takeo Kubo", "Masaaki Oyama", "Yuji Kohara", "Asao Fujiyama", "Kazuharu Arakawa", "Toshiaki Katayama", "Atsushi Toyoda", "Takekazu Kunieda"], "year": 2016, "venue": "Nature Communications", "url": "https://doi.org/10.1038/ncomms12808", "doi": "10.1038/ncomms12808", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2016-09", "2016-10", "2023-04", "2023-05"], "last_mentioned": "2023-05-09T05:15:07Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41386-022-01378-2", "slug": "892fa6fdf09eec23", "title": "Regulation of social hierarchy learning by serotonin transporter availability", "authors": ["Remi Janet", "Romain Ligneul", "Annabel B. Losecaat-Vermeer", "Remi Philippe", "Gabriele Bellucci", "Edmund Derrington", "Soyoung Q. Park", "Jean-Claude Dreher"], "year": 2022, "venue": "Neuropsychopharmacology", "url": "https://doi.org/10.1038/s41386-022-01378-2", "doi": "10.1038/s41386-022-01378-2", "topics": ["Neuroscience"], "description": "AbstractLearning one’s status in a group is a fundamental process in building social hierarchies.", "description_method": "abstract-excerpt", "abstract": "AbstractLearning one’s status in a group is a fundamental process in building social hierarchies. Although animal studies suggest that serotonin (5-HT) signaling modulates learning social hierarchies, direct evidence in humans is lacking. Here we determined the relationship between serotonin transporter (SERT) availability and brain systems engaged in learning social ranks combining computational approaches with simultaneous PET-fMRI acquisition in healthy males. We also investigated the link between SERT availability and brain activity in a non-social control condition involving learning the payoffs of slot machines. Learning social ranks was modulated by the dorsal raphe nucleus (DRN) 5-HT function. BOLD ventral striatal response, tracking the rank of opponents, decreased with DRN SERT levels. Moreover, this link was specific to the social learning task. These findings demonstrate that 5-HT plays an influence on the computations required to learn social ranks.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-01", "2023-06", "2026-05"], "last_mentioned": "2026-05-02T17:03:38Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41398-021-01682-3", "slug": "5d6485124fa29af2", "title": "Expectancy effects on serotonin and dopamine transporters during SSRI treatment of social anxiety disorder: a randomized clinical trial", "authors": ["Olof R. Hjorth", "Andreas Frick", "Malin Gingnell", "Johanna M. Hoppe", "Vanda Faria", "Sara Hultberg", "Iman Alaie", "Kristoffer N. T. Månsson", "Jörgen Rosén", "Margareta Reis", "Kurt Wahlstedt", "My Jonasson", "Mark Lubberink", "Gunnar Antoni", "Mats Fredrikson", "Tomas Furmark"], "year": 2021, "venue": "Translational Psychiatry", "url": "https://doi.org/10.1038/s41398-021-01682-3", "doi": "10.1038/s41398-021-01682-3", "topics": ["Biology & medicine", "Psychology", "Neuroscience"], "description": "AbstractIt has been extensively debated whether selective serotonin reuptake inhibitors (SSRIs) are more efficacious than placebo in affective disorders, and it is not fully understood how SSRIs exert their beneficial effects.", "description_method": "abstract-excerpt", "abstract": "AbstractIt has been extensively debated whether selective serotonin reuptake inhibitors (SSRIs) are more efficacious than placebo in affective disorders, and it is not fully understood how SSRIs exert their beneficial effects. Along with serotonin transporter blockade, altered dopamine signaling and psychological factors may contribute. In this randomized clinical trial of participants with social anxiety disorder (SAD) we investigated how manipulation of verbally-induced expectancies, vital for placebo response, affect brain monoamine transporters and symptom improvement during SSRI treatment. Twenty-seven participants with SAD (17 men, 10 women), were randomized, to 9 weeks of overt or covert treatment with escitalopram 20 mg. The overt group received correct treatment information whereas the covert group was treated deceptively with escitalopram, described as an active placebo in a cover story. Before and after treatment, patients underwent positron emission tomography (PET) assessments with the [11C]DASB and [11C]PE2I radiotracers, probing brain serotonin (SERT) and dopamine (DAT) transporters. SAD symptoms were measured by the Liebowitz Social Anxiety Scale. Overt was superior to covert SSRI treatment, resulting in almost a fourfold higher rate of responders. PET results showed that SERT occupancy after treatment was unrelated to anxiety reduction and equally high in both groups. In contrast, DAT binding decreased in the right putamen, pallidum, and the left thalamus with overt SSRI treatment, and increased with covert treatment, resulting in significant group differences. DAT binding potential changes in these regions correlated negatively with symptom improvement. Findings support that the anxiolytic effects of SSRIs involve psychological factors contingent on dopaminergic neurotransmission while serotonin transporter blockade alone is insufficient for clinical response.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-07", "2023-01", "2023-08", "2024-10"], "last_mentioned": "2024-10-12T19:08:52Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-021-24487-w", "slug": "96ddbc0f61918f58", "title": "The mortality cost of carbon", "authors": ["R. Daniel Bressler"], "year": 2021, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-021-24487-w", "doi": "10.1038/s41467-021-24487-w", "topics": ["Science & research"], "description": "Abstract Many studies project that climate change can cause a significant number of excess deaths.", "description_method": "abstract-excerpt", "abstract": "Abstract Many studies project that climate change can cause a significant number of excess deaths. Yet, in integrated assessment models (IAMs) that determine the social cost of carbon (SCC) and prescribe optimal climate policy, human mortality impacts are limited and not updated to the latest scientific understanding. This study extends the DICE-2016 IAM to explicitly include temperature-related mortality impacts by estimating a climate-mortality damage function. We introduce a metric, the mortality cost of carbon (MCC), that estimates the number of deaths caused by the emissions of one additional metric ton of CO2. In the baseline emissions scenario, the 2020 MCC is 2.26 × 10 ‒4 [low to high estimate −1.71× 10 ‒4 to 6.78 × 10 ‒4 ] excess deaths per metric ton of 2020 emissions. This implies that adding 4,434 metric tons of carbon dioxide in 2020—equivalent to the lifetime emissions of 3.5 average Americans—causes one excess death globally in expectation between 2020-2100. Incorporating mortality costs increases the 2020 SCC from $37 to $258 [−$69 to $545] per metric ton in the baseline emissions scenario. Optimal climate policy changes from gradual emissions reductions starting in 2050 to full decarbonization by 2050 when mortality is considered.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-08", "2021-10", "2022-04", "2022-09"], "last_mentioned": "2022-09-29T03:27:11Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-022-28457-8", "slug": "acfecd65b4ec5bcf", "title": "Highly efficient and salt rejecting solar evaporation via a wick-free confined water layer", "authors": ["Lenan Zhang", "Xiangyu Li", "Yang Zhong", "Arny Leroy", "Zhenyuan Xu", "Lin Zhao", "Evelyn N. Wang"], "year": 2022, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-022-28457-8", "doi": "10.1038/s41467-022-28457-8", "topics": ["Biology & medicine"], "description": "AbstractRecent advances in thermally localized solar evaporation hold significant promise for vapor generation, seawater desalination, wastewater treatment, and medical sterilization.", "description_method": "abstract-excerpt", "abstract": "AbstractRecent advances in thermally localized solar evaporation hold significant promise for vapor generation, seawater desalination, wastewater treatment, and medical sterilization. However, salt accumulation is one of the key bottlenecks for reliable adoption. Here, we demonstrate highly efficient (>80% solar-to-vapor conversion efficiency) and salt rejecting (20 weight % salinity) solar evaporation by engineering the fluidic flow in a wick-free confined water layer. With mechanistic modeling and experimental characterization of salt transport, we show that natural convection can be triggered in the confined water. More notably, there exists a regime enabling simultaneous thermal localization and salt rejection, i.e., natural convection significantly accelerates salt rejection while inducing negligible additional heat loss. Furthermore, we show the broad applicability by integrating this confined water layer with a recently developed contactless solar evaporator and report an improved efficiency. This work elucidates the fundamentals of salt transport and offers a low-cost strategy for high-performance solar evaporation.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2022-02", "2022-08", "2026-06"], "last_mentioned": "2026-06-06T14:18:33Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-022-35576-9", "slug": "5ffd48a022726935", "title": "Exposure to the Russian Internet Research Agency foreign influence campaign on Twitter in the 2016 US election and its relationship to attitudes and voting behavior", "authors": ["Gregory Eady", "Tom Paskhalis", "Jan Zilinsky", "Richard Bonneau", "Jonathan Nagler", "Joshua A. Tucker"], "year": 2023, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-022-35576-9", "doi": "10.1038/s41467-022-35576-9", "topics": ["Psychology", "Networks"], "description": "Abstract There is widespread concern that foreign actors are using social media to interfere in elections worldwide.", "description_method": "abstract-excerpt", "abstract": "Abstract There is widespread concern that foreign actors are using social media to interfere in elections worldwide. Yet data have been unavailable to investigate links between exposure to foreign influence campaigns and political behavior. Using longitudinal survey data from US respondents linked to their Twitter feeds, we quantify the relationship between exposure to the Russian foreign influence campaign and attitudes and voting behavior in the 2016 US election. We demonstrate, first, that exposure to Russian disinformation accounts was heavily concentrated: only 1% of users accounted for 70% of exposures. Second, exposure was concentrated among users who strongly identified as Republicans. Third, exposure to the Russian influence campaign was eclipsed by content from domestic news media and politicians. Finally, we find no evidence of a meaningful relationship between exposure to the Russian foreign influence campaign and changes in attitudes, polarization, or voting behavior. The results have implications for understanding the limits of election interference campaigns on social media.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2023-02", "2023-12", "2025-06"], "last_mentioned": "2025-06-10T17:38:34Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-024-47777-5", "slug": "3035fe4bb81a6a13", "title": "Ambient carbon dioxide concentration correlates with SARS-CoV-2 aerostability and infection risk", "authors": ["Allen Haddrell", "Henry Oswin", "Mara Otero-Fernandez", "Joshua F. Robinson", "Tristan Cogan", "Robert Alexander", "Jamie F. S. Mann", "Darryl Hill", "Adam Finn", "Andrew D. Davidson", "Jonathan P. Reid"], "year": 2024, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-024-47777-5", "doi": "10.1038/s41467-024-47777-5", "topics": ["Biology & medicine", "Physics"], "description": "Abstract An improved understanding of the underlying physicochemical properties of respiratory aerosol that influence viral infectivity may open new avenues to mitigate the transmission of respiratory diseases such as COVID-19.", "description_method": "abstract-excerpt", "abstract": "Abstract An improved understanding of the underlying physicochemical properties of respiratory aerosol that influence viral infectivity may open new avenues to mitigate the transmission of respiratory diseases such as COVID-19. Previous studies have shown that an increase in the pH of respiratory aerosols following generation due to changes in the gas-particle partitioning of pH buffering bicarbonate ions and carbon dioxide is a significant factor in reducing SARS-CoV-2 infectivity. We show here that a significant increase in SARS-CoV-2 aerostability results from a moderate increase in the atmospheric carbon dioxide concentration (e.g. 800 ppm), an effect that is more marked than that observed for changes in relative humidity. We model the likelihood of COVID-19 transmission on the ambient concentration of CO 2 , concluding that even this moderate increase in CO 2 concentration results in a significant increase in overall risk. These observations confirm the critical importance of ventilation and maintaining low CO 2 concentrations in indoor environments for mitigating disease transmission. Moreover, the correlation of increased CO 2 concentration with viral aerostability need to be better understood when considering the consequences of increases in ambient CO 2 levels in our atmosphere.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-06", "2024-08", "2025-03", "2026-01"], "last_mentioned": "2026-01-14T22:00:11Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-024-53922-x", "slug": "8ae10fc84675ceb7", "title": "The massed-spaced learning effect in non-neural human cells", "authors": ["N. V. Kukushkin", "R. E. Carney", "T. Tabassum", "T. J. Carew"], "year": 2024, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-024-53922-x", "doi": "10.1038/s41467-024-53922-x", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-11", "2024-12", "2025-01", "2025-05"], "last_mentioned": "2025-05-27T15:38:45Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41539-017-0012-7", "slug": "a6e987372d9de0f9", "title": "Montessori education: a review of the evidence base", "authors": ["Chloë Marshall"], "year": 2017, "venue": "npj Science of Learning", "url": "https://doi.org/10.1038/s41539-017-0012-7", "doi": "10.1038/s41539-017-0012-7", "topics": ["Other research"], "description": "AbstractThe Montessori educational method has existed for over 100 years, but evaluations of its effectiveness are scarce.", "description_method": "abstract-excerpt", "abstract": "AbstractThe Montessori educational method has existed for over 100 years, but evaluations of its effectiveness are scarce. This review paper has three aims, namely to (1) identify some key elements of the method, (2) review existing evaluations of Montessori education, and (3) review studies that do not explicitly evaluate Montessori education but which evaluate the key elements identified in (1). The goal of the paper is therefore to provide a review of the evidence base for Montessori education, with the dual aspirations of stimulating future research and helping teachers to better understand whether and why Montessori education might be effective.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2022-12", "2023-02", "2023-08"], "last_mentioned": "2023-08-26T20:36:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41550-018-0529-6", "slug": "42a15c0184cc9630", "title": "Inventory of CO2 available for terraforming Mars", "authors": ["Bruce M. Jakosky", "Christopher S. Edwards"], "year": 2018, "venue": "Nature Astronomy", "url": "https://doi.org/10.1038/s41550-018-0529-6", "doi": "10.1038/s41550-018-0529-6", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-07", "2018-09", "2018-10", "2019-11"], "last_mentioned": "2019-11-17T14:32:02Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41558-020-0797-x", "slug": "af6fb57ccd403667", "title": "Temporary reduction in daily global CO2 emissions during the COVID-19 forced confinement", "authors": ["Corinne Le Quéré", "Robert B. Jackson", "Matthew W. Jones", "Adam J. P. Smith", "Sam Abernethy", "Robbie M. Andrew", "Anthony J. De-Gol", "David R. Willis", "Yuli Shan", "Josep G. Canadell", "Pierre Friedlingstein", "Felix Creutzig", "Glen P. Peters"], "year": 2020, "venue": "Nature Climate Change", "url": "https://doi.org/10.1038/s41558-020-0797-x", "doi": "10.1038/s41558-020-0797-x", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2020-06", "2020-07", "2021-08"], "last_mentioned": "2021-08-26T14:55:12Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41562-017-0082", "slug": "c7470c7951b243a7", "title": "Why people prefer unequal societies", "authors": ["Christina Starmans", "Mark Sheskin", "Paul Bloom"], "year": 2017, "venue": "Nature Human Behaviour", "url": "https://doi.org/10.1038/s41562-017-0082", "doi": "10.1038/s41562-017-0082", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-04", "2017-05", "2017-06", "2023-04"], "last_mentioned": "2023-04-05T16:40:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41562-018-0399-z", "slug": "d29b74c31f1bd491", "title": "Evaluating the replicability of social science experiments in Nature and Science between 2010 and 2015", "authors": ["Colin F. Camerer", "Anna Dreber", "Felix Holzmeister", "Teck-Hua Ho", "Jürgen Huber", "Magnus Johannesson", "Michael Kirchler", "Gideon Nave", "Brian A. Nosek", "Thomas Pfeiffer", "Adam Altmejd", "Nick Buttrick", "Taizan Chan", "Yiling Chen", "Eskil Forsell", "Anup Gampa", "Emma Heikensten", "Lily Hummer", "Taisuke Imai", "Siri Isaksson", "Dylan Manfredi", "Julia Rose", "Eric-Jan Wagenmakers", "Hang Wu"], "year": 2018, "venue": "Nature Human Behaviour", "url": "https://doi.org/10.1038/s41562-018-0399-z", "doi": "10.1038/s41562-018-0399-z", "topics": ["Science & research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-09", "2021-08", "2022-06", "2025-02"], "last_mentioned": "2025-02-26T09:07:23Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41576-018-0055-5", "slug": "be2c63a82b94842d", "title": "The phylogenomics of evolving virus virulence", "authors": ["Jemma L. Geoghegan", "Edward C. Holmes"], "year": 2018, "venue": "Nature Reviews Genetics", "url": "https://doi.org/10.1038/s41576-018-0055-5", "doi": "10.1038/s41576-018-0055-5", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2021-07", "2021-08", "2021-12"], "last_mentioned": "2021-12-15T08:09:46Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41579-021-00573-0", "slug": "6906649af7330c39", "title": "SARS-CoV-2 variants, spike mutations and immune escape", "authors": ["William T. Harvey", "Alessandro M. Carabelli", "Ben Jackson", "Ravindra K. Gupta", "Emma C. Thomson", "Ewan M. Harrison", "Catherine Ludden", "Richard Reeve", "Andrew Rambaut", "COVID-19 Genomics UK (COG-UK) Consortium", "Sharon J. Peacock", "David L. Robertson"], "year": 2021, "venue": "Nature Reviews Microbiology", "url": "https://doi.org/10.1038/s41579-021-00573-0", "doi": "10.1038/s41579-021-00573-0", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2021-07", "2021-08", "2021-09"], "last_mentioned": "2021-09-22T19:54:45Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-018-0411-9", "slug": "b70fa69e5b0e6817", "title": "Global land change from 1982 to 2016", "authors": ["Xiao-Peng Song", "Matthew C. Hansen", "Stephen V. Stehman", "Peter V. Potapov", "Alexandra Tyukavina", "Eric F. Vermote", "John R. Townshend"], "year": 2018, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-018-0411-9", "doi": "10.1038/s41586-018-0411-9", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2019-07", "2020-08", "2021-07", "2022-09"], "last_mentioned": "2022-09-14T03:49:24Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-018-0707-9", "slug": "ab16a1c16ef617da", "title": "Flight of an aeroplane with solid-state propulsion", "authors": ["Haofeng Xu", "Yiou He", "Kieran L. Strobel", "Christopher K. Gilmore", "Sean P. Kelley", "Cooper C. Hennick", "Thomas Sebastian", "Mark R. Woolston", "David J. Perreault", "Steven R. H. Barrett"], "year": 2018, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-018-0707-9", "doi": "10.1038/s41586-018-0707-9", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-11", "2018-12", "2019-08", "2022-06"], "last_mentioned": "2022-06-30T15:52:09Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-019-1666-5", "slug": "928ec3d9f3eb78ea", "title": "Quantum supremacy using a programmable superconducting processor", "authors": ["Frank Arute", "Kunal Arya", "Ryan Babbush", "Dave Bacon", "Joseph C. Bardin", "Rami Barends", "Rupak Biswas", "Sergio Boixo", "Fernando G. S. L. Brandao", "David A. Buell", "Brian Burkett", "Yu Chen", "Zijun Chen", "Ben Chiaro", "Roberto Collins", "William Courtney", "Andrew Dunsworth", "Edward Farhi", "Brooks Foxen", "Austin Fowler", "Craig Gidney", "Marissa Giustina", "Rob Graff", "Keith Guerin", "Steve Habegger", "Matthew P. Harrigan", "Michael J. Hartmann", "Alan Ho", "Markus Hoffmann", "Trent Huang", "Travis S. Humble", "Sergei V. Isakov", "Evan Jeffrey", "Zhang Jiang", "Dvir Kafri", "Kostyantyn Kechedzhi", "Julian Kelly", "Paul V. Klimov", "Sergey Knysh", "Alexander Korotkov", "Fedor Kostritsa", "David Landhuis", "Mike Lindmark", "Erik Lucero", "Dmitry Lyakh", "Salvatore Mandrà", "Jarrod R. McClean", "Matthew McEwen", "Anthony Megrant", "Xiao Mi", "Kristel Michielsen", "Masoud Mohseni", "Josh Mutus", "Ofer Naaman", "Matthew Neeley", "Charles Neill", "Murphy Yuezhen Niu", "Eric Ostby", "Andre Petukhov", "John C. Platt", "Chris Quintana", "Eleanor G. Rieffel", "Pedram Roushan", "Nicholas C. Rubin", "Daniel Sank", "Kevin J. Satzinger", "Vadim Smelyanskiy", "Kevin J. Sung", "Matthew D. Trevithick", "Amit Vainsencher", "Benjamin Villalonga", "Theodore White", "Z. Jamie Yao", "Ping Yeh", "Adam Zalcman", "Hartmut Neven", "John M. Martinis"], "year": 2019, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-019-1666-5", "doi": "10.1038/s41586-019-1666-5", "topics": ["Physics", "Computer systems"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-05", "2021-07", "2022-06", "2023-07"], "last_mentioned": "2023-07-10T22:05:29Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-022-04473-y", "slug": "bd6d53f48d14938d", "title": "Thermophotovoltaic efficiency of 40%", "authors": ["Alina LaPotin", "Kevin L. Schulte", "Myles A. Steiner", "Kyle Buznitsky", "Colin C. Kelsall", "Daniel J. Friedman", "Eric J. Tervo", "Ryan M. France", "Michelle R. Young", "Andrew Rohskopf", "Shomik Verma", "Evelyn N. Wang", "Asegun Henry"], "year": 2022, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-022-04473-y", "doi": "10.1038/s41586-022-04473-y", "topics": ["Biology & medicine"], "description": "Abstract Thermophotovoltaics (TPVs) convert predominantly infrared wavelength light to electricity via the photovoltaic effect, and can enable approaches to energy storage 1,2 and conversion 3–9 that use higher temperature heat sources than the turbines that are ubiquitous in electricity production today.", "description_method": "abstract-excerpt", "abstract": "Abstract Thermophotovoltaics (TPVs) convert predominantly infrared wavelength light to electricity via the photovoltaic effect, and can enable approaches to energy storage 1,2 and conversion 3–9 that use higher temperature heat sources than the turbines that are ubiquitous in electricity production today. Since the first demonstration of 29% efficient TPVs (Fig. 1a) using an integrated back surface reflector and a tungsten emitter at 2,000 °C (ref. 10 ), TPV fabrication and performance have improved 11,12 . However, despite predictions that TPV efficiencies can exceed 50% (refs. 11,13,14 ), the demonstrated efficiencies are still only as high as 32%, albeit at much lower temperatures below 1,300 °C (refs. 13–15 ). Here we report the fabrication and measurement of TPV cells with efficiencies of more than 40% and experimentally demonstrate the efficiency of high-bandgap tandem TPV cells. The TPV cells are two-junction devices comprising III–V materials with bandgaps between 1.0 and 1.4 eV that are optimized for emitter temperatures of 1,900–2,400 °C. The cells exploit the concept of band-edge spectral filtering to obtain high efficiency, using highly reflective back surface reflectors to reject unusable sub-bandgap radiation back to the emitter. A 1.4/1.2 eV device reached a maximum efficiency of (41.1 ± 1)% operating at a power density of 2.39 W cm –2 and an emitter temperature of 2,400 °C. A 1.2/1.0 eV device reached a maximum efficiency of (39.3 ± 1)% operating at a power density of 1.8 W cm –2 and an emitter temperature of 2,127 °C. These cells can be integrated into a TPV system for thermal energy grid storage to enable dispatchable renewable energy. This creates a pathway for thermal energy grid storage to reach sufficiently high efficiency and sufficiently low cost to enable decarbonization of the electricity grid.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2023-10", "2023-11", "2025-01", "2025-02"], "last_mentioned": "2025-02-18T21:08:23Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-022-04569-5", "slug": "6005cd0f9b98ef79", "title": "SARS-CoV-2 is associated with changes in brain structure in UK Biobank", "authors": ["Gwenaëlle Douaud", "Soojin Lee", "Fidel Alfaro-Almagro", "Christoph Arthofer", "Chaoyue Wang", "Paul McCarthy", "Frederik Lange", "Jesper L. R. Andersson", "Ludovica Griffanti", "Eugene Duff", "Saad Jbabdi", "Bernd Taschler", "Peter Keating", "Anderson M. Winkler", "Rory Collins", "Paul M. Matthews", "Naomi Allen", "Karla L. Miller", "Thomas E. Nichols", "Stephen M. Smith"], "year": 2022, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-022-04569-5", "doi": "10.1038/s41586-022-04569-5", "topics": ["Neuroscience", "Biology & medicine"], "description": "Abstract There is strong evidence of brain-related abnormalities in COVID-19 1–13 .", "description_method": "abstract-excerpt", "abstract": "Abstract There is strong evidence of brain-related abnormalities in COVID-19 1–13 . However, it remains unknown whether the impact of SARS-CoV-2 infection can be detected in milder cases, and whether this can reveal possible mechanisms contributing to brain pathology. Here we investigated brain changes in 785 participants of UK Biobank (aged 51–81 years) who were imaged twice using magnetic resonance imaging, including 401 cases who tested positive for infection with SARS-CoV-2 between their two scans—with 141 days on average separating their diagnosis and the second scan—as well as 384 controls. The availability of pre-infection imaging data reduces the likelihood of pre-existing risk factors being misinterpreted as disease effects. We identified significant longitudinal effects when comparing the two groups, including (1) a greater reduction in grey matter thickness and tissue contrast in the orbitofrontal cortex and parahippocampal gyrus; (2) greater changes in markers of tissue damage in regions that are functionally connected to the primary olfactory cortex; and (3) a greater reduction in global brain size in the SARS-CoV-2 cases. The participants who were infected with SARS-CoV-2 also showed on average a greater cognitive decline between the two time points. Importantly, these imaging and cognitive longitudinal effects were still observed after excluding the 15 patients who had been hospitalised. These mainly limbic brain imaging results may be the in vivo hallmarks of a degenerative spread of the disease through olfactory pathways, of neuroinflammatory events, or of the loss of sensory input due to anosmia. Whether this deleterious effect can be partially reversed, or whether these effects will persist in the long term, remains to be investigated with additional follow-up.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-03", "2022-10", "2022-11", "2024-08"], "last_mentioned": "2024-08-31T07:55:38Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-023-06096-3", "slug": "27b5354100939031", "title": "Evidence for the utility of quantum computing before fault tolerance", "authors": ["Youngseok Kim", "Andrew Eddins", "Sajant Anand", "Ken Xuan Wei", "Ewout van den Berg", "Sami Rosenblatt", "Hasan Nayfeh", "Yantao Wu", "Michael Zaletel", "Kristan Temme", "Abhinav Kandala"], "year": 2023, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06096-3", "doi": "10.1038/s41586-023-06096-3", "topics": ["Physics", "Distributed systems", "Computer systems"], "description": "Abstract Quantum computing promises to offer substantial speed-ups over its classical counterpart for certain problems.", "description_method": "abstract-excerpt", "abstract": "Abstract Quantum computing promises to offer substantial speed-ups over its classical counterpart for certain problems. However, the greatest impediment to realizing its full potential is noise that is inherent to these systems. The widely accepted solution to this challenge is the implementation of fault-tolerant quantum circuits, which is out of reach for current processors. Here we report experiments on a noisy 127-qubit processor and demonstrate the measurement of accurate expectation values for circuit volumes at a scale beyond brute-force classical computation. We argue that this represents evidence for the utility of quantum computing in a pre-fault-tolerant era. These experimental results are enabled by advances in the coherence and calibration of a superconducting processor at this scale and the ability to characterize 1 and controllably manipulate noise across such a large device. We establish the accuracy of the measured expectation values by comparing them with the output of exactly verifiable circuits. In the regime of strong entanglement, the quantum computer provides correct results for which leading classical approximations such as pure-state-based 1D (matrix product states, MPS) and 2D (isometric tensor network states, isoTNS) tensor network methods 2,3 break down. These experiments demonstrate a foundational tool for the realization of near-term quantum applications 4,5 .", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2023-07", "2023-12", "2024-01", "2024-11"], "last_mentioned": "2024-11-09T22:40:57Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-023-06337-5", "slug": "ac71b9fa6dab4b17", "title": "An analog-AI chip for energy-efficient speech recognition and transcription", "authors": ["S. Ambrogio", "P. Narayanan", "A. Okazaki", "A. Fasoli", "C. Mackin", "K. Hosokawa", "A. Nomura", "T. Yasuda", "A. Chen", "A. Friz", "M. Ishii", "J. Luquin", "Y. Kohda", "N. Saulnier", "K. Brew", "S. Choi", "I. Ok", "T. Philip", "V. Chan", "C. Silvestre", "I. Ahsan", "V. Narayanan", "H. Tsai", "G. W. Burr"], "year": 2023, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06337-5", "doi": "10.1038/s41586-023-06337-5", "topics": ["Psychology", "Computer systems"], "description": "Abstract Models of artificial intelligence (AI) that have billions of parameters can achieve high accuracy across a range of tasks 1,2 , but they exacerbate the poor energy efficiency of conventional general-purpose processors, such as graphics processing units or central processing units.", "description_method": "abstract-excerpt", "abstract": "Abstract Models of artificial intelligence (AI) that have billions of parameters can achieve high accuracy across a range of tasks 1,2 , but they exacerbate the poor energy efficiency of conventional general-purpose processors, such as graphics processing units or central processing units. Analog in-memory computing (analog-AI) 3–7 can provide better energy efficiency by performing matrix–vector multiplications in parallel on ‘memory tiles’. However, analog-AI has yet to demonstrate software-equivalent (SW eq ) accuracy on models that require many such tiles and efficient communication of neural-network activations between the tiles. Here we present an analog-AI chip that combines 35 million phase-change memory devices across 34 tiles, massively parallel inter-tile communication and analog, low-power peripheral circuitry that can achieve up to 12.4 tera-operations per second per watt (TOPS/W) chip-sustained performance. We demonstrate fully end-to-end SW eq accuracy for a small keyword-spotting network and near-SW eq accuracy on the much larger MLPerf 8 recurrent neural-network transducer (RNNT), with more than 45 million weights mapped onto more than 140 million phase-change memory devices across five chips.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-08", "2023-10", "2024-02", "2024-04"], "last_mentioned": "2024-04-25T03:09:34Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-025-08800-x", "slug": "f7370a85e3b3e561", "title": "A natural experiment on the effect of herpes zoster vaccination on dementia", "authors": ["Markus Eyting", "Min Xie", "Felix Michalik", "Simon Heß", "Seunghun Chung", "Pascal Geldsetzer"], "year": 2025, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-025-08800-x", "doi": "10.1038/s41586-025-08800-x", "topics": ["Biology & medicine", "Statistics & probability", "Databases"], "description": "Abstract Neurotropic herpesviruses may be implicated in the development of dementia 1–5 .", "description_method": "abstract-excerpt", "abstract": "Abstract Neurotropic herpesviruses may be implicated in the development of dementia 1–5 . Moreover, vaccines may have important off-target immunological effects 6–9 . Here we aim to determine the effect of live-attenuated herpes zoster vaccination on the occurrence of dementia diagnoses. To provide causal as opposed to correlational evidence, we take advantage of the fact that, in Wales, eligibility for the zoster vaccine was determined on the basis of an individual’s exact date of birth. Those born before 2 September 1933 were ineligible and remained ineligible for life, whereas those born on or after 2 September 1933 were eligible for at least 1 year to receive the vaccine. Using large-scale electronic health record data, we first show that the percentage of adults who received the vaccine increased from 0.01% among patients who were merely 1 week too old to be eligible, to 47.2% among those who were just 1 week younger. Apart from this large difference in the probability of ever receiving the zoster vaccine, individuals born just 1 week before 2 September 1933 are unlikely to differ systematically from those born 1 week later. Using these comparison groups in a regression discontinuity design, we show that receiving the zoster vaccine reduced the probability of a new dementia diagnosis over a follow-up period of 7 years by 3.5 percentage points (95% confidence interval (CI) = 0.6–7.1, P = 0.019), corresponding to a 20.0% (95% CI = 6.5–33.4) relative reduction. This protective effect was stronger among women than men. We successfully confirm our findings in a different population (England and Wales’s combined population), with a different type of data (death certificates) and using an outcome (deaths with dementia as primary cause) that is closely related to dementia, but less reliant on a timely diagnosis of dementia by the healthcare system 10 . Through the use of a unique natural experiment, this study provides evidence of a dementia-preventing or dementia-delaying effect from zoster vaccination that is less vulnerable to confounding and bias than the existing associational evidence.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2025-04", "2025-10", "2025-12", "2026-02"], "last_mentioned": "2026-02-26T16:14:59Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-020-1083-1", "slug": "c49e303cedf99d12", "title": "Seasonal coronavirus protective immunity is short-lasting", "authors": ["Arthur W. D. Edridge", "Joanna Kaczorowska", "Alexis C. R. Hoste", "Margreet Bakker", "Michelle Klein", "Katherine Loens", "Maarten F. Jebbink", "Amy Matser", "Cormac M. Kinsella", "Paloma Rueda", "Margareta Ieven", "Herman Goossens", "Maria Prins", "Patricia Sastre", "Martin Deijs", "Lia van der Hoek"], "year": 2020, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-020-1083-1", "doi": "10.1038/s41591-020-1083-1", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-10", "2020-12", "2021-05", "2021-08"], "last_mentioned": "2021-08-08T17:49:42Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-021-01535-y", "slug": "48f15766fcb2ad81", "title": "The lesson of ivermectin: meta-analyses based on summary data alone are inherently unreliable", "authors": ["Jack M. Lawrence", "Gideon Meyerowitz-Katz", "James A. J. Heathers", "Nicholas J. L. Brown", "Kyle A. Sheldrick"], "year": 2021, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-021-01535-y", "doi": "10.1038/s41591-021-01535-y", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-10", "2022-01", "2022-09"], "last_mentioned": "2022-09-21T16:18:19Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-022-02051-3", "slug": "086c9d12d327133d", "title": "Acute and postacute sequelae associated with SARS-CoV-2 reinfection", "authors": ["Benjamin Bowe", "Yan Xie", "Ziyad Al-Aly"], "year": 2022, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-022-02051-3", "doi": "10.1038/s41591-022-02051-3", "topics": ["Statistics & probability", "Databases", "Biology & medicine"], "description": "Abstract First infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is associated with increased risk of acute and postacute death and sequelae in various organ systems.", "description_method": "abstract-excerpt", "abstract": "Abstract First infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is associated with increased risk of acute and postacute death and sequelae in various organ systems. Whether reinfection adds to risks incurred after first infection is unclear. Here we used the US Department of Veterans Affairs’ national healthcare database to build a cohort of individuals with one SARS-CoV-2 infection ( n = 443,588), reinfection (two or more infections, n = 40,947) and a noninfected control ( n = 5,334,729). We used inverse probability-weighted survival models to estimate risks and 6-month burdens of death, hospitalization and incident sequelae. Compared to no reinfection, reinfection contributed additional risks of death (hazard ratio (HR) = 2.17, 95% confidence intervals (CI) 1.93–2.45), hospitalization (HR = 3.32, 95% CI 3.13–3.51) and sequelae including pulmonary, cardiovascular, hematological, diabetes, gastrointestinal, kidney, mental health, musculoskeletal and neurological disorders. The risks were evident regardless of vaccination status. The risks were most pronounced in the acute phase but persisted in the postacute phase at 6 months. Compared to noninfected controls, cumulative risks and burdens of repeat infection increased according to the number of infections. Limitations included a cohort of mostly white males. The evidence shows that reinfection further increases risks of death, hospitalization and sequelae in multiple organ systems in the acute and postacute phase. Reducing overall burden of death and disease due to SARS-CoV-2 will require strategies for reinfection prevention.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-11", "2022-12", "2023-01"], "last_mentioned": "2023-01-14T07:04:45Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-017-13282-7", "slug": "00d1f83c69c6cecd", "title": "Psilocybin for treatment-resistant depression: fMRI-measured brain mechanisms", "authors": ["Robin L Carhart-Harris", "Leor Roseman", "Mark Bolstridge", "Lysia Demetriou", "J Nienke Pannekoek", "Matthew B Wall", "Mark Tanner", "Mendel Kaelen", "John McGonigle", "Kevin Murphy", "Robert Leech", "H Valerie Curran", "David J Nutt"], "year": 2017, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-017-13282-7", "doi": "10.1038/s41598-017-13282-7", "topics": ["Neuroscience", "Psychology", "Biology & medicine"], "description": "Abstract Psilocybin with psychological support is showing promise as a treatment model in psychiatry but its therapeutic mechanisms are poorly understood.", "description_method": "abstract-excerpt", "abstract": "Abstract Psilocybin with psychological support is showing promise as a treatment model in psychiatry but its therapeutic mechanisms are poorly understood. Here, cerebral blood flow (CBF) and blood oxygen-level dependent (BOLD) resting-state functional connectivity (RSFC) were measured with functional magnetic resonance imaging (fMRI) before and after treatment with psilocybin (serotonin agonist) for treatment-resistant depression (TRD). Quality pre and post treatment fMRI data were collected from 16 of 19 patients. Decreased depressive symptoms were observed in all 19 patients at 1-week post-treatment and 47% met criteria for response at 5 weeks. Whole-brain analyses revealed post-treatment decreases in CBF in the temporal cortex, including the amygdala. Decreased amygdala CBF correlated with reduced depressive symptoms. Focusing on a priori selected circuitry for RSFC analyses, increased RSFC was observed within the default-mode network (DMN) post-treatment. Increased ventromedial prefrontal cortex-bilateral inferior lateral parietal cortex RSFC was predictive of treatment response at 5-weeks, as was decreased parahippocampal-prefrontal cortex RSFC. These data fill an important knowledge gap regarding the post-treatment brain effects of psilocybin, and are the first in depressed patients. The post-treatment brain changes are different to previously observed acute effects of psilocybin and other ‘psychedelics’ yet were related to clinical outcomes. A ‘reset’ therapeutic mechanism is proposed.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2017-10", "2017-11", "2018-02", "2018-05"], "last_mentioned": "2018-05-17T13:26:26Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-018-21284-2", "slug": "5d49ccbaf4125d84", "title": "Impact of genetically engineered maize on agronomic, environmental and toxicological traits: a meta-analysis of 21 years of field data", "authors": ["Elisa Pellegrino", "Stefano Bedini", "Marco Nuti", "Laura Ercoli"], "year": 2018, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-018-21284-2", "doi": "10.1038/s41598-018-21284-2", "topics": ["Science & research"], "description": "Abstract Despite the extensive cultivation of genetically engineered (GE) maize and considerable number of scientific reports on its agro-environmental impact, the risks and benefits of GE maize are still being debated and concerns about safety remain.", "description_method": "abstract-excerpt", "abstract": "Abstract Despite the extensive cultivation of genetically engineered (GE) maize and considerable number of scientific reports on its agro-environmental impact, the risks and benefits of GE maize are still being debated and concerns about safety remain. This meta-analysis aimed at increasing knowledge on agronomic, environmental and toxicological traits of GE maize by analyzing the peer-reviewed literature (from 1996 to 2016) on yield, grain quality, non-target organisms (NTOs), target organisms (TOs) and soil biomass decomposition. Results provided strong evidence that GE maize performed better than its near isogenic line: grain yield was 5.6 to 24.5% higher with lower concentrations of mycotoxins (−28.8%), fumonisin (−30.6%) and thricotecens (−36.5%). The NTOs analyzed were not affected by GE maize, except for Braconidae, represented by a parasitoid of European corn borer, the target of Lepidoptera active Bt maize. Biogeochemical cycle parameters such as lignin content in stalks and leaves did not vary, whereas biomass decomposition was higher in GE maize. The results support the cultivation of GE maize, mainly due to enhanced grain quality and reduction of human exposure to mycotoxins. Furthermore, the reduction of the parasitoid of the target and the lack of consistent effects on other NTOs are confirmed.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2018-02", "2020-08", "2021-02"], "last_mentioned": "2021-02-17T21:43:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-018-32194-8", "slug": "b9a1e6f63c0372b9", "title": "Extracts of Polypore Mushroom Mycelia Reduce Viruses in Honey Bees", "authors": ["Paul E. Stamets", "Nicholas L. Naeger", "Jay D. Evans", "Jennifer O. Han", "Brandon K. Hopkins", "Dawn Lopez", "Henry M. Moershel", "Regan Nally", "David Sumerlin", "Alex W. Taylor", "Lori M. Carris", "Walter S. Sheppard"], "year": 2018, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-018-32194-8", "doi": "10.1038/s41598-018-32194-8", "topics": ["Other research"], "description": "AbstractWaves of highly infectious viruses sweeping through global honey bee populations have contributed to recent declines in honey bee health.", "description_method": "abstract-excerpt", "abstract": "AbstractWaves of highly infectious viruses sweeping through global honey bee populations have contributed to recent declines in honey bee health. Bees have been observed foraging on mushroom mycelium, suggesting that they may be deriving medicinal or nutritional value from fungi. Fungi are known to produce a wide array of chemicals with antimicrobial activity, including compounds active against bacteria, other fungi, or viruses. We tested extracts from the mycelium of multiple polypore fungal species known to have antiviral properties. Extracts from amadou (Fomes) and reishi (Ganoderma) fungi reduced the levels of honey bee deformed wing virus (DWV) and Lake Sinai virus (LSV) in a dose-dependent manner. In field trials, colonies fed Ganoderma resinaceum extract exhibited a 79-fold reduction in DWV and a 45,000-fold reduction in LSV compared to control colonies. These findings indicate honey bees may gain health benefits from fungi and their antimicrobial compounds.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2019-03", "2019-07", "2020-01", "2020-02"], "last_mentioned": "2020-02-05T14:09:25Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-019-50807-8", "slug": "99282be3068f428e", "title": "Oral selective serotonin reuptake inhibitors activate vagus nerve dependent gut-brain signalling", "authors": ["Karen-Anne McVey Neufeld", "John Bienenstock", "Aadil Bharwani", "Kevin Champagne-Jorgensen", "YuKang Mao", "Christine West", "Yunpeng Liu", "Michael G. Surette", "Wolfgang Kunze", "Paul Forsythe"], "year": 2019, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-019-50807-8", "doi": "10.1038/s41598-019-50807-8", "topics": ["Neuroscience", "Psychology"], "description": "Abstract The vagus nerve can transmit signals to the brain resulting in a reduction in depressive behavior as evidenced by the long-term beneficial effects of electrical stimulation of the vagus in patients with intractable depression.", "description_method": "abstract-excerpt", "abstract": "Abstract The vagus nerve can transmit signals to the brain resulting in a reduction in depressive behavior as evidenced by the long-term beneficial effects of electrical stimulation of the vagus in patients with intractable depression. The vagus is the major neural connection between gut and brain, and we have previously shown that ingestion of beneficial bacteria modulates behaviour and brain neurochemistry via this pathway. Given the high levels of serotonin in the gut, we considered if gut-brain signaling, and specifically the vagal pathway, might contribute to the therapeutic effect of oral selective serotonin reuptake inhibitors (SSRI). Mesenteric nerve recordings were conducted in mice after treatment with SSRI to ascertain if this class of drugs resulted in increased vagal excitability. Patch clamp recordings of enteric neurons were carried out to measure activity of primary afferent neurons in the gut in response to SSRI and to assess the importance of gut epithelium in transducing signal. The tail suspension test (TST) was used following 14d feeding of SSRI in vagotomised and surgical sham mice to measure depressive-like behaviour. Brain mRNA expression was examined via PCR and the intestinal microbiome was assessed. Mesenteric nerve recordings in BALB/c mice demonstrated that oral treatment with SSRI leads to a significant increase in vagal activity. This effect was not observed in mice treated with a representative noradrenaline-dopamine reuptake inhibitor. It is known that signals from the gut can be transmitted to the vagus via the enteric nervous system. Exposure of the gut to SSRI increased the excitability of intrinsic primary afferent neurons in the myenteric plexus, through an intestinal epithelium dependent mechanism, and alpha-diversity of gut microbiota was altered. Critically, blocking vagal signaling from gut to brain, via subdiaphragmatic vagotomy, abolished the antidepressive effects of oral SSRI treatment as determined by the tail suspension test. This work suggests that vagus nerve dependent gut-brain signaling contributes to the effects of oral SSRI and further, highlights the potential for pharmacological approaches to treatment of mood disorders that focus on vagal stimulation and may not even require therapeutic agents to enter the circulation.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-07", "2023-04", "2023-10", "2024-07"], "last_mentioned": "2024-07-05T18:39:05Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-020-58060-0", "slug": "dc31790657d32933", "title": "Emergence of life in an inflationary universe", "authors": ["Tomonori Totani"], "year": 2020, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-020-58060-0", "doi": "10.1038/s41598-020-58060-0", "topics": ["Biology & medicine", "Statistics & probability"], "description": "Abstract Abiotic emergence of ordered information stored in the form of RNA is an important unresolved problem concerning the origin of life.", "description_method": "abstract-excerpt", "abstract": "Abstract Abiotic emergence of ordered information stored in the form of RNA is an important unresolved problem concerning the origin of life. A polymer longer than 40–100 nucleotides is necessary to expect a self-replicating activity, but the formation of such a long polymer having a correct nucleotide sequence by random reactions seems statistically unlikely. However, our universe, created by a single inflation event, likely includes more than 10 100 Sun-like stars. If life can emerge at least once in such a large volume, it is not in contradiction with our observations of life on Earth, even if the expected number of abiogenesis events is negligibly small within the observable universe that contains only 10 22 stars. Here, a quantitative relation is derived between the minimum RNA length l min required to be the first biological polymer, and the universe size necessary to expect the formation of such a long and active RNA by randomly adding monomers. It is then shown that an active RNA can indeed be produced somewhere in an inflationary universe, giving a solution to the abiotic polymerization problem. On the other hand, l min must be shorter than ~20 nucleotides for the abiogenesis probability close to unity on a terrestrial planet, but a self-replicating activity is not expected for such a short RNA. Therefore, if extraterrestrial organisms of a different origin from those on Earth are discovered in the future, it would imply an unknown mechanism at work to polymerize nucleotides much faster than random statistical processes.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-05", "2022-07", "2023-12"], "last_mentioned": "2023-12-17T15:26:00Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-020-60867-w", "slug": "462de6d92e701e5b", "title": "Evidence of Cosmic Impact at Abu Hureyra, Syria at the Younger Dryas Onset (~12.8 ka): High-temperature melting at >2200 °C", "authors": ["Andrew M. T. Moore", "James P. Kennett", "William M. Napier", "Ted E. Bunch", "James C. Weaver", "Malcolm LeCompte", "A. Victor Adedeji", "Paul Hackley", "Gunther Kletetschka", "Robert E. Hermes", "James H. Wittke", "Joshua J. Razink", "Michael W. Gaultois", "Allen West"], "year": 2020, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-020-60867-w", "doi": "10.1038/s41598-020-60867-w", "topics": ["Statistics & probability", "Economics"], "description": "Abstract At Abu Hureyra (AH), Syria, the 12,800-year-old Younger Dryas boundary layer (YDB) contains peak abundances in meltglass, nanodiamonds, microspherules, and charcoal.", "description_method": "abstract-excerpt", "abstract": "Abstract At Abu Hureyra (AH), Syria, the 12,800-year-old Younger Dryas boundary layer (YDB) contains peak abundances in meltglass, nanodiamonds, microspherules, and charcoal. AH meltglass comprises 1.6 wt.% of bulk sediment, and crossed polarizers indicate that the meltglass is isotropic. High YDB concentrations of iridium, platinum, nickel, and cobalt suggest mixing of melted local sediment with small quantities of meteoritic material. Approximately 40% of AH glass display carbon-infused, siliceous plant imprints that laboratory experiments show formed at a minimum of 1200°–1300 °C; however, reflectance-inferred temperatures for the encapsulated carbon were lower by up to 1000 °C. Alternately, melted grains of quartz, chromferide, and magnetite in AH glass suggest exposure to minimum temperatures of 1720 °C ranging to >2200 °C. This argues against formation of AH meltglass in thatched hut fires at 1100°–1200 °C, and low values of remanent magnetism indicate the meltglass was not created by lightning. Low meltglass water content (0.02–0.05% H 2 O) is consistent with a formation process similar to that of tektites and inconsistent with volcanism and anthropogenesis. The wide range of evidence supports the hypothesis that a cosmic event occurred at Abu Hureyra ~12,800 years ago, coeval with impacts that deposited high-temperature meltglass, melted microspherules, and/or platinum at other YDB sites on four continents.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-09", "2022-01", "2025-07"], "last_mentioned": "2025-07-02T22:10:39Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41893-020-00678-z", "slug": "584d85631b70e126", "title": "Impacts of transportation network companies on urban mobility", "authors": ["Mi Diao", "Hui Kong", "Jinhua Zhao"], "year": 2021, "venue": "Nature Sustainability", "url": "https://doi.org/10.1038/s41893-020-00678-z", "doi": "10.1038/s41893-020-00678-z", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2022-12", "2024-03", "2024-04", "2026-04"], "last_mentioned": "2026-04-26T22:54:30Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s42003-020-01577-x", "slug": "52be17c134c0bc96", "title": "Identification of 3-chymotrypsin like protease (3CLPro) inhibitors as potential anti-SARS-CoV-2 agents", "authors": ["Vicky Mody", "Joanna Ho", "Savannah Wills", "Ahmed Mawri", "Latasha Lawson", "Maximilian C. C. J. C. Ebert", "Guillaume M. Fortin", "Srujana Rayalam", "Shashidharamurthy Taval"], "year": 2021, "venue": "Communications Biology", "url": "https://doi.org/10.1038/s42003-020-01577-x", "doi": "10.1038/s42003-020-01577-x", "topics": ["Biology & medicine", "Distributed systems"], "description": "Abstract Emerging outbreak of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection is a major threat to public health.", "description_method": "abstract-excerpt", "abstract": "Abstract Emerging outbreak of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection is a major threat to public health. The morbidity is increasing due to lack of SARS-CoV-2 specific drugs. Herein, we have identified potential drugs that target the 3-chymotrypsin like protease (3CLpro), the main protease that is pivotal for the replication of SARS-CoV-2. Computational molecular modeling was used to screen 3987 FDA approved drugs, and 47 drugs were selected to study their inhibitory effects on SARS-CoV-2 specific 3CLpro enzyme in vitro. Our results indicate that boceprevir, ombitasvir, paritaprevir, tipranavir, ivermectin, and micafungin exhibited inhibitory effect towards 3CLpro enzymatic activity. The 100 ns molecular dynamics simulation studies showed that ivermectin may require homodimeric form of 3CLpro enzyme for its inhibitory activity. In summary, these molecules could be useful to develop highly specific therapeutically viable drugs to inhibit the SARS-CoV-2 replication either alone or in combination with drugs specific for other SARS-CoV-2 viral targets.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-04", "2021-09", "2021-11", "2022-03"], "last_mentioned": "2022-03-18T12:27:44Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s43247-024-01442-3", "slug": "c067d42ce691068f", "title": "Abrupt reduction in shipping emission as an inadvertent geoengineering termination shock produces substantial radiative warming", "authors": ["Tianle Yuan", "Hua Song", "Lazaros Oreopoulos", "Robert Wood", "Huisheng Bian", "Katherine Breen", "Mian Chin", "Hongbin Yu", "Donifan Barahona", "Kerry Meyer", "Steven Platnick"], "year": 2024, "venue": "Communications Earth &amp; Environment", "url": "https://doi.org/10.1038/s43247-024-01442-3", "doi": "10.1038/s43247-024-01442-3", "topics": ["Other research"], "description": "Abstract Human activities affect the Earth’s climate through modifying the composition of the atmosphere, which then creates radiative forcing that drives climate change.", "description_method": "abstract-excerpt", "abstract": "Abstract Human activities affect the Earth’s climate through modifying the composition of the atmosphere, which then creates radiative forcing that drives climate change. The warming effect of anthropogenic greenhouse gases has been partially balanced by the cooling effect of anthropogenic aerosols. In 2020, fuel regulations abruptly reduced the emission of sulfur dioxide from international shipping by about 80% and created an inadvertent geoengineering termination shock with global impact. Here we estimate the regulation leads to a radiative forcing of $$+0.2\\pm 0.11$$ + 0.2 ± 0.11 Wm −2 averaged over the global ocean. The amount of radiative forcing could lead to a doubling (or more) of the warming rate in the 2020 s compared with the rate since 1980 with strong spatiotemporal heterogeneity. The warming effect is consistent with the recent observed strong warming in 2023 and expected to make the 2020 s anomalously warm. The forcing is equivalent in magnitude to 80% of the measured increase in planetary heat uptake since 2020. The radiative forcing also has strong hemispheric contrast, which has important implications for precipitation pattern changes. Our result suggests marine cloud brightening may be a viable geoengineering method in temporarily cooling the climate that has its unique challenges due to inherent spatiotemporal heterogeneity.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-06", "2024-09", "2025-02", "2025-11"], "last_mentioned": "2025-11-09T03:24:42Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/srep07861", "slug": "3b974d73c1bf38fb", "title": "Humans perceive flicker artifacts at 500 Hz", "authors": ["James Davis", "Yi-Hsuan Hsieh", "Hung-Chi Lee"], "year": 2015, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/srep07861", "doi": "10.1038/srep07861", "topics": ["Physics"], "description": "AbstractHumans perceive a stable average intensity image without flicker artifacts when a television or monitor updates at a sufficiently fast rate.", "description_method": "abstract-excerpt", "abstract": "AbstractHumans perceive a stable average intensity image without flicker artifacts when a television or monitor updates at a sufficiently fast rate. This rate, known as the critical flicker fusion rate, has been studied for both spatially uniform lights and spatio-temporal displays. These studies have included both stabilized and unstablized retinal images and report the maximum observable rate as 50–90 Hz. A separate line of research has reported that fast eye movements known as saccades allow simple modulated LEDs to be observed at very high rates. Here we show that humans perceive visual flicker artifacts at rates over 500 Hz when a display includes high frequency spatial edges. This rate is many times higher than previously reported. As a result, modern display designs which use complex spatio-temporal coding need to update much faster than conventional TVs, which traditionally presented a simple sequence of natural images.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2022-01", "2024-01", "2024-07", "2025-06"], "last_mentioned": "2025-06-19T01:58:48Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.aaw0341", "slug": "7aa6786863abb064", "title": "Extreme events reveal an alimentary limit on sustained maximal human energy expenditure", "authors": ["Caitlin Thurber", "Lara R. Dugas", "Cara Ocobock", "Bryce Carlson", "John R. Speakman", "Herman Pontzer"], "year": 2019, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.aaw0341", "doi": "10.1126/sciadv.aaw0341", "topics": ["Other research"], "description": "Maximum sustained energy expenditure in humans is limited by energy intake and strongly tied to event duration.", "description_method": "abstract-excerpt", "abstract": "Maximum sustained energy expenditure in humans is limited by energy intake and strongly tied to event duration.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-02", "2023-01", "2023-10"], "last_mentioned": "2023-10-14T14:34:01Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.aaz5803", "slug": "a78e94b40835b2f7", "title": "More than 1000 rivers account for 80% of global riverine plastic emissions into the ocean", "authors": ["Lourens J. J. Meijer", "Tim van Emmerik", "Ruud van der Ent", "Christian Schmidt", "Laurent Lebreton"], "year": 2021, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.aaz5803", "doi": "10.1126/sciadv.aaz5803", "topics": ["Other research"], "description": "More than 1000 rivers account for most plastic emissions, ranging from small urban drains to large rivers.", "description_method": "abstract-excerpt", "abstract": "More than 1000 rivers account for most plastic emissions, ranging from small urban drains to large rivers.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2021-09", "2022-11", "2024-05", "2024-09"], "last_mentioned": "2024-09-09T14:43:01Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.add4165", "slug": "f99823219f3a52a2", "title": "Not so smart? “Smart” drugs increase the level but decrease the quality of cognitive effort", "authors": ["Elizabeth Bowman", "David Coghill", "Carsten Murawski", "Peter Bossaerts"], "year": 2023, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.add4165", "doi": "10.1126/sciadv.add4165", "topics": ["Neuroscience", "Algorithms"], "description": "The efficacy of pharmaceutical cognitive enhancers in everyday complex tasks remains to be established.", "description_method": "abstract-excerpt", "abstract": "The efficacy of pharmaceutical cognitive enhancers in everyday complex tasks remains to be established. Using the knapsack optimization problem as a stylized representation of difficulty in tasks encountered in daily life, we discover that methylphenidate, dextroamphetamine, and modafinil cause knapsack value attained in the task to diminish significantly compared to placebo, even if the chance of finding the optimal solution (~50%) is not reduced significantly. Effort (decision time and number of steps taken to find a solution) increases significantly, but productivity (quality of effort) decreases significantly. At the same time, productivity differences across participants decrease, even reverse, to the extent that above-average performers end up below average and vice versa. The latter can be attributed to increased randomness of solution strategies. Our findings suggest that “smart drugs” increase motivation, but a reduction in quality of effort, crucial to solve complex problems, annuls this effect.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2023-06", "2023-11", "2024-06"], "last_mentioned": "2024-06-17T07:25:18Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.aaw1313", "slug": "20dc485eacece119", "title": "Decline of the North American avifauna", "authors": ["Kenneth V. Rosenberg", "Adriaan M. Dokter", "Peter J. Blancher", "John R. Sauer", "Adam C. Smith", "Paul A. Smith", "Jessica C. Stanton", "Arvind Panjabi", "Laura Helft", "Michael Parr", "Peter P. Marra"], "year": 2019, "venue": "Science", "url": "https://doi.org/10.1126/science.aaw1313", "doi": "10.1126/science.aaw1313", "topics": ["Other research"], "description": "Staggering decline of bird populations Because birds are conspicuous and easy to identify and count, reliable records of their occurrence have been gathered over many decades in many parts of the world.", "description_method": "abstract-excerpt", "abstract": "Staggering decline of bird populations Because birds are conspicuous and easy to identify and count, reliable records of their occurrence have been gathered over many decades in many parts of the world. Drawing on such data for North America, Rosenberg et al. report wide-spread population declines of birds over the past half-century, resulting in the cumulative loss of billions of breeding individuals across a wide range of species and habitats. They show that declines are not restricted to rare and threatened species—those once considered common and wide-spread are also diminished. These results have major implications for ecosystem integrity, the conservation of wildlife more broadly, and policies associated with the protection of birds and native ecosystems on which they depend. Science , this issue p. 120", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-08", "2024-01", "2026-04", "2026-05"], "last_mentioned": "2026-05-14T14:48:33Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abd9149", "slug": "16eb5ba50148b253", "title": "Airborne transmission of respiratory viruses", "authors": ["Chia C. Wang", "Kimberly A. Prather", "Josué Sznitman", "Jose L. Jimenez", "Seema S. Lakdawala", "Zeynep Tufekci", "Linsey C. Marr"], "year": 2021, "venue": "Science", "url": "https://doi.org/10.1126/science.abd9149", "doi": "10.1126/science.abd9149", "topics": ["Other research"], "description": "Mechanisms of airborne transmission The COVID-19 pandemic has highlighted controversies and unknowns about how respiratory pathogens spread between hosts.", "description_method": "abstract-excerpt", "abstract": "Mechanisms of airborne transmission The COVID-19 pandemic has highlighted controversies and unknowns about how respiratory pathogens spread between hosts. Traditionally, it was thought that respiratory pathogens spread between people through large droplets produced in coughs and through contact with contaminated surfaces (fomites). However, several respiratory pathogens are known to spread through small respiratory aerosols, which can float and travel in air flows, infecting people who inhale them at short and long distances from the infected person. Wang et al . review recent advances in understanding airborne transmission gained from studying the spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections and other respiratory pathogens. The authors suggest that airborne transmission may be the dominant form of transmission for several respiratory pathogens, including SARS-CoV-2, and that further understanding of the mechanisms underlying infection from the airborne route will better inform mitigation measures. —GKA", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2022-01", "2022-04", "2023-01", "2023-06"], "last_mentioned": "2023-06-02T18:37:14Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.ads9158", "slug": "34070f46657ccd4f", "title": "Confronting risks of mirror life", "authors": ["Katarzyna P. Adamala", "Deepa Agashe", "Yasmine Belkaid", "Daniela Matias de C. Bittencourt", "Yizhi Cai", "Matthew W. Chang", "Irene A. Chen", "George M. Church", "Vaughn S. Cooper", "Mark M. Davis", "Neal K. Devaraj", "Drew Endy", "Kevin M. Esvelt", "John I. Glass", "Timothy W. Hand", "Thomas V. Inglesby", "Farren J. Isaacs", "Wilmot G. James", "Jonathan D. G. Jones", "Michael S. Kay", "Richard E. Lenski", "Chenli Liu", "Ruslan Medzhitov", "Matthew L. Nicotra", "Sebastian B. Oehm", "Jaspreet Pannu", "David A. Relman", "Petra Schwille", "James A. Smith", "Hiroaki Suga", "Jack W. Szostak", "Nicholas J. Talbot", "James M. Tiedje", "J. Craig Venter", "Gregory Winter", "Weiwen Zhang", "Xinguang Zhu", "Maria T. Zuber"], "year": 2024, "venue": "Science", "url": "https://doi.org/10.1126/science.ads9158", "doi": "10.1126/science.ads9158", "topics": ["Other research"], "description": "Broad discussion is needed to chart a path forward", "description_method": "abstract-excerpt", "abstract": "Broad discussion is needed to chart a path forward", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2025-10", "2026-06", "2026-08"], "last_mentioned": "2026-08-16T15:24:21Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/121132.121151", "slug": "98c4f454de6fc8bd", "title": "Scheduler activations", "authors": ["Thomas E. Anderson", "Brian N. Bershad", "Edward D. Lazowska", "Henry M. Levy"], "year": 1991, "venue": "Proceedings of the thirteenth ACM symposium on Operating systems principles", "url": "https://doi.org/10.1145/121132.121151", "doi": "10.1145/121132.121151", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-02", "2025-10", "2026-05"], "last_mentioned": "2026-05-05T15:11:38Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3276490", "slug": "10e805d27a31f48f", "title": "Julia: dynamism and performance reconciled by design", "authors": ["Jeff Bezanson", "Jiahao Chen", "Benjamin Chung", "Stefan Karpinski", "Viral B. Shah", "Jan Vitek", "Lionel Zoubritzky"], "year": 2018, "venue": "Proceedings of the ACM on Programming Languages", "url": "https://doi.org/10.1145/3276490", "doi": "10.1145/3276490", "topics": ["Programming languages", "Science & research", "Human–computer interaction"], "description": "Julia is a programming language for the scientific community that combines features of productivity languages, such as Python or MATLAB, with characteristics of performance-oriented languages, such as C++ or Fortran.", "description_method": "abstract-excerpt", "abstract": "Julia is a programming language for the scientific community that combines features of productivity languages, such as Python or MATLAB, with characteristics of performance-oriented languages, such as C++ or Fortran. Julia's productivity features include: dynamic typing, automatic memory management, rich type annotations, and multiple dispatch. At the same time, Julia allows programmers to control memory layout and leverages a specializing just-in-time compiler to eliminate much of the overhead of those features. This paper details the design choices made by the creators of Julia and reflects on the implications of those choices for performance and usability.", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2020-10", "2021-02", "2021-04", "2022-04"], "last_mentioned": "2022-04-26T23:49:49Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3317550.3321435", "slug": "5e1d0f9f4501c6a4", "title": "A fork() in the road", "authors": ["Andrew Baumann", "Jonathan Appavoo", "Orran Krieger", "Timothy Roscoe"], "year": 2019, "venue": "Proceedings of the Workshop on Hot Topics in Operating Systems", "url": "https://doi.org/10.1145/3317550.3321435", "doi": "10.1145/3317550.3321435", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-11", "2022-03", "2023-03", "2023-12"], "last_mentioned": "2023-12-25T10:16:30Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3445814.3446723", "slug": "a7873a223a2272c4", "title": "Warehouse-scale video acceleration: co-design and deployment in the wild", "authors": ["Parthasarathy Ranganathan", "Daniel Stodolsky", "Jeff Calow", "Jeremy Dorfman", "Marisabel Guevara", "Clinton Wills Smullen IV", "Aki Kuusela", "Raghu Balasubramanian", "Sandeep Bhatia", "Prakash Chauhan", "Anna Cheung", "In Suk Chong", "Niranjani Dasharathi", "Jia Feng", "Brian Fosco", "Samuel Foss", "Ben Gelb", "Sara J. Gwin", "Yoshiaki Hase", "Da-ke He", "C. Richard Ho", "Roy W. Huffman Jr.", "Elisha Indupalli", "Indira Jayaram", "Poonacha Kongetira", "Cho Mon Kyaw", "Aaron Laursen", "Yuan Li", "Fong Lou", "Kyle A. Lucke", "JP Maaninen", "Ramon Macias", "Maire Mahony", "David Alexander Munday", "Srikanth Muroor", "Narayana Penukonda", "Eric Perkins-Argueta", "Devin Persaud", "Alex Ramirez", "Ville-Mikko Rautio", "Yolanda Ripley", "Amir Salek", "Sathish Sekar", "Sergey N. Sokolov", "Rob Springer", "Don Stark", "Mercedes Tan", "Mark S. Wachsler", "Andrew C. Walton", "David A. Wickeraad", "Alvin Wijaya", "Hon Kwan Wu"], "year": 2021, "venue": "Proceedings of the 26th ACM International Conference on Architectural Support for Programming Languages and Operating Systems", "url": "https://doi.org/10.1145/3445814.3446723", "doi": "10.1145/3445814.3446723", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2021-04", "2021-05", "2022-02", "2022-08"], "last_mentioned": "2022-08-27T14:46:05Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3519939.3523440", "slug": "2fdc2b0ee8a338af", "title": "Low-latency, high-throughput garbage collection", "authors": ["Wenyu Zhao", "Stephen M. Blackburn", "Kathryn S. McKinley"], "year": 2022, "venue": "Proceedings of the 43rd ACM SIGPLAN International Conference on Programming Language Design and Implementation", "url": "https://doi.org/10.1145/3519939.3523440", "doi": "10.1145/3519939.3523440", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2023-03", "2024-03", "2025-07"], "last_mentioned": "2025-07-10T11:39:09Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/363831.363839", "slug": "75f4d0253224ccdc", "title": "Proposed revised American standard code for information interchange", "authors": [], "year": 1965, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/363831.363839", "doi": "10.1145/363831.363839", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2024-03", "2024-05", "2025-01"], "last_mentioned": "2025-01-22T02:30:15Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/365153.365168", "slug": "217983091ccc12a4", "title": "ELIZA—A Computer Program For the Study of Natural Language Communication Between Man and Machine", "authors": ["Joseph Weizenbaum"], "year": 1966, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/365153.365168", "doi": "10.1145/365153.365168", "topics": ["Psychology"], "description": "ELIZA is a program operating within the MAC time-sharing system at MIT which makes certain kinds of natural language conversation between man and computer possible.", "description_method": "abstract-excerpt", "abstract": "ELIZA is a program operating within the MAC time-sharing system at MIT which makes certain kinds of natural language conversation between man and computer possible. Input sentences are analyzed on the basis of decomposition rules which are triggered by key words appearing in the input text. Responses are generated by reassembly rules associated with selected decomposition rules. The fundamental technical problems with which ELIZA is concerned are: (1) the identification of key words, (2) the discovery of minimal context, (3) the choice of appropriate transformations, (4) generation of responses in the absence of key words, and (5) the provision of an editing capability for ELIZA “scripts”. A discussion of some psychological issues relevant to the ELIZA approach as well as of future developments concludes the paper.", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2024-01", "2024-10", "2026-04"], "last_mentioned": "2026-04-21T00:16:40Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/367177.367199", "slug": "d31779e054b64763", "title": "Recursive functions of symbolic expressions and their computation by machine, Part I", "authors": ["John McCarthy"], "year": 1960, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/367177.367199", "doi": "10.1145/367177.367199", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 4, "months": ["2024-04", "2024-10", "2025-08", "2025-10"], "last_mentioned": "2025-10-14T14:43:26Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/512760.512762", "slug": "2a5fe05975f3ad25", "title": "The Smalltalk-76 programming system design and implementation", "authors": ["Daniel H. H. Ingalls"], "year": 1978, "venue": "Proceedings of the 5th ACM SIGACT-SIGPLAN symposium on Principles of programming languages  - POPL '78", "url": "https://doi.org/10.1145/512760.512762", "doi": "10.1145/512760.512762", "topics": ["Programming languages"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 4, "user_count": 3, "months": ["2020-05", "2020-10", "2025-02", "2025-07"], "last_mentioned": "2025-07-30T07:40:27Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0711.0770", "slug": "13a6b01cc12f2233", "title": "An Exceptionally Simple Theory of Everything", "authors": ["A. Garrett Lisi"], "year": 2007, "venue": "arXiv", "url": "https://arxiv.org/abs/0711.0770", "doi": null, "topics": ["Mathematics", "Physics"], "description": "All fields of the standard model and gravity are unified as an E8 principal bundle connection.", "description_method": "abstract-excerpt", "abstract": "All fields of the standard model and gravity are unified as an E8 principal bundle connection. A non-compact real form of the E8 Lie algebra has G2 and F4 subalgebras which break down to strong su(3), electroweak su(2) x u(1), gravitational so(3,1), the frame-Higgs, and three generations of fermions related by triality. The interactions and dynamics of these 1-form and Grassmann valued parts of an E8 superconnection are described by the curvature and action over a four dimensional base manifold.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2008-02", "2016-08", "2018-07"], "last_mentioned": "2018-07-20T17:31:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0808.3283", "slug": "26be1e08fdb7b36b", "title": "Evidence for Correlations Between Nuclear Decay Rates and Earth-Sun Distance", "authors": ["Jere H. Jenkins", "Ephraim Fischbach", "John B. Buncher", "John T. Gruenwald", "Dennis E. Krause", "Joshua J. Mattes"], "year": 2008, "venue": "arXiv", "url": "https://arxiv.org/abs/0808.3283", "doi": "10.1016/j.astropartphys.2009.05.004", "topics": ["Economics"], "description": "Unexplained periodic fluctuations in the decay rates of Si-32 and Ra-226 have been reported by groups at Brookhaven National Laboratory (Si-32), and at the Physikalisch-Technische-Bundesandstalt in Germany (Ra-226).", "description_method": "abstract-excerpt", "abstract": "Unexplained periodic fluctuations in the decay rates of Si-32 and Ra-226 have been reported by groups at Brookhaven National Laboratory (Si-32), and at the Physikalisch-Technische-Bundesandstalt in Germany (Ra-226). We show from an analysis of the raw data in these experiments that the observed fluctuations are strongly correlated in time, not only with each other, but also with the distance between the Earth and the Sun. Some implications of these results are also discussed, including the suggestion that discrepancies in published half-life determinations for these and other nuclides may be attributable in part to differences in solar activity during the course of the various experiments, or to seasonal variations in fundamental constants.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2008-08", "2010-08", "2011-06"], "last_mentioned": "2011-06-08T19:03:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:0909.0801", "slug": "f1b8bda5d2108715", "title": "A Monte Carlo AIXI Approximation", "authors": ["Joel Veness", "Kee Siong Ng", "Marcus Hutter", "William Uther", "David Silver"], "year": 2009, "venue": "arXiv", "url": "https://arxiv.org/abs/0909.0801", "doi": null, "topics": ["Statistics & probability", "Algorithms", "Machine learning"], "description": "This paper introduces a principled approach for the design of a scalable general reinforcement learning agent.", "description_method": "abstract-excerpt", "abstract": "This paper introduces a principled approach for the design of a scalable general reinforcement learning agent. Our approach is based on a direct approximation of AIXI, a Bayesian optimality notion for general reinforcement learning agents. Previously, it has been unclear whether the theory of AIXI could motivate the design of practical algorithms. We answer this hitherto open question in the affirmative, by providing the first computationally feasible approximation to the AIXI agent. To develop our approximation, we introduce a new Monte-Carlo Tree Search algorithm along with an agent-specific extension to the Context Tree Weighting algorithm. Empirically, we present a set of encouraging results on a variety of stochastic and partially observable domains. We conclude by proposing a number of directions for future research.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2010-03", "2018-11", "2024-01"], "last_mentioned": "2024-01-31T02:39:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1002.2858", "slug": "a66bdf61e3b6cb0b", "title": "PageRank: Standing on the shoulders of giants", "authors": ["Massimo Franceschet"], "year": 2010, "venue": "arXiv", "url": "https://arxiv.org/abs/1002.2858", "doi": null, "topics": ["Other research"], "description": "PageRank is a Web page ranking technique that has been a fundamental ingredient in the development and success of the Google search engine.", "description_method": "abstract-excerpt", "abstract": "PageRank is a Web page ranking technique that has been a fundamental ingredient in the development and success of the Google search engine. The method is still one of the many signals that Google uses to determine which pages are most important. The main idea behind PageRank is to determine the importance of a Web page in terms of the importance assigned to the pages hyperlinking to it. In fact, this thesis is not new, and has been previously successfully exploited in different contexts. We review the PageRank method and link it to some renowned previous techniques that we have found in the fields of Web information retrieval, bibliometrics, sociometry, and econometrics.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2014-05", "2022-06", "2023-06"], "last_mentioned": "2023-06-20T20:07:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1004.5534", "slug": "377ff0c9070a296d", "title": "Fully Countering Trusting Trust through Diverse Double-Compiling", "authors": ["David A. Wheeler"], "year": 2010, "venue": "arXiv", "url": "https://arxiv.org/abs/1004.5534", "doi": null, "topics": ["Programming languages", "Security & cryptography", "Mathematics"], "description": "An Air Force evaluation of Multics, and Ken Thompson's Turing award lecture (\"Reflections on Trusting Trust\"), showed that compilers can be subverted to insert malicious Trojan horses into critical software, including themselves.", "description_method": "abstract-excerpt", "abstract": "An Air Force evaluation of Multics, and Ken Thompson's Turing award lecture (\"Reflections on Trusting Trust\"), showed that compilers can be subverted to insert malicious Trojan horses into critical software, including themselves. If this \"trusting trust\" attack goes undetected, even complete analysis of a system's source code will not find the malicious code that is running. Previously-known countermeasures have been grossly inadequate. If this attack cannot be countered, attackers can quietly subvert entire classes of computer systems, gaining complete control over financial, infrastructure, military, and/or business systems worldwide. This dissertation's thesis is that the trusting trust attack can be detected and effectively countered using the \"Diverse Double-Compiling\" (DDC) technique, as demonstrated by (1) a formal proof that DDC can determine if source code and generated executable code correspond, (2) a demonstration of DDC with four compilers (a small C compiler, a small Lisp compiler, a small maliciously corrupted Lisp compiler, and a large industrial-strength C compiler, GCC), and (3) a description of approaches for applying DDC in various real-world scenarios. In the DDC technique, source code is compiled twice: the source code of the compiler's parent is compiled using a trusted compiler, and then the putative compiler source code is compiled using the result of the first compilation. If the DDC result is bit-for-bit identical with the original compiler-under-test's executable, and certain other assumptions hold, then the compiler-under-test's executable corresponds with its putative source code.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-12", "2026-03", "2026-04"], "last_mentioned": "2026-04-03T13:33:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1110.1409", "slug": "3f862f4ec16ab99f", "title": "Good Fences: The Importance of Setting Boundaries for Peaceful Coexistence", "authors": ["Alex Rutherford", "Dion Harmon", "Justin Werfel", "Shlomiya Bar-Yam", "Alexander Gard-Murray", "Andreas Gros", "Yaneer Bar-Yam"], "year": 2011, "venue": "arXiv", "url": "https://arxiv.org/abs/1110.1409", "doi": null, "topics": ["Statistics & probability"], "description": "We consider the conditions of peace and violence among ethnic groups, testing a theory designed to predict the locations of violence and interventions that can promote peace.", "description_method": "abstract-excerpt", "abstract": "We consider the conditions of peace and violence among ethnic groups, testing a theory designed to predict the locations of violence and interventions that can promote peace. Characterizing the model's success in predicting peace requires examples where peace prevails despite diversity. Switzerland is recognized as a country of peace, stability and prosperity. This is surprising because of its linguistic and religious diversity that in other parts of the world lead to conflict and violence. Here we analyze how peaceful stability is maintained. Our analysis shows that peace does not depend on integrated coexistence, but rather on well defined topographical and political boundaries separating groups. Mountains and lakes are an important part of the boundaries between sharply defined linguistic areas. Political canton and circle (sub-canton) boundaries often separate religious groups. Where such boundaries do not appear to be sufficient, we find that specific aspects of the population distribution either guarantee sufficient separation or sufficient mixing to inhibit intergroup violence according to the quantitative theory of conflict. In exactly one region, a porous mountain range does not adequately separate linguistic groups and violent conflict has led to the recent creation of the canton of Jura. Our analysis supports the hypothesis that violence between groups can be inhibited by physical and political boundaries. A similar analysis of the area of the former Yugoslavia shows that during widespread ethnic violence existing political boundaries did not coincide with the boundaries of distinct groups, but peace prevailed in specific areas where they did coincide. The success of peace in Switzerland may serve as a model to resolve conflict in other ethnically diverse countries and regions of the world.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-11", "2019-01", "2020-07"], "last_mentioned": "2020-07-15T03:37:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1111.1567", "slug": "965132008757fb97", "title": "Generalization of Conway's \"Game of Life\" to a continuous domain - SmoothLife", "authors": ["Stephan Rafler"], "year": 2011, "venue": "arXiv", "url": "https://arxiv.org/abs/1111.1567", "doi": null, "topics": ["Machine learning"], "description": "We present what we argue is the generic generalization of Conway's \"Game of Life\" to a continuous domain.", "description_method": "abstract-excerpt", "abstract": "We present what we argue is the generic generalization of Conway's \"Game of Life\" to a continuous domain. We describe the theoretical model and the explicit implementation on a computer.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2012-10", "2018-05", "2021-06"], "last_mentioned": "2021-06-12T10:09:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1202.2539", "slug": "cfb8693fbd7a818a", "title": "Quantum Time Crystals", "authors": ["Frank Wilczek"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1202.2539", "doi": "10.1103/PhysRevLett.109.160401", "topics": ["Physics"], "description": "Difficulties around the idea of spontaneous breaking of time translation symmetry in a closed quantum mechanical system are identified, and then overcome in a simple model.", "description_method": "abstract-excerpt", "abstract": "Difficulties around the idea of spontaneous breaking of time translation symmetry in a closed quantum mechanical system are identified, and then overcome in a simple model. The possibility of ordering in imaginary time is also discussed.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2012-08", "2013-05", "2018-06"], "last_mentioned": "2018-06-04T22:14:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1202.6429", "slug": "365ed2ab4bd0040a", "title": "Stable image reconstruction using total variation minimization", "authors": ["Deanna Needell", "Rachel Ward"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1202.6429", "doi": null, "topics": ["Machine learning", "Economics"], "description": "This article presents near-optimal guarantees for accurate and robust image recovery from under-sampled noisy measurements using total variation minimization.", "description_method": "abstract-excerpt", "abstract": "This article presents near-optimal guarantees for accurate and robust image recovery from under-sampled noisy measurements using total variation minimization. In particular, we show that from O(slog(N)) nonadaptive linear measurements, an image can be reconstructed to within the best s-term approximation of its gradient up to a logarithmic factor, and this factor can be removed by taking slightly more measurements. Along the way, we prove a strengthened Sobolev inequality for functions lying in the null space of suitably incoherent matrices.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2013-09", "2019-02", "2024-03"], "last_mentioned": "2024-03-30T08:53:04Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1203.5813", "slug": "81a2b75770cd0117", "title": "Quantum computing and the entanglement frontier", "authors": ["John Preskill"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1203.5813", "doi": null, "topics": ["Physics", "Statistics & probability", "Algorithms"], "description": "Quantum information science explores the frontier of highly complex quantum states, the \"entanglement frontier.\" This study is motivated by the observation (widely believed but unproven) that classical systems cannot simulate highly entangled quantum systems efficiently, and we hope to hasten the day when well controlled quantum systems c", "description_method": "abstract-excerpt", "abstract": "Quantum information science explores the frontier of highly complex quantum states, the \"entanglement frontier.\" This study is motivated by the observation (widely believed but unproven) that classical systems cannot simulate highly entangled quantum systems efficiently, and we hope to hasten the day when well controlled quantum systems can perform tasks surpassing what can be done in the classical world. One way to achieve such \"quantum supremacy\" would be to run an algorithm on a quantum computer which solves a problem with a super-polynomial speedup relative to classical computers, but there may be other ways that can be achieved sooner, such as simulating exotic quantum states of strongly correlated matter. To operate a large scale quantum computer reliably we will need to overcome the debilitating effects of decoherence, which might be done using \"standard\" quantum hardware protected by quantum error-correcting codes, or by exploiting the nonabelian quantum statistics of anyons realized in solid state systems, or by combining both methods. Only by challenging the entanglement frontier will we learn whether Nature provides extravagant resources far beyond what the classical world would allow.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-10", "2019-09", "2019-10"], "last_mentioned": "2019-10-23T13:59:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1204.0162", "slug": "f7388d1f37b52d6d", "title": "The Proof of Innocence", "authors": ["Dmitri Krioukov"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1204.0162", "doi": null, "topics": ["Other research"], "description": "A way to fight your traffic tickets.", "description_method": "abstract-excerpt", "abstract": "A way to fight your traffic tickets. The paper was awarded a special prize of $400 that the author did not have to pay to the state of California. In view of enormous, extremely surprising and completely unexpected public interest to this work, we have added an appendix answering the two most common questions.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2012-08", "2017-07", "2023-10"], "last_mentioned": "2023-10-05T20:18:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1205.0458", "slug": "726f2f5cf166ee3b", "title": "Better Balance by Being Biased: A 0.8776-Approximation for Max Bisection", "authors": ["Per Austrin", "Siavosh Benabbas", "Konstantinos Georgiou"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1205.0458", "doi": null, "topics": ["Algorithms", "Psychology"], "description": "Recently Raghavendra and Tan (SODA 2012) gave a 0.85-approximation algorithm for the Max Bisection problem.", "description_method": "abstract-excerpt", "abstract": "Recently Raghavendra and Tan (SODA 2012) gave a 0.85-approximation algorithm for the Max Bisection problem. We improve their algorithm to a 0.8776-approximation. As Max Bisection is hard to approximate within $α_{GW} + ε\\approx 0.8786$ under the Unique Games Conjecture (UGC), our algorithm is nearly optimal. We conjecture that Max Bisection is approximable within $α_{GW}-ε$, i.e., the bisection constraint (essentially) does not make Max Cut harder. We also obtain an optimal algorithm (assuming the UGC) for the analogous variant of Max 2-Sat. Our approximation ratio for this problem exactly matches the optimal approximation ratio for Max 2-Sat, i.e., $α_{LLZ} + ε\\approx 0.9401$, showing that the bisection constraint does not make Max 2-Sat harder. This improves on a 0.93-approximation for this problem due to Raghavendra and Tan.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-01", "2017-08", "2021-10"], "last_mentioned": "2021-10-28T06:12:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1206.5533", "slug": "9b21b268ab50de57", "title": "Practical recommendations for gradient-based training of deep architectures", "authors": ["Yoshua Bengio"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1206.5533", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "Learning algorithms related to artificial neural networks and in particular for Deep Learning may seem to involve many bells and whistles, called hyper-parameters.", "description_method": "abstract-excerpt", "abstract": "Learning algorithms related to artificial neural networks and in particular for Deep Learning may seem to involve many bells and whistles, called hyper-parameters. This chapter is meant as a practical guide with recommendations for some of the most commonly used hyper-parameters, in particular in the context of learning algorithms based on back-propagated gradient and gradient-based optimization. It also discusses how to deal with the fact that more interesting results can be obtained when allowing one to adjust many hyper-parameters. Overall, it describes elements of the practice used to successfully and efficiently train and debug large-scale and often deep multi-layer neural networks. It closes with open questions about the training difficulties observed with deeper architectures.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2013-11", "2014-05", "2014-12"], "last_mentioned": "2014-12-30T09:15:26Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1210.0530", "slug": "aacb4520b06bb568", "title": "Best Practices for Scientific Computing", "authors": ["Greg Wilson", "D. A. Aruliah", "C. Titus Brown", "Neil P. Chue Hong", "Matt Davis", "Richard T. Guy", "Steven H. D. Haddock", "Katy Huff", "Ian M. Mitchell", "Mark Plumbley", "Ben Waugh", "Ethan P. White", "Paul Wilson"], "year": 2012, "venue": "arXiv", "url": "https://arxiv.org/abs/1210.0530", "doi": "10.1371/journal.pbio.1001745", "topics": ["Science & research"], "description": "Scientists spend an increasing amount of time building and using software.", "description_method": "abstract-excerpt", "abstract": "Scientists spend an increasing amount of time building and using software. However, most scientists are never taught how to do this efficiently. As a result, many are unaware of tools and practices that would allow them to write more reliable and maintainable code with less effort. We describe a set of best practices for scientific software development that have solid foundations in research and experience, and that improve scientists' productivity and the reliability of their software.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2012-11", "2012-12", "2014-10"], "last_mentioned": "2014-10-10T17:18:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1302.5966", "slug": "cae3943f95a4f8f3", "title": "Information Transmission Between Financial Markets in Chicago and New York", "authors": ["Gregory Laughlin", "Anthony Aguirre", "Joseph Grundfest"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1302.5966", "doi": null, "topics": ["Economics"], "description": "High frequency trading has led to widespread efforts to reduce information propagation delays between physically distant exchanges.", "description_method": "abstract-excerpt", "abstract": "High frequency trading has led to widespread efforts to reduce information propagation delays between physically distant exchanges. Using relativistically correct millisecond-resolution tick data, we document a 3-millisecond decrease in one-way communication time between the Chicago and New York areas that has occurred from April 27th, 2010 to August 17th, 2012. We attribute the first segment of this decline to the introduction of a latency-optimized fiber optic connection in late 2010. A second phase of latency decrease can be attributed to line-of-sight microwave networks, operating primarily in the 6-11 GHz region of the spectrum, licensed during 2011 and 2012. Using publicly available information, we estimate these networks' latencies and bandwidths. We estimate the total infrastructure and 5-year operations costs associated with these latency improvements to exceed $500 million.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2014-09", "2018-01", "2018-05"], "last_mentioned": "2018-05-09T15:31:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1302.6165", "slug": "eaf7abfdbcd39144", "title": "The quantum vacuum as the origin of the speed of light", "authors": ["Marcel Urban", "Francois Couchot", "Xavier Sarazin", "Arache Djannati-Atai"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1302.6165", "doi": "10.1140/epjd/e2013-30578-7", "topics": ["Physics", "Statistics & probability"], "description": "We show that the vacuum permeability and permittivity may originate from the magnetization and the polarization of continuously appearing and disappearing fermion pairs.", "description_method": "abstract-excerpt", "abstract": "We show that the vacuum permeability and permittivity may originate from the magnetization and the polarization of continuously appearing and disappearing fermion pairs. We then show that if we simply model the propagation of the photon in vacuum as a series of transient captures within these ephemeral pairs, we can derive a finite photon velocity. Requiring that this velocity is equal to the speed of light constrains our model of vacuum. Within this approach, the propagation of a photon is a statistical process at scales much larger than the Planck scale. Therefore we expect its time of flight to fluctuate. We propose an experimental test of this prediction.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2013-03", "2015-09", "2016-04"], "last_mentioned": "2016-04-20T19:52:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1303.5113", "slug": "491a1f9c6e44a00f", "title": "A theory of regularity structures", "authors": ["Martin Hairer"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1303.5113", "doi": "10.1007/s00222-014-0505-4", "topics": ["Mathematics", "Physics", "Statistics & probability"], "description": "We introduce a new notion of \"regularity structure\" that provides an algebraic framework allowing to describe functions and / or distributions via a kind of \"jet\" or local Taylor expansion around each point.", "description_method": "abstract-excerpt", "abstract": "We introduce a new notion of \"regularity structure\" that provides an algebraic framework allowing to describe functions and / or distributions via a kind of \"jet\" or local Taylor expansion around each point. The main novel idea is to replace the classical polynomial model which is suitable for describing smooth functions by arbitrary models that are purpose-built for the problem at hand. In particular, this allows to describe the local behaviour not only of functions but also of large classes of distributions. We then build a calculus allowing to perform the various operations (multiplication, composition with smooth functions, integration against singular kernels) necessary to formulate fixed point equations for a very large class of semilinear PDEs driven by some very singular (typically random) input. This allows, for the first time, to give a mathematically rigorous meaning to many interesting stochastic PDEs arising in physics. The theory comes with convergence results that allow to interpret the solutions obtained in this way as limits of classical solutions to regularised problems, possibly modified by the addition of diverging counterterms. These counterterms arise naturally through the action of a \"renormalisation group\" which is defined canonically in terms of the regularity structure associated to the given class of PDEs. As an example of a novel application, we solve the long-standing problem of building a natural Markov process that is symmetric with respect to the (finite volume) measure describing the $Φ^4_3$ Euclidean quantum field theory. It is natural to conjecture that the Markov process built in this way describes the Glauber dynamic of 3-dimensional ferromagnets near their critical temperature.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2014-08", "2020-09", "2026-07"], "last_mentioned": "2026-07-31T14:04:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1306.0533", "slug": "e7d828eaf3db4f1a", "title": "Cool horizons for entangled black holes", "authors": ["Juan Maldacena", "Leonard Susskind"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1306.0533", "doi": "10.1002/prop.201300020", "topics": ["Physics"], "description": "General relativity contains solutions in which two distant black holes are connected through the interior via a wormhole, or Einstein-Rosen bridge.", "description_method": "abstract-excerpt", "abstract": "General relativity contains solutions in which two distant black holes are connected through the interior via a wormhole, or Einstein-Rosen bridge. These solutions can be interpreted as maximally entangled states of two black holes that form a complex EPR pair. We suggest that similar bridges might be present for more general entangled states. In the case of entangled black holes one can formulate versions of the AMPS(S) paradoxes and resolve them. This suggests possible resolutions of the firewall paradoxes for more general situations.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2013-07", "2018-11", "2022-11"], "last_mentioned": "2022-11-24T18:06:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1309.4168", "slug": "b3dd3791f2c6b4e7", "title": "Exploiting Similarities among Languages for Machine Translation", "authors": ["Tomas Mikolov", "Quoc V. Le", "Ilya Sutskever"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1309.4168", "doi": null, "topics": ["Statistics & probability", "Distributed systems"], "description": "Dictionaries and phrase tables are the basis of modern statistical machine translation systems.", "description_method": "abstract-excerpt", "abstract": "Dictionaries and phrase tables are the basis of modern statistical machine translation systems. This paper develops a method that can automate the process of generating and extending dictionaries and phrase tables. Our method can translate missing word and phrase entries by learning language structures based on large monolingual data and mapping between languages from small bilingual data. It uses distributed representation of words and learns a linear mapping between vector spaces of languages. Despite its simplicity, our method is surprisingly effective: we can achieve almost 90% precision@5 for translation of words between English and Spanish. This method makes little assumption about the languages, so it can be used to extend and refine dictionaries and translation tables for any language pairs.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2013-10", "2015-06", "2017-03"], "last_mentioned": "2017-03-02T14:59:05Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1310.4691", "slug": "de173decca6af68e", "title": "Time from quantum entanglement: an experimental illustration", "authors": ["Ekaterina Moreva", "Giorgio Brida", "Marco Gramegna", "Vittorio Giovannetti", "Lorenzo Maccone", "Marco Genovese"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1310.4691", "doi": "10.1103/PhysRevA.89.052122", "topics": ["Physics"], "description": "In the last years several theoretical papers discussed if time can be an emergent property deriving from quantum correlations.", "description_method": "abstract-excerpt", "abstract": "In the last years several theoretical papers discussed if time can be an emergent property deriving from quantum correlations. Here, to provide an insight into how this phenomenon can occur, we present an experiment that illustrates Page and Wootters' mechanism of \"static\" time, and Gambini et al. subsequent refinements. A static, entangled state between a clock system and the rest of the universe is perceived as evolving by internal observers that test the correlations between the two subsystems. We implement this mechanism using an entangled state of the polarization of two photons, one of which is used as a clock to gauge the evolution of the second: an \"internal\" observer that becomes correlated with the clock photon sees the other system evolve, while an \"external\" observer that only observes global properties of the two photons can prove it is static.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-02", "2017-08", "2021-05"], "last_mentioned": "2021-05-12T01:51:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.0658", "slug": "5e883b344f9aa2b7", "title": "A Theory of Changes for Higher-Order Languages - Incrementalizing λ-Calculi by Static Differentiation", "authors": ["Yufei Cai", "Paolo G. Giarrusso", "Tillmann Rendel", "Klaus Ostermann"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.0658", "doi": null, "topics": ["Algorithms"], "description": "If the result of an expensive computation is invalidated by a small change to the input, the old result should be updated incrementally instead of reexecuting the whole computation.", "description_method": "abstract-excerpt", "abstract": "If the result of an expensive computation is invalidated by a small change to the input, the old result should be updated incrementally instead of reexecuting the whole computation. We incrementalize programs through their derivative. A derivative maps changes in the program's input directly to changes in the program's output, without reexecuting the original program. We present a program transformation taking programs to their derivatives, which is fully static and automatic, supports first-class functions, and produces derivatives amenable to standard optimization. We prove the program transformation correct in Agda for a family of simply-typed λ-calculi, parameterized by base types and primitives. A precise interface specifies what is required to incrementalize the chosen primitives. We investigate performance by a case study: We implement in Scala the program transformation, a plugin and improve performance of a nontrivial program by orders of magnitude.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-11", "2019-05", "2020-08"], "last_mentioned": "2020-08-07T19:18:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.2007", "slug": "be41e010dbeb2d19", "title": "The Amplituhedron", "authors": ["Nima Arkani-Hamed", "Jaroslav Trnka"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.2007", "doi": "10.1007/JHEP10(2014)030", "topics": ["Mathematics"], "description": "Perturbative scattering amplitudes in gauge theories have remarkable simplicity and hidden infinite dimensional symmetries that are completely obscured in the conventional formulation of field theory using Feynman diagrams.", "description_method": "abstract-excerpt", "abstract": "Perturbative scattering amplitudes in gauge theories have remarkable simplicity and hidden infinite dimensional symmetries that are completely obscured in the conventional formulation of field theory using Feynman diagrams. This suggests the existence of a new understanding for scattering amplitudes where locality and unitarity do not play a central role but are derived consequences from a different starting point. In this note we provide such an understanding for N=4 SYM scattering amplitudes in the planar limit, which we identify as ``the volume\" of a new mathematical object--the Amplituhedron--generalizing the positive Grassmannian. Locality and unitarity emerge hand-in-hand from positive geometry.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-02", "2020-06", "2022-10"], "last_mentioned": "2022-10-19T23:23:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1312.6114", "slug": "03271ba130d423fe", "title": "Auto-Encoding Variational Bayes", "authors": ["Diederik P Kingma", "Max Welling"], "year": 2013, "venue": "arXiv", "url": "https://arxiv.org/abs/1312.6114", "doi": null, "topics": ["Statistics & probability", "Machine learning", "Psychology"], "description": "How can we perform efficient inference and learning in directed probabilistic models, in the presence of continuous latent variables with intractable posterior distributions, and large datasets?", "description_method": "abstract-excerpt", "abstract": "How can we perform efficient inference and learning in directed probabilistic models, in the presence of continuous latent variables with intractable posterior distributions, and large datasets? We introduce a stochastic variational inference and learning algorithm that scales to large datasets and, under some mild differentiability conditions, even works in the intractable case. Our contributions are two-fold. First, we show that a reparameterization of the variational lower bound yields a lower bound estimator that can be straightforwardly optimized using standard stochastic gradient methods. Second, we show that for i.i.d. datasets with continuous latent variables per datapoint, posterior inference can be made especially efficient by fitting an approximate inference model (also called a recognition model) to the intractable posterior using the proposed lower bound estimator. Theoretical advantages are reflected in experimental results.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-09", "2017-10", "2018-10"], "last_mentioned": "2018-10-14T16:34:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1405.1429", "slug": "5adfa5344ca2893e", "title": "How Community Feedback Shapes User Behavior", "authors": ["Justin Cheng", "Cristian Danescu-Niculescu-Mizil", "Jure Leskovec"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1405.1429", "doi": null, "topics": ["Psychology"], "description": "Social media systems rely on user feedback and rating mechanisms for personalization, ranking, and content filtering.", "description_method": "abstract-excerpt", "abstract": "Social media systems rely on user feedback and rating mechanisms for personalization, ranking, and content filtering. However, when users evaluate content contributed by fellow users (e.g., by liking a post or voting on a comment), these evaluations create complex social feedback effects. This paper investigates how ratings on a piece of content affect its author's future behavior. By studying four large comment-based news communities, we find that negative feedback leads to significant behavioral changes that are detrimental to the community. Not only do authors of negatively-evaluated content contribute more, but also their future posts are of lower quality, and are perceived by the community as such. Moreover, these authors are more likely to subsequently evaluate their fellow users negatively, percolating these effects through the community. In contrast, positive feedback does not carry similar effects, and neither encourages rewarded authors to write more, nor improves the quality of their posts. Interestingly, the authors that receive no feedback are most likely to leave a community. Furthermore, a structural analysis of the voter network reveals that evaluations polarize the community the most when positive and negative votes are equally split.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2014-05", "2017-10", "2018-03"], "last_mentioned": "2018-03-13T06:11:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1405.7577", "slug": "348f73b6ed08b29c", "title": "Self-Locating Uncertainty and the Origin of Probability in Everettian Quantum Mechanics", "authors": ["Charles T. Sebens", "Sean M. Carroll"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1405.7577", "doi": "10.1093/bjps/axw004", "topics": ["Physics", "Statistics & probability"], "description": "A longstanding issue in attempts to understand the Everett (Many-Worlds) approach to quantum mechanics is the origin of the Born rule: why is the probability given by the square of the amplitude?", "description_method": "abstract-excerpt", "abstract": "A longstanding issue in attempts to understand the Everett (Many-Worlds) approach to quantum mechanics is the origin of the Born rule: why is the probability given by the square of the amplitude? Following Vaidman, we note that observers are in a position of self-locating uncertainty during the period between the branches of the wave function splitting via decoherence and the observer registering the outcome of the measurement. In this period it is tempting to regard each branch as equiprobable, but we argue that the temptation should be resisted. Applying lessons from this analysis, we demonstrate (using methods similar to those of Zurek's envariance-based derivation) that the Born rule is the uniquely rational way of apportioning credence in Everettian quantum mechanics. In doing so, we rely on a single key principle: changes purely to the environment do not affect the probabilities one ought to assign to measurement outcomes in a local subsystem. We arrive at a method for assigning probabilities in cases that involve both classical and quantum self-locating uncertainty. This method provides unique answers to quantum Sleeping Beauty problems, as well as a well-defined procedure for calculating probabilities in quantum cosmological multiverses with multiple similar observers.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2014-06", "2018-10", "2019-09"], "last_mentioned": "2019-09-10T15:32:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1405.7907", "slug": "45e4024aa8662dc1", "title": "Many Worlds, the Born Rule, and Self-Locating Uncertainty", "authors": ["Sean M. Carroll", "Charles T. Sebens"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1405.7907", "doi": "10.1007/978-88-470-5217-8_10", "topics": ["Physics"], "description": "We provide a derivation of the Born Rule in the context of the Everett (Many-Worlds) approach to quantum mechanics.", "description_method": "abstract-excerpt", "abstract": "We provide a derivation of the Born Rule in the context of the Everett (Many-Worlds) approach to quantum mechanics. Our argument is based on the idea of self-locating uncertainty: in the period between the wave function branching via decoherence and an observer registering the outcome of the measurement, that observer can know the state of the universe precisely without knowing which branch they are on. We show that there is a uniquely rational way to apportion credence in such cases, which leads directly to the Born Rule. Our analysis generalizes straightforwardly to cases of combined classical and quantum self-locating uncertainty, as in the cosmological multiverse.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-03", "2021-09", "2026-03"], "last_mentioned": "2026-03-02T00:50:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1406.2294", "slug": "75e764260fd04e0f", "title": "A Fast, Minimal Memory, Consistent Hash Algorithm", "authors": ["John Lamping", "Eric Veach"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1406.2294", "doi": null, "topics": ["Algorithms", "Distributed systems"], "description": "We present jump consistent hash, a fast, minimal memory, consistent hash algorithm that can be expressed in about 5 lines of code.", "description_method": "abstract-excerpt", "abstract": "We present jump consistent hash, a fast, minimal memory, consistent hash algorithm that can be expressed in about 5 lines of code. In comparison to the algorithm of Karger et al., jump consistent hash requires no storage, is faster, and does a better job of evenly dividing the key space among the buckets and of evenly dividing the workload when the number of buckets changes. Its main limitation is that the buckets must be numbered sequentially, which makes it more suitable for data storage applications than for distributed web caching.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-06", "2021-10", "2022-03"], "last_mentioned": "2022-03-13T23:38:44Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1407.6981", "slug": "f9e1a1721e9a2a9f", "title": "RAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal Response", "authors": ["Úlfar Erlingsson", "Vasyl Pihur", "Aleksandra Korolova"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1407.6981", "doi": "10.1145/2660267.2660348", "topics": ["Security & cryptography", "Statistics & probability"], "description": "Randomized Aggregatable Privacy-Preserving Ordinal Response, or RAPPOR, is a technology for crowdsourcing statistics from end-user client software, anonymously, with strong privacy guarantees.", "description_method": "abstract-excerpt", "abstract": "Randomized Aggregatable Privacy-Preserving Ordinal Response, or RAPPOR, is a technology for crowdsourcing statistics from end-user client software, anonymously, with strong privacy guarantees. In short, RAPPORs allow the forest of client data to be studied, without permitting the possibility of looking at individual trees. By applying randomized response in a novel manner, RAPPOR provides the mechanisms for such collection as well as for efficient, high-utility analysis of the collected data. In particular, RAPPOR permits statistics to be collected on the population of client-side strings with strong privacy guarantees for each client, and without linkability of their reports. This paper describes and motivates RAPPOR, details its differential-privacy and utility guarantees, discusses its practical deployment and properties in the face of different attack models, and, finally, gives results of its application to both synthetic and real-world data.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-06", "2017-08", "2017-09"], "last_mentioned": "2017-09-12T11:02:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1408.3106", "slug": "03c8a0f2cb5a81a4", "title": "Quantum algorithms for topological and geometric analysis of big data", "authors": ["Seth Lloyd", "Silvano Garnerone", "Paolo Zanardi"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1408.3106", "doi": null, "topics": ["Algorithms", "Physics"], "description": "Extracting useful information from large data sets can be a daunting task.", "description_method": "abstract-excerpt", "abstract": "Extracting useful information from large data sets can be a daunting task. Topological methods for analyzing data sets provide a powerful technique for extracting such information. Persistent homology is a sophisticated tool for identifying such topological features -- connected components, holes, or voids -- and for determining how such features persist as the data is viewed at different scales. This paper provides quantum algorithms for calculating Betti numbers in persistent homology, and for finding eigenvectors and eigenvalues of the combinatorial Laplacian. The algorithms provide an exponential speedup over classical algorithms for topological data analysis.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2014-10", "2016-01", "2016-10"], "last_mentioned": "2016-10-17T20:51:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1411.4389", "slug": "44f8febfb6c69c09", "title": "Long-term Recurrent Convolutional Networks for Visual Recognition and Description", "authors": ["Jeff Donahue", "Lisa Anne Hendricks", "Marcus Rohrbach", "Subhashini Venugopalan", "Sergio Guadarrama", "Kate Saenko", "Trevor Darrell"], "year": 2014, "venue": "arXiv", "url": "https://arxiv.org/abs/1411.4389", "doi": null, "topics": ["Computer vision", "Psychology", "Machine learning"], "description": "Models based on deep convolutional networks have dominated recent image interpretation tasks; we investigate whether models which are also recurrent, or \"temporally deep\", are effective for tasks involving sequences, visual and otherwise.", "description_method": "abstract-excerpt", "abstract": "Models based on deep convolutional networks have dominated recent image interpretation tasks; we investigate whether models which are also recurrent, or \"temporally deep\", are effective for tasks involving sequences, visual and otherwise. We develop a novel recurrent convolutional architecture suitable for large-scale visual learning which is end-to-end trainable, and demonstrate the value of these models on benchmark video recognition tasks, image description and retrieval problems, and video narration challenges. In contrast to current models which assume a fixed spatio-temporal receptive field or simple temporal averaging for sequential processing, recurrent convolutional models are \"doubly deep\"' in that they can be compositional in spatial and temporal \"layers\". Such models may have advantages when target concepts are complex and/or training data are limited. Learning long-term dependencies is possible when nonlinearities are incorporated into the network state updates. Long-term RNN models are appealing in that they directly can map variable-length inputs (e.g., video frames) to variable length outputs (e.g., natural language text) and can model complex temporal dynamics; yet they can be optimized with backpropagation. Our recurrent long-term models are directly connected to modern visual convnet models and can be jointly trained to simultaneously learn temporal dynamics and convolutional perceptual representations. Our results show such models have distinct advantages over state-of-the-art models for recognition or generation which are separately defined and/or optimized.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2014-11", "2015-01", "2019-04"], "last_mentioned": "2019-04-29T22:20:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.03492", "slug": "07322644c72d2f29", "title": "Gradient-based Hyperparameter Optimization through Reversible Learning", "authors": ["Dougal Maclaurin", "David Duvenaud", "Ryan P. Adams"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.03492", "doi": null, "topics": ["Machine learning", "Algorithms", "Statistics & probability"], "description": "Tuning hyperparameters of learning algorithms is hard because gradients are usually unavailable.", "description_method": "abstract-excerpt", "abstract": "Tuning hyperparameters of learning algorithms is hard because gradients are usually unavailable. We compute exact gradients of cross-validation performance with respect to all hyperparameters by chaining derivatives backwards through the entire training procedure. These gradients allow us to optimize thousands of hyperparameters, including step-size and momentum schedules, weight initialization distributions, richly parameterized regularization schemes, and neural network architectures. We compute hyperparameter gradients by exactly reversing the dynamics of stochastic gradient descent with momentum.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-05", "2019-10", "2025-06"], "last_mentioned": "2025-06-02T12:20:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.04623", "slug": "01ac98ad639fe746", "title": "DRAW: A Recurrent Neural Network For Image Generation", "authors": ["Karol Gregor", "Ivo Danihelka", "Alex Graves", "Danilo Jimenez Rezende", "Daan Wierstra"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.04623", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "This paper introduces the Deep Recurrent Attentive Writer (DRAW) neural network architecture for image generation.", "description_method": "abstract-excerpt", "abstract": "This paper introduces the Deep Recurrent Attentive Writer (DRAW) neural network architecture for image generation. DRAW networks combine a novel spatial attention mechanism that mimics the foveation of the human eye, with a sequential variational auto-encoding framework that allows for the iterative construction of complex images. The system substantially improves on the state of the art for generative models on MNIST, and, when trained on the Street View House Numbers dataset, it generates images that cannot be distinguished from real data with the naked eye.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-09", "2016-09", "2020-02"], "last_mentioned": "2020-02-20T13:07:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.05700", "slug": "87cc91b0ab44e4a6", "title": "Scalable Bayesian Optimization Using Deep Neural Networks", "authors": ["Jasper Snoek", "Oren Rippel", "Kevin Swersky", "Ryan Kiros", "Nadathur Satish", "Narayanan Sundaram", "Md. Mostofa Ali Patwary", " Prabhat", "Ryan P. Adams"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.05700", "doi": null, "topics": ["Algorithms", "Statistics & probability", "Machine learning"], "description": "Bayesian optimization is an effective methodology for the global optimization of functions with expensive evaluations.", "description_method": "abstract-excerpt", "abstract": "Bayesian optimization is an effective methodology for the global optimization of functions with expensive evaluations. It relies on querying a distribution over functions defined by a relatively cheap surrogate model. An accurate model for this distribution over functions is critical to the effectiveness of the approach, and is typically fit using Gaussian processes (GPs). However, since GPs scale cubically with the number of observations, it has been challenging to handle objectives whose optimization requires many evaluations, and as such, massively parallelizing the optimization. In this work, we explore the use of neural networks as an alternative to GPs to model distributions over functions. We show that performing adaptive basis function regression with a neural network as the parametric form performs competitively with state-of-the-art GP-based approaches, but scales linearly with the number of data rather than cubically. This allows us to achieve a previously intractable degree of parallelism, which we apply to large scale hyperparameter optimization, rapidly finding competitive models on benchmark object recognition tasks using convolutional networks, and image caption generation using neural language models.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-12", "2016-06", "2019-01"], "last_mentioned": "2019-01-03T09:32:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1502.07373", "slug": "96bf499ff34fbd2b", "title": "The Spy in the Sandbox -- Practical Cache Attacks in Javascript", "authors": ["Yossef Oren", "Vasileios P. Kemerlis", "Simha Sethumadhavan", "Angelos D. Keromytis"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1502.07373", "doi": null, "topics": ["Security & cryptography", "Networks", "Computer systems"], "description": "We present the first micro-architectural side-channel attack which runs entirely in the browser.", "description_method": "abstract-excerpt", "abstract": "We present the first micro-architectural side-channel attack which runs entirely in the browser. In contrast to other works in this genre, this attack does not require the attacker to install any software on the victim's machine -- to facilitate the attack, the victim needs only to browse to an untrusted webpage with attacker-controlled content. This makes the attack model highly scalable and extremely relevant and practical to today's web, especially since most desktop browsers currently accessing the Internet are vulnerable to this attack. Our attack, which is an extension of the last-level cache attacks of Yarom et al., allows a remote adversary recover information belonging to other processes, other users and even other virtual machines running on the same physical host as the victim web browser. We describe the fundamentals behind our attack, evaluate its performance using a high bandwidth covert channel and finally use it to construct a system-wide mouse/network activity logger. Defending against this attack is possible, but the required countermeasures can exact an impractical cost on other benign uses of the web browser and of the computer.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-07", "2017-04", "2017-09"], "last_mentioned": "2017-09-08T08:46:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1503.03465", "slug": "34468988bb6d8cf4", "title": "Faster 64-bit universal hashing using carry-less multiplications", "authors": ["Daniel Lemire", "Owen Kaser"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1503.03465", "doi": "10.1007/s13389-015-0110-5", "topics": ["Computer systems", "Algorithms"], "description": "Intel and AMD support the Carry-less Multiplication (CLMUL) instruction set in their x64 processors.", "description_method": "abstract-excerpt", "abstract": "Intel and AMD support the Carry-less Multiplication (CLMUL) instruction set in their x64 processors. We use CLMUL to implement an almost universal 64-bit hash family (CLHASH). We compare this new family with what might be the fastest almost universal family on x64 processors (VHASH). We find that CLHASH is at least 60% faster. We also compare CLHASH with a popular hash function designed for speed (Google's CityHash). We find that CLHASH is 40% faster than CityHash on inputs larger than 64 bytes and just as fast otherwise.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-10", "2016-03", "2016-12"], "last_mentioned": "2016-12-04T18:54:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1503.08895", "slug": "a295691400d97936", "title": "End-To-End Memory Networks", "authors": ["Sainbayar Sukhbaatar", "Arthur Szlam", "Jason Weston", "Rob Fergus"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1503.08895", "doi": null, "topics": ["Machine learning", "Language models"], "description": "We introduce a neural network with a recurrent attention model over a possibly large external memory.", "description_method": "abstract-excerpt", "abstract": "We introduce a neural network with a recurrent attention model over a possibly large external memory. The architecture is a form of Memory Network (Weston et al., 2015) but unlike the model in that work, it is trained end-to-end, and hence requires significantly less supervision during training, making it more generally applicable in realistic settings. It can also be seen as an extension of RNNsearch to the case where multiple computational steps (hops) are performed per output symbol. The flexibility of the model allows us to apply it to tasks as diverse as (synthetic) question answering and to language modeling. For the former our approach is competitive with Memory Networks, but with less supervision. For the latter, on the Penn TreeBank and Text8 datasets our approach demonstrates comparable performance to RNNs and LSTMs. In both cases we show that the key concept of multiple computational hops yields improved results.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-08", "2015-09", "2016-04"], "last_mentioned": "2016-04-05T21:41:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1505.00387", "slug": "83470645fe3a429d", "title": "Highway Networks", "authors": ["Rupesh Kumar Srivastava", "Klaus Greff", "Jürgen Schmidhuber"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1505.00387", "doi": null, "topics": ["Machine learning", "Statistics & probability"], "description": "There is plenty of theoretical and empirical evidence that depth of neural networks is a crucial ingredient for their success.", "description_method": "abstract-excerpt", "abstract": "There is plenty of theoretical and empirical evidence that depth of neural networks is a crucial ingredient for their success. However, network training becomes more difficult with increasing depth and training of very deep networks remains an open problem. In this extended abstract, we introduce a new architecture designed to ease gradient-based training of very deep networks. We refer to networks with this architecture as highway networks, since they allow unimpeded information flow across several layers on \"information highways\". The architecture is characterized by the use of gating units which learn to regulate the flow of information through a network. Highway networks with hundreds of layers can be trained directly using stochastic gradient descent and with a variety of activation functions, opening up the possibility of studying extremely deep and efficient architectures.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-08", "2016-04", "2019-03"], "last_mentioned": "2019-03-27T14:14:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1505.00521", "slug": "41cc9bd681d18e6d", "title": "Reinforcement Learning Neural Turing Machines - Revised", "authors": ["Wojciech Zaremba", "Ilya Sutskever"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1505.00521", "doi": null, "topics": ["Machine learning", "Databases", "Algorithms"], "description": "The Neural Turing Machine (NTM) is more expressive than all previously considered models because of its external memory.", "description_method": "abstract-excerpt", "abstract": "The Neural Turing Machine (NTM) is more expressive than all previously considered models because of its external memory. It can be viewed as a broader effort to use abstract external Interfaces and to learn a parametric model that interacts with them. The capabilities of a model can be extended by providing it with proper Interfaces that interact with the world. These external Interfaces include memory, a database, a search engine, or a piece of software such as a theorem verifier. Some of these Interfaces are provided by the developers of the model. However, many important existing Interfaces, such as databases and search engines, are discrete. We examine feasibility of learning models to interact with discrete Interfaces. We investigate the following discrete Interfaces: a memory Tape, an input Tape, and an output Tape. We use a Reinforcement Learning algorithm to train a neural network that interacts with such Interfaces to solve simple algorithmic tasks. Our Interfaces are expressive enough to make our model Turing complete.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-05", "2016-05", "2020-10"], "last_mentioned": "2020-10-21T21:19:38Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1505.04722", "slug": "fe11f5d4d3ac2cc8", "title": "On the statistical properties and tail risk of violent conflicts", "authors": ["Pasquale Cirillo", "Nassim Nicholas Taleb"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1505.04722", "doi": "10.1016/j.physa.2016.01.050", "topics": ["Statistics & probability"], "description": "We examine statistical pictures of violent conflicts over the last 2000 years, finding techniques for dealing with incompleteness and unreliability of historical data.", "description_method": "abstract-excerpt", "abstract": "We examine statistical pictures of violent conflicts over the last 2000 years, finding techniques for dealing with incompleteness and unreliability of historical data. We introduce a novel approach to apply extreme value theory to fat-tailed variables that have a remote, but nonetheless finite upper bound, by defining a corresponding unbounded dual distribution (given that potential war casualties are bounded by the world population). We apply methods from extreme value theory on the dual distribution and derive its tail properties. The dual method allows us to calculate the real mean of war casualties, which proves to be considerably larger than the sample mean, meaning severe underestimation of the tail risks of conflicts from naive observation. We analyze the robustness of our results to errors in historical reports, taking into account the unreliability of accounts by historians and absence of critical data. We study inter-arrival times between tail events and find that no particular trend can be asserted. All the statistical pictures obtained are at variance with the prevailing claims about \"long peace\", namely that violence has been declining over time.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-03", "2022-07", "2024-02"], "last_mentioned": "2024-02-22T21:00:50Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1506.03134", "slug": "ef7d7dc78b44de3f", "title": "Pointer Networks", "authors": ["Oriol Vinyals", "Meire Fortunato", "Navdeep Jaitly"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1506.03134", "doi": null, "topics": ["Algorithms", "Statistics & probability"], "description": "We introduce a new neural architecture to learn the conditional probability of an output sequence with elements that are discrete tokens corresponding to positions in an input sequence.", "description_method": "abstract-excerpt", "abstract": "We introduce a new neural architecture to learn the conditional probability of an output sequence with elements that are discrete tokens corresponding to positions in an input sequence. Such problems cannot be trivially addressed by existent approaches such as sequence-to-sequence and Neural Turing Machines, because the number of target classes in each step of the output depends on the length of the input, which is variable. Problems such as sorting variable sized sequences, and various combinatorial optimization problems belong to this class. Our model solves the problem of variable size output dictionaries using a recently proposed mechanism of neural attention. It differs from the previous attention attempts in that, instead of using attention to blend hidden units of an encoder to a context vector at each decoder step, it uses attention as a pointer to select a member of the input sequence as the output. We call this architecture a Pointer Net (Ptr-Net). We show Ptr-Nets can be used to learn approximate solutions to three challenging geometric problems -- finding planar convex hulls, computing Delaunay triangulations, and the planar Travelling Salesman Problem -- using training examples alone. Ptr-Nets not only improve over sequence-to-sequence with input attention, but also allow us to generalize to variable size output dictionaries. We show that the learnt models generalize beyond the maximum lengths they were trained on. We hope our results on these tasks will encourage a broader exploration of neural learning for discrete problems.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-05", "2016-10", "2026-07"], "last_mentioned": "2026-07-07T19:03:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1506.06668", "slug": "2b2f729f64ca166d", "title": "Fairy Lights in Femtoseconds: Aerial and Volumetric Graphics Rendered by Focused Femtosecond Laser Combined with Computational Holographic Fields", "authors": ["Yoichi Ochiai", "Kota Kumagai", "Takayuki Hoshi", "Jun Rekimoto", "Satoshi Hasegawa", "Yoshio Hayasaki"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1506.06668", "doi": null, "topics": ["Other research"], "description": "We present a method of rendering aerial and volumetric graphics using femtosecond lasers.", "description_method": "abstract-excerpt", "abstract": "We present a method of rendering aerial and volumetric graphics using femtosecond lasers. A high-intensity laser excites a physical matter to emit light at an arbitrary 3D position. Popular applications can then be explored especially since plasma induced by a femtosecond laser is safer than that generated by a nanosecond laser. There are two methods of rendering graphics with a femtosecond laser in air: Producing holograms using spatial light modulation technology, and scanning of a laser beam by a galvano mirror. The holograms and workspace of the system proposed here occupy a volume of up to 1 cm^3; however, this size is scalable depending on the optical devices and their setup. This paper provides details of the principles, system setup, and experimental evaluation, and discussions on scalability, design space, and applications of this system. We tested two laser sources: an adjustable (30-100 fs) laser which projects up to 1,000 pulses per second at energy up to 7 mJ per pulse, and a 269-fs laser which projects up to 200,000 pulses per second at an energy up to 50 uJ per pulse. We confirmed that the spatiotemporal resolution of volumetric displays, implemented with these laser sources, is 4,000 and 200,000 dots per second. Although we focus on laser-induced plasma in air, the discussion presented here is also applicable to other rendering principles such as fluorescence and microbubble in solid/liquid materials.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-07", "2019-05", "2024-05"], "last_mentioned": "2024-05-14T18:02:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1509.08731", "slug": "1fcb8ae93f75cc91", "title": "Variational Information Maximisation for Intrinsically Motivated Reinforcement Learning", "authors": ["Shakir Mohamed", "Danilo Jimenez Rezende"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1509.08731", "doi": null, "topics": ["Machine learning", "Algorithms", "Statistics & probability"], "description": "The mutual information is a core statistical quantity that has applications in all areas of machine learning, whether this is in training of density models over multiple data modalities, in maximising the efficiency of noisy transmission channels, or when learning behaviour policies for exploration by artificial agents.", "description_method": "abstract-excerpt", "abstract": "The mutual information is a core statistical quantity that has applications in all areas of machine learning, whether this is in training of density models over multiple data modalities, in maximising the efficiency of noisy transmission channels, or when learning behaviour policies for exploration by artificial agents. Most learning algorithms that involve optimisation of the mutual information rely on the Blahut-Arimoto algorithm --- an enumerative algorithm with exponential complexity that is not suitable for modern machine learning applications. This paper provides a new approach for scalable optimisation of the mutual information by merging techniques from variational inference and deep learning. We develop our approach by focusing on the problem of intrinsically-motivated learning, where the mutual information forms the definition of a well-known internal drive known as empowerment. Using a variational lower bound on the mutual information, combined with convolutional networks for handling visual input streams, we develop a stochastic optimisation algorithm that allows for scalable information maximisation and empowerment-based reasoning directly from pixels to actions.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-10", "2015-12", "2016-06"], "last_mentioned": "2016-06-08T16:37:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1510.04931", "slug": "70c1fbe1414c0b17", "title": "Bad Universal Priors and Notions of Optimality", "authors": ["Jan Leike", "Marcus Hutter"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1510.04931", "doi": null, "topics": ["Mathematics", "Algorithms", "Statistics & probability"], "description": "A big open question of algorithmic information theory is the choice of the universal Turing machine (UTM).", "description_method": "abstract-excerpt", "abstract": "A big open question of algorithmic information theory is the choice of the universal Turing machine (UTM). For Kolmogorov complexity and Solomonoff induction we have invariance theorems: the choice of the UTM changes bounds only by a constant. For the universally intelligent agent AIXI (Hutter, 2005) no invariance theorem is known. Our results are entirely negative: we discuss cases in which unlucky or adversarial choices of the UTM cause AIXI to misbehave drastically. We show that Legg-Hutter intelligence and thus balanced Pareto optimality is entirely subjective, and that every policy is Pareto optimal in the class of all computable environments. This undermines all existing optimality properties for AIXI. While it may still serve as a gold standard for AI, our results imply that AIXI is a relative theory, dependent on the choice of the UTM.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-08", "2024-05", "2024-11"], "last_mentioned": "2024-11-24T21:51:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1511.00363", "slug": "65a1eb97b7c64d84", "title": "BinaryConnect: Training Deep Neural Networks with binary weights during propagations", "authors": ["Matthieu Courbariaux", "Yoshua Bengio", "Jean-Pierre David"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1511.00363", "doi": null, "topics": ["Machine learning"], "description": "Deep Neural Networks (DNN) have achieved state-of-the-art results in a wide range of tasks, with the best results obtained with large training sets and large models.", "description_method": "abstract-excerpt", "abstract": "Deep Neural Networks (DNN) have achieved state-of-the-art results in a wide range of tasks, with the best results obtained with large training sets and large models. In the past, GPUs enabled these breakthroughs because of their greater computational speed. In the future, faster computation at both training and test time is likely to be crucial for further progress and for consumer applications on low-power devices. As a result, there is much interest in research and development of dedicated hardware for Deep Learning (DL). Binary weights, i.e., weights which are constrained to only two possible values (e.g. -1 or 1), would bring great benefits to specialized DL hardware by replacing many multiply-accumulate operations by simple accumulations, as multipliers are the most space and power-hungry components of the digital implementation of neural networks. We introduce BinaryConnect, a method which consists in training a DNN with binary weights during the forward and backward propagations, while retaining precision of the stored weights in which gradients are accumulated. Like other dropout schemes, we show that BinaryConnect acts as regularizer and we obtain near state-of-the-art results with BinaryConnect on the permutation-invariant MNIST, CIFAR-10 and SVHN.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-03", "2024-03", "2024-11"], "last_mentioned": "2024-11-09T18:14:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1511.05946", "slug": "b52f859e80926025", "title": "ACDC: A Structured Efficient Linear Layer", "authors": ["Marcin Moczulski", "Misha Denil", "Jeremy Appleyard", "Nando de Freitas"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1511.05946", "doi": null, "topics": ["Machine learning", "Computer vision", "Psychology"], "description": "The linear layer is one of the most pervasive modules in deep learning representations.", "description_method": "abstract-excerpt", "abstract": "The linear layer is one of the most pervasive modules in deep learning representations. However, it requires $O(N^2)$ parameters and $O(N^2)$ operations. These costs can be prohibitive in mobile applications or prevent scaling in many domains. Here, we introduce a deep, differentiable, fully-connected neural network module composed of diagonal matrices of parameters, $\\mathbf{A}$ and $\\mathbf{D}$, and the discrete cosine transform $\\mathbf{C}$. The core module, structured as $\\mathbf{ACDC^{-1}}$, has $O(N)$ parameters and incurs $O(N log N )$ operations. We present theoretical results showing how deep cascades of ACDC layers approximate linear layers. ACDC is, however, a stand-alone module and can be used in combination with any other types of module. In our experiments, we show that it can indeed be successfully interleaved with ReLU modules in convolutional neural networks for image recognition. Our experiments also study critical factors in the training of these structured modules, including initialization and depth. Finally, this paper also provides a connection between structured linear transforms used in deep learning and the field of Fourier optics, illustrating how ACDC could in principle be implemented with lenses and diffractive elements.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-08", "2018-02", "2019-01"], "last_mentioned": "2019-01-26T09:09:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1511.09249", "slug": "07fb644e05fbafbc", "title": "On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models", "authors": ["Juergen Schmidhuber"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1511.09249", "doi": null, "topics": ["Machine learning", "Algorithms", "Statistics & probability"], "description": "This paper addresses the general problem of reinforcement learning (RL) in partially observable environments.", "description_method": "abstract-excerpt", "abstract": "This paper addresses the general problem of reinforcement learning (RL) in partially observable environments. In 2013, our large RL recurrent neural networks (RNNs) learned from scratch to drive simulated cars from high-dimensional video input. However, real brains are more powerful in many ways. In particular, they learn a predictive model of their initially unknown environment, and somehow use it for abstract (e.g., hierarchical) planning and reasoning. Guided by algorithmic information theory, we describe RNN-based AIs (RNNAIs) designed to do the same. Such an RNNAI can be trained on never-ending sequences of tasks, some of them provided by the user, others invented by the RNNAI itself in a curious, playful fashion, to improve its RNN-based world model. Unlike our previous model-building RNN-based RL machines dating back to 1990, the RNNAI learns to actively query its model for abstract reasoning and planning and decision making, essentially \"learning to think.\" The basic ideas of this report can be applied to many other cases where one RNN-like system exploits the algorithmic information content of another. They are taken from a grant proposal submitted in Fall 2014, and also explain concepts such as \"mirror neurons.\" Experimental results will be described in separate papers.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-12", "2018-04", "2023-08"], "last_mentioned": "2023-08-16T17:36:21Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1512.02325", "slug": "a0b89457781ad27b", "title": "SSD: Single Shot MultiBox Detector", "authors": ["Wei Liu", "Dragomir Anguelov", "Dumitru Erhan", "Christian Szegedy", "Scott Reed", "Cheng-Yang Fu", "Alexander C. Berg"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1512.02325", "doi": "10.1007/978-3-319-46448-0_2", "topics": ["Machine learning"], "description": "We present a method for detecting objects in images using a single deep neural network.", "description_method": "abstract-excerpt", "abstract": "We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. Our SSD model is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stage and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, MS COCO, and ILSVRC datasets confirm that SSD has comparable accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. Compared to other single stage methods, SSD has much better accuracy, even with a smaller input image size. For $300\\times 300$ input, SSD achieves 72.1% mAP on VOC2007 test at 58 FPS on a Nvidia Titan X and for $500\\times 500$ input, SSD achieves 75.1% mAP, outperforming a comparable state of the art Faster R-CNN model. Code is available at https://github.com/weiliu89/caffe/tree/ssd .", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-11", "2019-10", "2020-02"], "last_mentioned": "2020-02-22T16:01:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1512.03547", "slug": "310b2d11a43c1b53", "title": "Graph Isomorphism in Quasipolynomial Time", "authors": ["László Babai"], "year": 2015, "venue": "arXiv", "url": "https://arxiv.org/abs/1512.03547", "doi": null, "topics": ["Algorithms", "Security & cryptography"], "description": "We show that the Graph Isomorphism (GI) problem and the related problems of String Isomorphism (under group action) (SI) and Coset Intersection (CI) can be solved in quasipolynomial ($\\exp((\\log n)^{O(1)})$) time.", "description_method": "abstract-excerpt", "abstract": "We show that the Graph Isomorphism (GI) problem and the related problems of String Isomorphism (under group action) (SI) and Coset Intersection (CI) can be solved in quasipolynomial ($\\exp((\\log n)^{O(1)})$) time. The best previous bound for GI was $\\exp(O(\\sqrt{n\\log n}))$, where $n$ is the number of vertices (Luks, 1983); for the other two problems, the bound was similar, $\\exp(\\tilde{O}(\\sqrt{n}))$, where $n$ is the size of the permutation domain (Babai, 1983). The algorithm builds on Luks's SI framework and attacks the barrier configurations for Luks's algorithm by group theoretic \"local certificates\" and combinatorial canonical partitioning techniques. We show that in a well-defined sense, Johnson graphs are the only obstructions to effective canonical partitioning. Luks's barrier situation is characterized by a homomorphism φ that maps a given permutation group $G$ onto $S_k$ or $A_k$, the symmetric or alternating group of degree $k$, where $k$ is not too small. We say that an element $x$ in the permutation domain on which $G$ acts is affected by φ if the φ-image of the stabilizer of $x$ does not contain $A_k$. The affected/unaffected dichotomy underlies the core \"local certificates\" routine and is the central divide-and-conquer tool of the algorithm.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-03", "2016-12", "2022-06"], "last_mentioned": "2022-06-10T12:58:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1601.04589", "slug": "90f15a2499530a87", "title": "Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis", "authors": ["Chuan Li", "Michael Wand"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1601.04589", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "This paper studies a combination of generative Markov random field (MRF) models and discriminatively trained deep convolutional neural networks (dCNNs) for synthesizing 2D images.", "description_method": "abstract-excerpt", "abstract": "This paper studies a combination of generative Markov random field (MRF) models and discriminatively trained deep convolutional neural networks (dCNNs) for synthesizing 2D images. The generative MRF acts on higher-levels of a dCNN feature pyramid, controling the image layout at an abstract level. We apply the method to both photographic and non-photo-realistic (artwork) synthesis tasks. The MRF regularizer prevents over-excitation artifacts and reduces implausible feature mixtures common to previous dCNN inversion approaches, permitting synthezing photographic content with increased visual plausibility. Unlike standard MRF-based texture synthesis, the combined system can both match and adapt local features with considerable variability, yielding results far out of reach of classic generative MRF methods.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-03", "2016-04", "2016-09"], "last_mentioned": "2016-09-30T14:14:19Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1602.01103", "slug": "52d66a2d6a3b1627", "title": "Winning Arguments: Interaction Dynamics and Persuasion Strategies in Good-faith Online Discussions", "authors": ["Chenhao Tan", "Vlad Niculae", "Cristian Danescu-Niculescu-Mizil", "Lillian Lee"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1602.01103", "doi": "10.1145/2872427.2883081", "topics": ["Psychology"], "description": "Changing someone's opinion is arguably one of the most important challenges of social interaction.", "description_method": "abstract-excerpt", "abstract": "Changing someone's opinion is arguably one of the most important challenges of social interaction. The underlying process proves difficult to study: it is hard to know how someone's opinions are formed and whether and how someone's views shift. Fortunately, ChangeMyView, an active community on Reddit, provides a platform where users present their own opinions and reasoning, invite others to contest them, and acknowledge when the ensuing discussions change their original views. In this work, we study these interactions to understand the mechanisms behind persuasion. We find that persuasive arguments are characterized by interesting patterns of interaction dynamics, such as participant entry-order and degree of back-and-forth exchange. Furthermore, by comparing similar counterarguments to the same opinion, we show that language factors play an essential role. In particular, the interplay between the language of the opinion holder and that of the counterargument provides highly predictive cues of persuasiveness. Finally, since even in this favorable setting people may not be persuaded, we investigate the problem of determining whether someone's opinion is susceptible to being changed at all. For this more difficult task, we show that stylistic choices in how the opinion is expressed carry predictive power.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-04", "2019-07", "2025-04"], "last_mentioned": "2025-04-27T01:15:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1602.05314", "slug": "eca3a1dff8883842", "title": "PlaNet - Photo Geolocation with Convolutional Neural Networks", "authors": ["Tobias Weyand", "Ilya Kostrikov", "James Philbin"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1602.05314", "doi": "10.1007/978-3-319-46484-8_3", "topics": ["Computer vision", "Machine learning", "Biology & medicine"], "description": "Is it possible to build a system to determine the location where a photo was taken using just its pixels?", "description_method": "abstract-excerpt", "abstract": "Is it possible to build a system to determine the location where a photo was taken using just its pixels? In general, the problem seems exceptionally difficult: it is trivial to construct situations where no location can be inferred. Yet images often contain informative cues such as landmarks, weather patterns, vegetation, road markings, and architectural details, which in combination may allow one to determine an approximate location and occasionally an exact location. Websites such as GeoGuessr and View from your Window suggest that humans are relatively good at integrating these cues to geolocate images, especially en-masse. In computer vision, the photo geolocation problem is usually approached using image retrieval methods. In contrast, we pose the problem as one of classification by subdividing the surface of the earth into thousands of multi-scale geographic cells, and train a deep network using millions of geotagged images. While previous approaches only recognize landmarks or perform approximate matching using global image descriptors, our model is able to use and integrate multiple visible cues. We show that the resulting model, called PlaNet, outperforms previous approaches and even attains superhuman levels of accuracy in some cases. Moreover, we extend our model to photo albums by combining it with a long short-term memory (LSTM) architecture. By learning to exploit temporal coherence to geolocate uncertain photos, we demonstrate that this model achieves a 50% performance improvement over the single-image model.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-11", "2017-05", "2019-11"], "last_mentioned": "2019-11-21T14:14:33Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1603.00788", "slug": "67f4a36e6e6c579e", "title": "Automatic Differentiation Variational Inference", "authors": ["Alp Kucukelbir", "Dustin Tran", "Rajesh Ranganath", "Andrew Gelman", "David M. Blei"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1603.00788", "doi": null, "topics": ["Algorithms", "Programming languages", "Mathematics"], "description": "Probabilistic modeling is iterative.", "description_method": "abstract-excerpt", "abstract": "Probabilistic modeling is iterative. A scientist posits a simple model, fits it to her data, refines it according to her analysis, and repeats. However, fitting complex models to large data is a bottleneck in this process. Deriving algorithms for new models can be both mathematically and computationally challenging, which makes it difficult to efficiently cycle through the steps. To this end, we develop automatic differentiation variational inference (ADVI). Using our method, the scientist only provides a probabilistic model and a dataset, nothing else. ADVI automatically derives an efficient variational inference algorithm, freeing the scientist to refine and explore many models. ADVI supports a broad class of models-no conjugacy assumptions are required. We study ADVI across ten different models and apply it to a dataset with millions of observations. ADVI is integrated into Stan, a probabilistic programming system; it is available for immediate use.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-09", "2017-03", "2018-10"], "last_mentioned": "2018-10-14T16:34:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1603.01121", "slug": "72224fa9a1e0380d", "title": "Deep Reinforcement Learning from Self-Play in Imperfect-Information Games", "authors": ["Johannes Heinrich", "David Silver"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1603.01121", "doi": null, "topics": ["Machine learning", "Algorithms"], "description": "Many real-world applications can be described as large-scale games of imperfect information.", "description_method": "abstract-excerpt", "abstract": "Many real-world applications can be described as large-scale games of imperfect information. To deal with these challenging domains, prior work has focused on computing Nash equilibria in a handcrafted abstraction of the domain. In this paper we introduce the first scalable end-to-end approach to learning approximate Nash equilibria without prior domain knowledge. Our method combines fictitious self-play with deep reinforcement learning. When applied to Leduc poker, Neural Fictitious Self-Play (NFSP) approached a Nash equilibrium, whereas common reinforcement learning methods diverged. In Limit Texas Holdem, a poker game of real-world scale, NFSP learnt a strategy that approached the performance of state-of-the-art, superhuman algorithms based on significant domain expertise.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-01", "2017-12", "2021-12"], "last_mentioned": "2021-12-13T22:08:31Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1603.01529", "slug": "0c4a46ac3a039a33", "title": "Delta State Replicated Data Types", "authors": ["Paulo Sérgio Almeida", "Ali Shoker", "Carlos Baquero"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1603.01529", "doi": "10.1016/j.jpdc.2017.08.003", "topics": ["Distributed systems", "Statistics & probability", "Algorithms"], "description": "CRDTs are distributed data types that make eventual consistency of a distributed object possible and non ad-hoc.", "description_method": "abstract-excerpt", "abstract": "CRDTs are distributed data types that make eventual consistency of a distributed object possible and non ad-hoc. Specifically, state-based CRDTs ensure convergence through disseminating the entire state, that may be large, and merging it to other replicas; whereas operation-based CRDTs disseminate operations (i.e., small states) assuming an exactly-once reliable dissemination layer. We introduce Delta State Conflict-Free Replicated Data Types ($δ$-CRDTs) that can achieve the best of both worlds: small messages with an incremental nature, as in operation-based CRDTs, disseminated over unreliable communication channels, as in traditional state-based CRDTs. This is achieved by defining delta mutators to return a delta-state, typically with a much smaller size than the full state, that to be joined with both local and remote states. We introduce the $δ$-CRDT framework, and we explain it through establishing a correspondence to current state-based CRDTs. In addition, we present an anti-entropy algorithm for eventual convergence, and another one that ensures causal consistency. Finally, we introduce several $δ$-CRDT specifications of both well-known replicated datatypes and novel datatypes, including a generic map composition.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-09", "2022-04", "2026-03"], "last_mentioned": "2026-03-03T21:35:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1603.09320", "slug": "4072b69e3f23c876", "title": "Efficient and robust approximate nearest neighbor search using Hierarchical Navigable Small World graphs", "authors": ["Yu. A. Malkov", "D. A. Yashunin"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1603.09320", "doi": null, "topics": ["Algorithms", "Statistics & probability", "Distributed systems"], "description": "We present a new approach for the approximate K-nearest neighbor search based on navigable small world graphs with controllable hierarchy (Hierarchical NSW, HNSW).", "description_method": "abstract-excerpt", "abstract": "We present a new approach for the approximate K-nearest neighbor search based on navigable small world graphs with controllable hierarchy (Hierarchical NSW, HNSW). The proposed solution is fully graph-based, without any need for additional search structures, which are typically used at the coarse search stage of the most proximity graph techniques. Hierarchical NSW incrementally builds a multi-layer structure consisting from hierarchical set of proximity graphs (layers) for nested subsets of the stored elements. The maximum layer in which an element is present is selected randomly with an exponentially decaying probability distribution. This allows producing graphs similar to the previously studied Navigable Small World (NSW) structures while additionally having the links separated by their characteristic distance scales. Starting search from the upper layer together with utilizing the scale separation boosts the performance compared to NSW and allows a logarithmic complexity scaling. Additional employment of a heuristic for selecting proximity graph neighbors significantly increases performance at high recall and in case of highly clustered data. Performance evaluation has demonstrated that the proposed general metric space search index is able to strongly outperform previous opensource state-of-the-art vector-only approaches. Similarity of the algorithm to the skip list structure allows straightforward balanced distributed implementation.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-07", "2023-05", "2024-09"], "last_mentioned": "2024-09-05T03:33:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1605.05396", "slug": "dcc4c07ac8bf25cf", "title": "Generative Adversarial Text to Image Synthesis", "authors": ["Scott Reed", "Zeynep Akata", "Xinchen Yan", "Lajanugen Logeswaran", "Bernt Schiele", "Honglak Lee"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1605.05396", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "Automatic synthesis of realistic images from text would be interesting and useful, but current AI systems are still far from this goal.", "description_method": "abstract-excerpt", "abstract": "Automatic synthesis of realistic images from text would be interesting and useful, but current AI systems are still far from this goal. However, in recent years generic and powerful recurrent neural network architectures have been developed to learn discriminative text feature representations. Meanwhile, deep convolutional generative adversarial networks (GANs) have begun to generate highly compelling images of specific categories, such as faces, album covers, and room interiors. In this work, we develop a novel deep architecture and GAN formulation to effectively bridge these advances in text and image model- ing, translating visual concepts from characters to pixels. We demonstrate the capability of our model to generate plausible images of birds and flowers from detailed text descriptions.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-06", "2016-10", "2016-12"], "last_mentioned": "2016-12-08T02:42:04Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1606.01299", "slug": "dc4d056310764b6c", "title": "RAISR: Rapid and Accurate Image Super Resolution", "authors": ["Yaniv Romano", "John Isidoro", "Peyman Milanfar"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1606.01299", "doi": null, "topics": ["Algorithms"], "description": "Given an image, we wish to produce an image of larger size with significantly more pixels and higher image quality.", "description_method": "abstract-excerpt", "abstract": "Given an image, we wish to produce an image of larger size with significantly more pixels and higher image quality. This is generally known as the Single Image Super-Resolution (SISR) problem. The idea is that with sufficient training data (corresponding pairs of low and high resolution images) we can learn set of filters (i.e. a mapping) that when applied to given image that is not in the training set, will produce a higher resolution version of it, where the learning is preferably low complexity. In our proposed approach, the run-time is more than one to two orders of magnitude faster than the best competing methods currently available, while producing results comparable or better than state-of-the-art. A closely related topic is image sharpening and contrast enhancement, i.e., improving the visual quality of a blurry image by amplifying the underlying details (a wide range of frequencies). Our approach additionally includes an extremely efficient way to produce an image that is significantly sharper than the input blurry one, without introducing artifacts such as halos and noise amplification. We illustrate how this effective sharpening algorithm, in addition to being of independent interest, can be used as a pre-processing step to induce the learning of more effective upscaling filters with built-in sharpening and contrast enhancement effect.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-11", "2017-02", "2018-10"], "last_mentioned": "2018-10-10T00:56:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1606.03498", "slug": "66559f00ba26cb80", "title": "Improved Techniques for Training GANs", "authors": ["Tim Salimans", "Ian Goodfellow", "Wojciech Zaremba", "Vicki Cheung", "Alec Radford", "Xi Chen"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1606.03498", "doi": null, "topics": ["Machine learning"], "description": "We present a variety of new architectural features and training procedures that we apply to the generative adversarial networks (GANs) framework.", "description_method": "abstract-excerpt", "abstract": "We present a variety of new architectural features and training procedures that we apply to the generative adversarial networks (GANs) framework. We focus on two applications of GANs: semi-supervised learning, and the generation of images that humans find visually realistic. Unlike most work on generative models, our primary goal is not to train a model that assigns high likelihood to test data, nor do we require the model to be able to learn well without using any labels. Using our new techniques, we achieve state-of-the-art results in semi-supervised classification on MNIST, CIFAR-10 and SVHN. The generated images are of high quality as confirmed by a visual Turing test: our model generates MNIST samples that humans cannot distinguish from real data, and CIFAR-10 samples that yield a human error rate of 21.3%. We also present ImageNet samples with unprecedented resolution and show that our methods enable the model to learn recognizable features of ImageNet classes.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-06", "2016-08", "2017-01"], "last_mentioned": "2017-01-16T02:51:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1606.04442", "slug": "e0e419d463e21c4d", "title": "DeepMath - Deep Sequence Models for Premise Selection", "authors": ["Alex A. Alemi", "Francois Chollet", "Niklas Een", "Geoffrey Irving", "Christian Szegedy", "Josef Urban"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1606.04442", "doi": null, "topics": ["Mathematics", "Machine learning"], "description": "We study the effectiveness of neural sequence models for premise selection in automated theorem proving, one of the main bottlenecks in the formalization of mathematics.", "description_method": "abstract-excerpt", "abstract": "We study the effectiveness of neural sequence models for premise selection in automated theorem proving, one of the main bottlenecks in the formalization of mathematics. We propose a two stage approach for this task that yields good results for the premise selection task on the Mizar corpus while avoiding the hand-engineered features of existing state-of-the-art models. To our knowledge, this is the first time deep learning has been applied to theorem proving on a large scale.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-07", "2016-11", "2020-10"], "last_mentioned": "2020-10-19T07:10:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1606.05915", "slug": "514b203a246e4f44", "title": "Fansmitter: Acoustic Data Exfiltration from (Speakerless) Air-Gapped Computers", "authors": ["Mordechai Guri", "Yosef Solewicz", "Andrey Daidakulov", "Yuval Elovici"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1606.05915", "doi": null, "topics": ["Security & cryptography", "Networks"], "description": "Because computers may contain or interact with sensitive information, they are often air-gapped and in this way kept isolated and disconnected from the Internet.", "description_method": "abstract-excerpt", "abstract": "Because computers may contain or interact with sensitive information, they are often air-gapped and in this way kept isolated and disconnected from the Internet. In recent years the ability of malware to communicate over an air-gap by transmitting sonic and ultrasonic signals from a computer speaker to a nearby receiver has been shown. In order to eliminate such acoustic channels, current best practice recommends the elimination of speakers (internal or external) in secure computers, thereby creating a so-called 'audio-gap'. In this paper, we present Fansmitter, a malware that can acoustically exfiltrate data from air-gapped computers, even when audio hardware and speakers are not present. Our method utilizes the noise emitted from the CPU and chassis fans which are present in virtually every computer today. We show that a software can regulate the internal fans' speed in order to control the acoustic waveform emitted from a computer. Binary data can be modulated and transmitted over these audio signals to a remote microphone (e.g., on a nearby mobile phone). We present Fansmitter's design considerations, including acoustic signature analysis, data modulation, and data transmission. We also evaluate the acoustic channel, present our results, and discuss countermeasures. Using our method we successfully transmitted data from air-gapped computer without audio hardware, to a smartphone receiver in the same room. We demonstrated the effective transmission of encryption keys and passwords from a distance of zero to eight meters, with bit rate of up to 900 bits/hour. We show that our method can also be used to leak data from different types of IT equipment, embedded systems, and IoT devices that have no audio hardware, but contain fans of various types and sizes.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-07", "2025-07", "2025-12"], "last_mentioned": "2025-12-06T16:04:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1608.00263", "slug": "1d5a9cdca68e7042", "title": "Characterizing Quantum Supremacy in Near-Term Devices", "authors": ["Sergio Boixo", "Sergei V. Isakov", "Vadim N. Smelyanskiy", "Ryan Babbush", "Nan Ding", "Zhang Jiang", "Michael J. Bremner", "John M. Martinis", "Hartmut Neven"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1608.00263", "doi": "10.1038/s41567-018-0124-x", "topics": ["Physics", "Statistics & probability", "Algorithms"], "description": "A critical question for the field of quantum computing in the near future is whether quantum devices without error correction can perform a well-defined computational task beyond the capabilities of state-of-the-art classical computers, achieving so-called quantum supremacy.", "description_method": "abstract-excerpt", "abstract": "A critical question for the field of quantum computing in the near future is whether quantum devices without error correction can perform a well-defined computational task beyond the capabilities of state-of-the-art classical computers, achieving so-called quantum supremacy. We study the task of sampling from the output distributions of (pseudo-)random quantum circuits, a natural task for benchmarking quantum computers. Crucially, sampling this distribution classically requires a direct numerical simulation of the circuit, with computational cost exponential in the number of qubits. This requirement is typical of chaotic systems. We extend previous results in computational complexity to argue more formally that this sampling task must take exponential time in a classical computer. We study the convergence to the chaotic regime using extensive supercomputer simulations, modeling circuits with up to 42 qubits - the largest quantum circuits simulated to date for a computational task that approaches quantum supremacy. We argue that while chaotic states are extremely sensitive to errors, quantum supremacy can be achieved in the near-term with approximately fifty superconducting qubits. We introduce cross entropy as a useful benchmark of quantum circuits which approximates the circuit fidelity. We show that the cross entropy can be efficiently measured when circuit simulations are available. Beyond the classically tractable regime, the cross entropy can be extrapolated and compared with theoretical estimates of circuit fidelity to define a practical quantum supremacy test.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-01", "2017-06", "2019-09"], "last_mentioned": "2019-09-22T21:50:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1608.03676", "slug": "af05be1f16e999ec", "title": "Coz: Finding Code that Counts with Causal Profiling", "authors": ["Charlie Curtsinger", "Emery D. Berger"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1608.03676", "doi": "10.1145/2815400.2815409", "topics": ["Algorithms"], "description": "Improving performance is a central concern for software developers.", "description_method": "abstract-excerpt", "abstract": "Improving performance is a central concern for software developers. To locate optimization opportunities, developers rely on software profilers. However, these profilers only report where programs spent their time: optimizing that code may have no impact on performance. Past profilers thus both waste developer time and make it difficult for them to uncover significant optimization opportunities. This paper introduces causal profiling. Unlike past profiling approaches, causal profiling indicates exactly where programmers should focus their optimization efforts, and quantifies their potential impact. Causal profiling works by running performance experiments during program execution. Each experiment calculates the impact of any potential optimization by virtually speeding up code: inserting pauses that slow down all other code running concurrently. The key insight is that this slowdown has the same relative effect as running that line faster, thus \"virtually\" speeding it up. We present Coz, a causal profiler, which we evaluate on a range of highly-tuned applications: Memcached, SQLite, and the PARSEC benchmark suite. Coz identifies previously unknown optimization opportunities that are both significant and targeted. Guided by Coz, we improve the performance of Memcached by 9%, SQLite by 25%, and accelerate six PARSEC applications by as much as 68%; in most cases, these optimizations involve modifying under 10 lines of code.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-02", "2020-06", "2025-01"], "last_mentioned": "2025-01-24T03:59:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.02943", "slug": "3f99ce6d94584d59", "title": "Stealing Machine Learning Models via Prediction APIs", "authors": ["Florian Tramèr", "Fan Zhang", "Ari Juels", "Michael K. Reiter", "Thomas Ristenpart"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.02943", "doi": null, "topics": ["Security & cryptography", "Machine learning", "Databases"], "description": "Machine learning (ML) models may be deemed confidential due to their sensitive training data, commercial value, or use in security applications.", "description_method": "abstract-excerpt", "abstract": "Machine learning (ML) models may be deemed confidential due to their sensitive training data, commercial value, or use in security applications. Increasingly often, confidential ML models are being deployed with publicly accessible query interfaces. ML-as-a-service (\"predictive analytics\") systems are an example: Some allow users to train models on potentially sensitive data and charge others for access on a pay-per-query basis. The tension between model confidentiality and public access motivates our investigation of model extraction attacks. In such attacks, an adversary with black-box access, but no prior knowledge of an ML model's parameters or training data, aims to duplicate the functionality of (i.e., \"steal\") the model. Unlike in classical learning theory settings, ML-as-a-service offerings may accept partial feature vectors as inputs and include confidence values with predictions. Given these practices, we show simple, efficient attacks that extract target ML models with near-perfect fidelity for popular model classes including logistic regression, neural networks, and decision trees. We demonstrate these attacks against the online services of BigML and Amazon Machine Learning. We further show that the natural countermeasure of omitting confidence values from model outputs still admits potentially harmful model extraction attacks. Our results highlight the need for careful ML model deployment and new model extraction countermeasures.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-01", "2017-03", "2018-06"], "last_mentioned": "2018-06-13T16:13:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1609.04938", "slug": "a4219a9ca0a17203", "title": "Image-to-Markup Generation with Coarse-to-Fine Attention", "authors": ["Yuntian Deng", "Anssi Kanervisto", "Jeffrey Ling", "Alexander M. Rush"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1609.04938", "doi": null, "topics": ["Mathematics", "Algorithms"], "description": "We present a neural encoder-decoder model to convert images into presentational markup based on a scalable coarse-to-fine attention mechanism.", "description_method": "abstract-excerpt", "abstract": "We present a neural encoder-decoder model to convert images into presentational markup based on a scalable coarse-to-fine attention mechanism. Our method is evaluated in the context of image-to-LaTeX generation, and we introduce a new dataset of real-world rendered mathematical expressions paired with LaTeX markup. We show that unlike neural OCR techniques using CTC-based models, attention-based approaches can tackle this non-standard OCR task. Our approach outperforms classical mathematical OCR systems by a large margin on in-domain rendered data, and, with pretraining, also performs well on out-of-domain handwritten data. To reduce the inference complexity associated with the attention-based approaches, we introduce a new coarse-to-fine attention layer that selects a support region before applying attention.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-05", "2017-12", "2023-11"], "last_mentioned": "2023-11-03T15:42:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1610.00161", "slug": "df23b1da131ae22f", "title": "Towards deep learning with segregated dendrites", "authors": ["Jordan Guergiuev", "Timothy P. Lillicrap", "Blake A. Richards"], "year": 2016, "venue": "arXiv", "url": "https://arxiv.org/abs/1610.00161", "doi": null, "topics": ["Neuroscience", "Machine learning", "Algorithms"], "description": "Deep learning has led to significant advances in artificial intelligence, in part, by adopting strategies motivated by neurophysiology.", "description_method": "abstract-excerpt", "abstract": "Deep learning has led to significant advances in artificial intelligence, in part, by adopting strategies motivated by neurophysiology. However, it is unclear whether deep learning could occur in the real brain. Here, we show that a deep learning algorithm that utilizes multi-compartment neurons might help us to understand how the brain optimizes cost functions. Like neocortical pyramidal neurons, neurons in our model receive sensory information and higher-order feedback in electrotonically segregated compartments. Thanks to this segregation, the neurons in different layers of the network can coordinate synaptic weight updates. As a result, the network can learn to categorize images better than a single layer network. Furthermore, we show that our algorithm takes advantage of multilayer architectures to identify useful representations---the hallmark of deep learning. This work demonstrates that deep learning can be achieved using segregated dendritic compartments, which may help to explain the dendritic morphology of neocortical pyramidal neurons.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-05", "2018-11", "2019-06"], "last_mentioned": "2019-06-02T21:59:46Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1701.02434", "slug": "d703a62be67862a2", "title": "A Conceptual Introduction to Hamiltonian Monte Carlo", "authors": ["Michael Betancourt"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1701.02434", "doi": null, "topics": ["Statistics & probability", "Mathematics"], "description": "Hamiltonian Monte Carlo has proven a remarkable empirical success, but only recently have we begun to develop a rigorous understanding of why it performs so well on difficult problems and how it is best applied in practice.", "description_method": "abstract-excerpt", "abstract": "Hamiltonian Monte Carlo has proven a remarkable empirical success, but only recently have we begun to develop a rigorous understanding of why it performs so well on difficult problems and how it is best applied in practice. Unfortunately, that understanding is confined within the mathematics of differential geometry which has limited its dissemination, especially to the applied communities for which it is particularly important. In this review I provide a comprehensive conceptual account of these theoretical foundations, focusing on developing a principled intuition behind the method and its optimal implementations rather of any exhaustive rigor. Whether a practitioner or a statistician, the dedicated reader will acquire a solid grasp of how Hamiltonian Monte Carlo works, when it succeeds, and, perhaps most importantly, when it fails.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-11", "2024-03", "2025-12"], "last_mentioned": "2025-12-04T04:48:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1701.06972", "slug": "8819a1f44d552ed0", "title": "Deep Network Guided Proof Search", "authors": ["Sarah Loos", "Geoffrey Irving", "Christian Szegedy", "Cezary Kaliszyk"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1701.06972", "doi": null, "topics": ["Mathematics", "Machine learning", "Psychology"], "description": "Deep learning techniques lie at the heart of several significant AI advances in recent years including object recognition and detection, image captioning, machine translation, speech recognition and synthesis, and playing the game of Go.", "description_method": "abstract-excerpt", "abstract": "Deep learning techniques lie at the heart of several significant AI advances in recent years including object recognition and detection, image captioning, machine translation, speech recognition and synthesis, and playing the game of Go. Automated first-order theorem provers can aid in the formalization and verification of mathematical theorems and play a crucial role in program analysis, theory reasoning, security, interpolation, and system verification. Here we suggest deep learning based guidance in the proof search of the theorem prover E. We train and compare several deep neural network models on the traces of existing ATP proofs of Mizar statements and use them to select processed clauses during proof search. We give experimental evidence that with a hybrid, two-phase approach, deep learning based guidance can significantly reduce the average number of proof search steps while increasing the number of theorems proved. Using a few proof guidance strategies that leverage deep neural networks, we have found first-order proofs of 7.36% of the first-order logic translations of the Mizar Mathematical Library theorems that did not previously have ATP generated proofs. This increases the ratio of statements in the corpus with ATP generated proofs from 56% to 59%.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-08", "2020-10", "2022-11"], "last_mentioned": "2022-11-15T22:06:57Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1701.08734", "slug": "2f747376460dad3a", "title": "PathNet: Evolution Channels Gradient Descent in Super Neural Networks", "authors": ["Chrisantha Fernando", "Dylan Banarse", "Charles Blundell", "Yori Zwols", "David Ha", "Andrei A. Rusu", "Alexander Pritzel", "Daan Wierstra"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1701.08734", "doi": null, "topics": ["Machine learning", "Algorithms", "Psychology"], "description": "For artificial general intelligence (AGI) it would be efficient if multiple users trained the same giant neural network, permitting parameter reuse, without catastrophic forgetting.", "description_method": "abstract-excerpt", "abstract": "For artificial general intelligence (AGI) it would be efficient if multiple users trained the same giant neural network, permitting parameter reuse, without catastrophic forgetting. PathNet is a first step in this direction. It is a neural network algorithm that uses agents embedded in the neural network whose task is to discover which parts of the network to re-use for new tasks. Agents are pathways (views) through the network which determine the subset of parameters that are used and updated by the forwards and backwards passes of the backpropogation algorithm. During learning, a tournament selection genetic algorithm is used to select pathways through the neural network for replication and mutation. Pathway fitness is the performance of that pathway measured according to a cost function. We demonstrate successful transfer learning; fixing the parameters along a path learned on task A and re-evolving a new population of paths for task B, allows task B to be learned faster than it could be learned from scratch or after fine-tuning. Paths evolved on task B re-use parts of the optimal path evolved on task A. Positive transfer was demonstrated for binary MNIST, CIFAR, and SVHN supervised learning classification tasks, and a set of Atari and Labyrinth reinforcement learning tasks, suggesting PathNets have general applicability for neural network training. Finally, PathNet also significantly improves the robustness to hyperparameter choices of a parallel asynchronous reinforcement learning algorithm (A3C).", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-02", "2017-03", "2020-06"], "last_mentioned": "2020-06-29T14:05:02Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1702.06230", "slug": "6e99978cae0c0a7c", "title": "Beating the World's Best at Super Smash Bros. with Deep Reinforcement Learning", "authors": ["Vlad Firoiu", "William F. Whitney", "Joshua B. Tenenbaum"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1702.06230", "doi": null, "topics": ["Machine learning", "Psychology", "Algorithms"], "description": "There has been a recent explosion in the capabilities of game-playing artificial intelligence.", "description_method": "abstract-excerpt", "abstract": "There has been a recent explosion in the capabilities of game-playing artificial intelligence. Many classes of RL tasks, from Atari games to motor control to board games, are now solvable by fairly generic algorithms, based on deep learning, that learn to play from experience with minimal knowledge of the specific domain of interest. In this work, we will investigate the performance of these methods on Super Smash Bros. Melee (SSBM), a popular console fighting game. The SSBM environment has complex dynamics and partial observability, making it challenging for human and machine alike. The multi-player aspect poses an additional challenge, as the vast majority of recent advances in RL have focused on single-agent environments. Nonetheless, we will show that it is possible to train agents that are competitive against and even surpass human professionals, a new result for the multi-player video game setting.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-04", "2017-12", "2018-07"], "last_mentioned": "2018-07-18T17:21:04Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1702.07825", "slug": "cb2cfdf13bc5855e", "title": "Deep Voice: Real-time Neural Text-to-Speech", "authors": ["Sercan O. Arik", "Mike Chrzanowski", "Adam Coates", "Gregory Diamos", "Andrew Gibiansky", "Yongguo Kang", "Xian Li", "John Miller", "Andrew Ng", "Jonathan Raiman", "Shubho Sengupta", "Mohammad Shoeybi"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1702.07825", "doi": null, "topics": ["Machine learning", "Computer vision", "Economics"], "description": "We present Deep Voice, a production-quality text-to-speech system constructed entirely from deep neural networks.", "description_method": "abstract-excerpt", "abstract": "We present Deep Voice, a production-quality text-to-speech system constructed entirely from deep neural networks. Deep Voice lays the groundwork for truly end-to-end neural speech synthesis. The system comprises five major building blocks: a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conversion model, a phoneme duration prediction model, a fundamental frequency prediction model, and an audio synthesis model. For the segmentation model, we propose a novel way of performing phoneme boundary detection with deep neural networks using connectionist temporal classification (CTC) loss. For the audio synthesis model, we implement a variant of WaveNet that requires fewer parameters and trains faster than the original. By using a neural network for each component, our system is simpler and more flexible than traditional text-to-speech systems, where each component requires laborious feature engineering and extensive domain expertise. Finally, we show that inference with our system can be performed faster than real time and describe optimized WaveNet inference kernels on both CPU and GPU that achieve up to 400x speedups over existing implementations.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-02", "2017-03", "2024-10"], "last_mentioned": "2024-10-04T01:31:57Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1703.03400", "slug": "16ebd01aca85309c", "title": "Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks", "authors": ["Chelsea Finn", "Pieter Abbeel", "Sergey Levine"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1703.03400", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Computer vision"], "description": "We propose an algorithm for meta-learning that is model-agnostic, in the sense that it is compatible with any model trained with gradient descent and applicable to a variety of different learning problems, including classification, regression, and reinforcement learning.", "description_method": "abstract-excerpt", "abstract": "We propose an algorithm for meta-learning that is model-agnostic, in the sense that it is compatible with any model trained with gradient descent and applicable to a variety of different learning problems, including classification, regression, and reinforcement learning. The goal of meta-learning is to train a model on a variety of learning tasks, such that it can solve new learning tasks using only a small number of training samples. In our approach, the parameters of the model are explicitly trained such that a small number of gradient steps with a small amount of training data from a new task will produce good generalization performance on that task. In effect, our method trains the model to be easy to fine-tune. We demonstrate that this approach leads to state-of-the-art performance on two few-shot image classification benchmarks, produces good results on few-shot regression, and accelerates fine-tuning for policy gradient reinforcement learning with neural network policies.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-01", "2018-06", "2019-03"], "last_mentioned": "2019-03-28T21:24:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1703.03779", "slug": "eef926d0008ba8d2", "title": "Dissecting Ponzi schemes on Ethereum: identification, analysis, and impact", "authors": ["Massimo Bartoletti", "Salvatore Carta", "Tiziana Cimoli", "Roberto Saia"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1703.03779", "doi": "10.1016/J.FUTURE.2019.08.014", "topics": ["Psychology", "Economics"], "description": "Ponzi schemes are financial frauds which lure users under the promise of high profits.", "description_method": "abstract-excerpt", "abstract": "Ponzi schemes are financial frauds which lure users under the promise of high profits. Actually, users are repaid only with the investments of new users joining the scheme: consequently, a Ponzi scheme implodes soon after users stop joining it. Originated in the offline world 150 years ago, Ponzi schemes have since then migrated to the digital world, approaching first the Web, and more recently hanging over cryptocurrencies like Bitcoin. Smart contract platforms like Ethereum have provided a new opportunity for scammers, who have now the possibility of creating \"trustworthy\" frauds that still make users lose money, but at least are guaranteed to execute \"correctly\". We present a comprehensive survey of Ponzi schemes on Ethereum, analysing their behaviour and their impact from various viewpoints.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-06", "2017-08", "2017-10"], "last_mentioned": "2017-10-17T14:02:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1703.03906", "slug": "22d282f9e40f3304", "title": "Massive Exploration of Neural Machine Translation Architectures", "authors": ["Denny Britz", "Anna Goldie", "Minh-Thang Luong", "Quoc Le"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1703.03906", "doi": null, "topics": ["Machine learning"], "description": "Neural Machine Translation (NMT) has shown remarkable progress over the past few years with production systems now being deployed to end-users.", "description_method": "abstract-excerpt", "abstract": "Neural Machine Translation (NMT) has shown remarkable progress over the past few years with production systems now being deployed to end-users. One major drawback of current architectures is that they are expensive to train, typically requiring days to weeks of GPU time to converge. This makes exhaustive hyperparameter search, as is commonly done with other neural network architectures, prohibitively expensive. In this work, we present the first large-scale analysis of NMT architecture hyperparameters. We report empirical results and variance numbers for several hundred experimental runs, corresponding to over 250,000 GPU hours on the standard WMT English to German translation task. Our experiments lead to novel insights and practical advice for building and extending NMT architectures. As part of this contribution, we release an open-source NMT framework that enables researchers to easily experiment with novel techniques and reproduce state of the art results.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-03", "2017-06", "2024-10"], "last_mentioned": "2024-10-17T22:13:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1703.04416", "slug": "f63eab4f8109c09d", "title": "Users prefer Guetzli JPEG over same-sized libjpeg", "authors": ["Jyrki Alakuijala", "Robert Obryk", "Zoltan Szabadka", "Jan Wassenberg"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1703.04416", "doi": null, "topics": ["Psychology"], "description": "We report on pairwise comparisons by human raters of JPEG images from libjpeg and our new Guetzli encoder.", "description_method": "abstract-excerpt", "abstract": "We report on pairwise comparisons by human raters of JPEG images from libjpeg and our new Guetzli encoder. Although both files are size-matched, 75% of ratings are in favor of Guetzli. This implies the Butteraugli psychovisual image similarity metric which guides Guetzli is reasonably close to human perception at high quality levels. We provide access to the raw ratings and source images for further analysis and study.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-03", "2017-05", "2018-05"], "last_mentioned": "2018-05-07T08:27:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1703.04730", "slug": "4940709e5cec7aeb", "title": "Understanding Black-box Predictions via Influence Functions", "authors": ["Pang Wei Koh", "Percy Liang"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1703.04730", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Security & cryptography"], "description": "How can we explain the predictions of a black-box model?", "description_method": "abstract-excerpt", "abstract": "How can we explain the predictions of a black-box model? In this paper, we use influence functions -- a classic technique from robust statistics -- to trace a model's prediction through the learning algorithm and back to its training data, thereby identifying training points most responsible for a given prediction. To scale up influence functions to modern machine learning settings, we develop a simple, efficient implementation that requires only oracle access to gradients and Hessian-vector products. We show that even on non-convex and non-differentiable models where the theory breaks down, approximations to influence functions can still provide valuable information. On linear models and convolutional neural networks, we demonstrate that influence functions are useful for multiple purposes: understanding model behavior, debugging models, detecting dataset errors, and even creating visually-indistinguishable training-set attacks.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-07", "2023-02", "2023-03"], "last_mentioned": "2023-03-21T20:26:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1703.06907", "slug": "07c510cf5c518a00", "title": "Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World", "authors": ["Josh Tobin", "Rachel Fong", "Alex Ray", "Jonas Schneider", "Wojciech Zaremba", "Pieter Abbeel"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1703.06907", "doi": null, "topics": ["Robotics", "Machine learning"], "description": "Bridging the 'reality gap' that separates simulated robotics from experiments on hardware could accelerate robotic research through improved data availability.", "description_method": "abstract-excerpt", "abstract": "Bridging the 'reality gap' that separates simulated robotics from experiments on hardware could accelerate robotic research through improved data availability. This paper explores domain randomization, a simple technique for training models on simulated images that transfer to real images by randomizing rendering in the simulator. With enough variability in the simulator, the real world may appear to the model as just another variation. We focus on the task of object localization, which is a stepping stone to general robotic manipulation skills. We find that it is possible to train a real-world object detector that is accurate to $1.5$cm and robust to distractors and partial occlusions using only data from a simulator with non-realistic random textures. To demonstrate the capabilities of our detectors, we show they can be used to perform grasping in a cluttered environment. To our knowledge, this is the first successful transfer of a deep neural network trained only on simulated RGB images (without pre-training on real images) to the real world for the purpose of robotic control.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-04", "2021-02", "2021-04"], "last_mentioned": "2021-04-21T10:43:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1704.01212", "slug": "5a56e2c5511d23e0", "title": "Neural Message Passing for Quantum Chemistry", "authors": ["Justin Gilmer", "Samuel S. Schoenholz", "Patrick F. Riley", "Oriol Vinyals", "George E. Dahl"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1704.01212", "doi": null, "topics": ["Machine learning", "Physics", "Biology & medicine"], "description": "Supervised learning on molecules has incredible potential to be useful in chemistry, drug discovery, and materials science.", "description_method": "abstract-excerpt", "abstract": "Supervised learning on molecules has incredible potential to be useful in chemistry, drug discovery, and materials science. Luckily, several promising and closely related neural network models invariant to molecular symmetries have already been described in the literature. These models learn a message passing algorithm and aggregation procedure to compute a function of their entire input graph. At this point, the next step is to find a particularly effective variant of this general approach and apply it to chemical prediction benchmarks until we either solve them or reach the limits of the approach. In this paper, we reformulate existing models into a single common framework we call Message Passing Neural Networks (MPNNs) and explore additional novel variations within this framework. Using MPNNs we demonstrate state of the art results on an important molecular property prediction benchmark; these results are strong enough that we believe future work should focus on datasets with larger molecules or more accurate ground truth labels.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-08", "2021-01", "2026-07"], "last_mentioned": "2026-07-07T19:03:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1704.04861", "slug": "b1cfac874995617c", "title": "MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications", "authors": ["Andrew G. Howard", "Menglong Zhu", "Bo Chen", "Dmitry Kalenichenko", "Weijun Wang", "Tobias Weyand", "Marco Andreetto", "Hartwig Adam"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1704.04861", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "We present a class of efficient models called MobileNets for mobile and embedded vision applications.", "description_method": "abstract-excerpt", "abstract": "We present a class of efficient models called MobileNets for mobile and embedded vision applications. MobileNets are based on a streamlined architecture that uses depth-wise separable convolutions to build light weight deep neural networks. We introduce two simple global hyper-parameters that efficiently trade off between latency and accuracy. These hyper-parameters allow the model builder to choose the right sized model for their application based on the constraints of the problem. We present extensive experiments on resource and accuracy tradeoffs and show strong performance compared to other popular models on ImageNet classification. We then demonstrate the effectiveness of MobileNets across a wide range of applications and use cases including object detection, finegrain classification, face attributes and large scale geo-localization.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-05", "2017-07", "2020-02"], "last_mentioned": "2020-02-24T22:44:39Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1705.01088", "slug": "e524ce81d92c18b8", "title": "Visual Attribute Transfer through Deep Image Analogy", "authors": ["Jing Liao", "Yuan Yao", "Lu Yuan", "Gang Hua", "Sing Bing Kang"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1705.01088", "doi": null, "topics": ["Computer vision"], "description": "We propose a new technique for visual attribute transfer across images that may have very different appearance but have perceptually similar semantic structure.", "description_method": "abstract-excerpt", "abstract": "We propose a new technique for visual attribute transfer across images that may have very different appearance but have perceptually similar semantic structure. By visual attribute transfer, we mean transfer of visual information (such as color, tone, texture, and style) from one image to another. For example, one image could be that of a painting or a sketch while the other is a photo of a real scene, and both depict the same type of scene. Our technique finds semantically-meaningful dense correspondences between two input images. To accomplish this, it adapts the notion of \"image analogy\" with features extracted from a Deep Convolutional Neutral Network for matching; we call our technique Deep Image Analogy. A coarse-to-fine strategy is used to compute the nearest-neighbor field for generating the results. We validate the effectiveness of our proposed method in a variety of cases, including style/texture transfer, color/style swap, sketch/painting to photo, and time lapse.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-05", "2017-09", "2017-11"], "last_mentioned": "2017-11-14T07:25:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1705.10694", "slug": "05f6d70266614f16", "title": "Deep Learning is Robust to Massive Label Noise", "authors": ["David Rolnick", "Andreas Veit", "Serge Belongie", "Nir Shavit"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1705.10694", "doi": null, "topics": ["Machine learning", "Psychology", "Computer vision"], "description": "Deep neural networks trained on large supervised datasets have led to impressive results in image classification and other tasks.", "description_method": "abstract-excerpt", "abstract": "Deep neural networks trained on large supervised datasets have led to impressive results in image classification and other tasks. However, well-annotated datasets can be time-consuming and expensive to collect, lending increased interest to larger but noisy datasets that are more easily obtained. In this paper, we show that deep neural networks are capable of generalizing from training data for which true labels are massively outnumbered by incorrect labels. We demonstrate remarkably high test performance after training on corrupted data from MNIST, CIFAR, and ImageNet. For example, on MNIST we obtain test accuracy above 90 percent even after each clean training example has been diluted with 100 randomly-labeled examples. Such behavior holds across multiple patterns of label noise, even when erroneous labels are biased towards confusing classes. We show that training in this regime requires a significant but manageable increase in dataset size that is related to the factor by which correct labels have been diluted. Finally, we provide an analysis of our results that shows how increasing noise decreases the effective batch size.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-03", "2022-10", "2024-07"], "last_mentioned": "2024-07-25T05:40:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1707.03501", "slug": "fda72718909693ed", "title": "NO Need to Worry about Adversarial Examples in Object Detection in Autonomous Vehicles", "authors": ["Jiajun Lu", "Hussein Sibai", "Evan Fabry", "David Forsyth"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1707.03501", "doi": null, "topics": ["Robotics", "Machine learning", "Computer vision"], "description": "It has been shown that most machine learning algorithms are susceptible to adversarial perturbations.", "description_method": "abstract-excerpt", "abstract": "It has been shown that most machine learning algorithms are susceptible to adversarial perturbations. Slightly perturbing an image in a carefully chosen direction in the image space may cause a trained neural network model to misclassify it. Recently, it was shown that physical adversarial examples exist: printing perturbed images then taking pictures of them would still result in misclassification. This raises security and safety concerns. However, these experiments ignore a crucial property of physical objects: the camera can view objects from different distances and at different angles. In this paper, we show experiments that suggest that current constructions of physical adversarial examples do not disrupt object detection from a moving platform. Instead, a trained neural network classifies most of the pictures taken from different distances and angles of a perturbed image correctly. We believe this is because the adversarial property of the perturbation is sensitive to the scale at which the perturbed picture is viewed, so (for example) an autonomous car will misclassify a stop sign only from a small range of distances. Our work raises an important question: can one construct examples that are adversarial for many or most viewing conditions? If so, the construction should offer very significant insights into the internal representation of patterns by deep networks. If not, there is a good prospect that adversarial examples can be reduced to a curiosity with little practical impact.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-07", "2017-08", "2017-11"], "last_mentioned": "2017-11-03T00:14:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1711.08028", "slug": "90653ca1287e37bb", "title": "Recurrent Relational Networks", "authors": ["Rasmus Berg Palm", "Ulrich Paquet", "Ole Winther"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1711.08028", "doi": null, "topics": ["Databases", "Machine learning"], "description": "This paper is concerned with learning to solve tasks that require a chain of interdependent steps of relational inference, like answering complex questions about the relationships between objects, or solving puzzles where the smaller elements of a solution mutually constrain each other.", "description_method": "abstract-excerpt", "abstract": "This paper is concerned with learning to solve tasks that require a chain of interdependent steps of relational inference, like answering complex questions about the relationships between objects, or solving puzzles where the smaller elements of a solution mutually constrain each other. We introduce the recurrent relational network, a general purpose module that operates on a graph representation of objects. As a generalization of Santoro et al. [2017]'s relational network, it can augment any neural network model with the capacity to do many-step relational reasoning. We achieve state of the art results on the bAbI textual question-answering dataset with the recurrent relational network, consistently solving 20/20 tasks. As bAbI is not particularly challenging from a relational reasoning point of view, we introduce Pretty-CLEVR, a new diagnostic dataset for relational reasoning. In the Pretty-CLEVR set-up, we can vary the question to control for the number of relational reasoning steps that are required to obtain the answer. Using Pretty-CLEVR, we probe the limitations of multi-layer perceptrons, relational and recurrent relational networks. Finally, we show how recurrent relational networks can learn to solve Sudoku puzzles from supervised training data, a challenging task requiring upwards of 64 steps of relational reasoning. We achieve state-of-the-art results amongst comparable methods by solving 96.6% of the hardest Sudoku puzzles.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-12", "2019-02", "2023-07"], "last_mentioned": "2023-07-06T23:55:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1711.10925", "slug": "141b85a3f2f68ae4", "title": "Deep Image Prior", "authors": ["Dmitry Ulyanov", "Andrea Vedaldi", "Victor Lempitsky"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1711.10925", "doi": "10.1007/s11263-020-01303-4", "topics": ["Computer vision", "Statistics & probability", "Psychology"], "description": "Deep convolutional networks have become a popular tool for image generation and restoration.", "description_method": "abstract-excerpt", "abstract": "Deep convolutional networks have become a popular tool for image generation and restoration. Generally, their excellent performance is imputed to their ability to learn realistic image priors from a large number of example images. In this paper, we show that, on the contrary, the structure of a generator network is sufficient to capture a great deal of low-level image statistics prior to any learning. In order to do so, we show that a randomly-initialized neural network can be used as a handcrafted prior with excellent results in standard inverse problems such as denoising, super-resolution, and inpainting. Furthermore, the same prior can be used to invert deep neural representations to diagnose them, and to restore images based on flash-no flash input pairs. Apart from its diverse applications, our approach highlights the inductive bias captured by standard generator network architectures. It also bridges the gap between two very popular families of image restoration methods: learning-based methods using deep convolutional networks and learning-free methods based on handcrafted image priors such as self-similarity. Code and supplementary material are available at https://dmitryulyanov.github.io/deep_image_prior .", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-08", "2021-02", "2021-12"], "last_mentioned": "2021-12-27T08:50:22Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1711.11053", "slug": "ed11423137af8328", "title": "A Multi-Horizon Quantile Recurrent Forecaster", "authors": ["Ruofeng Wen", "Kari Torkkola", "Balakrishnan Narayanaswamy", "Dhruv Madeka"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1711.11053", "doi": null, "topics": ["Statistics & probability", "Machine learning", "Computer vision"], "description": "We propose a framework for general probabilistic multi-step time series regression.", "description_method": "abstract-excerpt", "abstract": "We propose a framework for general probabilistic multi-step time series regression. Specifically, we exploit the expressiveness and temporal nature of Sequence-to-Sequence Neural Networks (e.g. recurrent and convolutional structures), the nonparametric nature of Quantile Regression and the efficiency of Direct Multi-Horizon Forecasting. A new training scheme, *forking-sequences*, is designed for sequential nets to boost stability and performance. We show that the approach accommodates both temporal and static covariates, learning across multiple related series, shifting seasonality, future planned event spikes and cold-starts in real life large-scale forecasting. The performance of the framework is demonstrated in an application to predict the future demand of items sold on Amazon.com, and in a public probabilistic forecasting competition to predict electricity price and load.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-11", "2021-07", "2022-12"], "last_mentioned": "2022-12-13T02:16:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1712.02950", "slug": "816397ae152c64b9", "title": "CycleGAN, a Master of Steganography", "authors": ["Casey Chu", "Andrey Zhmoginov", "Mark Sandler"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1712.02950", "doi": null, "topics": ["Security & cryptography", "Distributed systems"], "description": "CycleGAN (Zhu et al.", "description_method": "abstract-excerpt", "abstract": "CycleGAN (Zhu et al. 2017) is one recent successful approach to learn a transformation between two image distributions. In a series of experiments, we demonstrate an intriguing property of the model: CycleGAN learns to \"hide\" information about a source image into the images it generates in a nearly imperceptible, high-frequency signal. This trick ensures that the generator can recover the original sample and thus satisfy the cyclic consistency requirement, while the generated image remains realistic. We connect this phenomenon with adversarial attacks by viewing CycleGAN's training procedure as training a generator of adversarial examples and demonstrate that the cyclic consistency loss causes CycleGAN to be especially vulnerable to adversarial attacks.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-09", "2025-08", "2026-03"], "last_mentioned": "2026-03-05T10:12:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1712.05090", "slug": "c020351fc1f1c3c3", "title": "Secure Encrypted Virtualization is Unsecure", "authors": ["Zhao-Hui Du", "Zhiwei Ying", "Zhenke Ma", "Yufei Mai", "Phoebe Wang", "Jesse Liu", "Jesse Fang"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1712.05090", "doi": null, "topics": ["Security & cryptography", "Computer systems", "Economics"], "description": "Virtualization has become more important since cloud computing is getting more and more popular than before.", "description_method": "abstract-excerpt", "abstract": "Virtualization has become more important since cloud computing is getting more and more popular than before. There is an increasing demand for security among the cloud customers. AMD plans to provide Secure Encrypted Virtualization (SEV) technology in its latest processor EPYC to protect virtual machines by encrypting its memory but without integrity protection. In this paper, we analyzed the weakness in the SEV design due to lack of integrity protection thus it is not so secure. Using different design flaw in physical address-based tweak algorithm to protect against ciphertext block move attacks, we found a realistic attack against SEV which could obtain the root privilege of an encrypted virtual machine protected by SEV. A demo to simulate the attack against a virtual machine protected by SEV is done in a Ryzen machine which supports Secure Memory Encryption (SME) technology since SEV enabled machine is still not available in market.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-06", "2020-07", "2021-03"], "last_mentioned": "2021-03-10T02:04:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1712.06567", "slug": "7191dc11e224beb3", "title": "Deep Neuroevolution: Genetic Algorithms Are a Competitive Alternative for Training Deep Neural Networks for Reinforcement Learning", "authors": ["Felipe Petroski Such", "Vashisht Madhavan", "Edoardo Conti", "Joel Lehman", "Kenneth O. Stanley", "Jeff Clune"], "year": 2017, "venue": "arXiv", "url": "https://arxiv.org/abs/1712.06567", "doi": null, "topics": ["Machine learning", "Algorithms", "Statistics & probability"], "description": "Deep artificial neural networks (DNNs) are typically trained via gradient-based learning algorithms, namely backpropagation.", "description_method": "abstract-excerpt", "abstract": "Deep artificial neural networks (DNNs) are typically trained via gradient-based learning algorithms, namely backpropagation. Evolution strategies (ES) can rival backprop-based algorithms such as Q-learning and policy gradients on challenging deep reinforcement learning (RL) problems. However, ES can be considered a gradient-based algorithm because it performs stochastic gradient descent via an operation similar to a finite-difference approximation of the gradient. That raises the question of whether non-gradient-based evolutionary algorithms can work at DNN scales. Here we demonstrate they can: we evolve the weights of a DNN with a simple, gradient-free, population-based genetic algorithm (GA) and it performs well on hard deep RL problems, including Atari and humanoid locomotion. The Deep GA successfully evolves networks with over four million free parameters, the largest neural networks ever evolved with a traditional evolutionary algorithm. These results (1) expand our sense of the scale at which GAs can operate, (2) suggest intriguingly that in some cases following the gradient is not the best choice for optimizing performance, and (3) make immediately available the multitude of neuroevolution techniques that improve performance. We demonstrate the latter by showing that combining DNNs with novelty search, which encourages exploration on tasks with deceptive or sparse reward functions, can solve a high-dimensional problem on which reward-maximizing algorithms (e.g.\\ DQN, A3C, ES, and the GA) fail. Additionally, the Deep GA is faster than ES, A3C, and DQN (it can train Atari in ${\\raise.17ex\\hbox{$\\scriptstyle\\sim$}}$4 hours on one desktop or ${\\raise.17ex\\hbox{$\\scriptstyle\\sim$}}$1 hour distributed on 720 cores), and enables a state-of-the-art, up to 10,000-fold compact encoding technique.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-12", "2018-06", "2018-07"], "last_mentioned": "2018-07-26T15:08:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1802.01561", "slug": "b9447d3c450b8d29", "title": "IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures", "authors": ["Lasse Espeholt", "Hubert Soyer", "Remi Munos", "Karen Simonyan", "Volodymir Mnih", "Tom Ward", "Yotam Doron", "Vlad Firoiu", "Tim Harley", "Iain Dunning", "Shane Legg", "Koray Kavukcuoglu"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1802.01561", "doi": null, "topics": ["Distributed systems", "Machine learning"], "description": "In this work we aim to solve a large collection of tasks using a single reinforcement learning agent with a single set of parameters.", "description_method": "abstract-excerpt", "abstract": "In this work we aim to solve a large collection of tasks using a single reinforcement learning agent with a single set of parameters. A key challenge is to handle the increased amount of data and extended training time. We have developed a new distributed agent IMPALA (Importance Weighted Actor-Learner Architecture) that not only uses resources more efficiently in single-machine training but also scales to thousands of machines without sacrificing data efficiency or resource utilisation. We achieve stable learning at high throughput by combining decoupled acting and learning with a novel off-policy correction method called V-trace. We demonstrate the effectiveness of IMPALA for multi-task reinforcement learning on DMLab-30 (a set of 30 tasks from the DeepMind Lab environment (Beattie et al., 2016)) and Atari-57 (all available Atari games in Arcade Learning Environment (Bellemare et al., 2013a)). Our results show that IMPALA is able to achieve better performance than previous agents with less data, and crucially exhibits positive transfer between tasks as a result of its multi-task approach.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-02", "2018-05", "2018-08"], "last_mentioned": "2018-08-22T16:05:58Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1802.02353", "slug": "dda07e7097123783", "title": "Recent Advances in Neural Program Synthesis", "authors": ["Neel Kant"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1802.02353", "doi": null, "topics": ["Psychology", "Machine learning"], "description": "In recent years, deep learning has made tremendous progress in a number of fields that were previously out of reach for artificial intelligence.", "description_method": "abstract-excerpt", "abstract": "In recent years, deep learning has made tremendous progress in a number of fields that were previously out of reach for artificial intelligence. The successes in these problems has led researchers to consider the possibilities for intelligent systems to tackle a problem that humans have only recently themselves considered: program synthesis. This challenge is unlike others such as object recognition and speech translation, since its abstract nature and demand for rigor make it difficult even for human minds to attempt. While it is still far from being solved or even competitive with most existing methods, neural program synthesis is a rapidly growing discipline which holds great promise if completely realized. In this paper, we start with exploring the problem statement and challenges of program synthesis. Then, we examine the fascinating evolution of program induction models, along with how they have succeeded, failed and been reimagined since. Finally, we conclude with a contrastive look at program synthesis and future research recommendations for the field.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-09", "2019-03", "2020-01"], "last_mentioned": "2020-01-16T12:03:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1802.10026", "slug": "44c66c7ef8acd737", "title": "Loss Surfaces, Mode Connectivity, and Fast Ensembling of DNNs", "authors": ["Timur Garipov", "Pavel Izmailov", "Dmitrii Podoprikhin", "Dmitry Vetrov", "Andrew Gordon Wilson"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1802.10026", "doi": null, "topics": ["Machine learning"], "description": "The loss functions of deep neural networks are complex and their geometric properties are not well understood.", "description_method": "abstract-excerpt", "abstract": "The loss functions of deep neural networks are complex and their geometric properties are not well understood. We show that the optima of these complex loss functions are in fact connected by simple curves over which training and test accuracy are nearly constant. We introduce a training procedure to discover these high-accuracy pathways between modes. Inspired by this new geometric insight, we also propose a new ensembling method entitled Fast Geometric Ensembling (FGE). Using FGE we can train high-performing ensembles in the time required to train a single model. We achieve improved performance compared to the recent state-of-the-art Snapshot Ensembles, on CIFAR-10, CIFAR-100, and ImageNet.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-02", "2024-06", "2026-06"], "last_mentioned": "2026-06-14T20:06:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.07055", "slug": "1862d4c25ee6c382", "title": "Simple random search provides a competitive approach to reinforcement learning", "authors": ["Horia Mania", "Aurelia Guy", "Benjamin Recht"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.07055", "doi": null, "topics": ["Machine learning", "Algorithms", "Robotics"], "description": "A common belief in model-free reinforcement learning is that methods based on random search in the parameter space of policies exhibit significantly worse sample complexity than those that explore the space of actions.", "description_method": "abstract-excerpt", "abstract": "A common belief in model-free reinforcement learning is that methods based on random search in the parameter space of policies exhibit significantly worse sample complexity than those that explore the space of actions. We dispel such beliefs by introducing a random search method for training static, linear policies for continuous control problems, matching state-of-the-art sample efficiency on the benchmark MuJoCo locomotion tasks. Our method also finds a nearly optimal controller for a challenging instance of the Linear Quadratic Regulator, a classical problem in control theory, when the dynamics are not known. Computationally, our random search algorithm is at least 15 times more efficient than the fastest competing model-free methods on these benchmarks. We take advantage of this computational efficiency to evaluate the performance of our method over hundreds of random seeds and many different hyperparameter configurations for each benchmark task. Our simulations highlight a high variability in performance in these benchmark tasks, suggesting that commonly used estimations of sample efficiency do not adequately evaluate the performance of RL algorithms.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-06", "2018-11", "2022-06"], "last_mentioned": "2022-06-09T20:25:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.08930", "slug": "8861b3cb76041ced", "title": "The Big Bang, CPT, and neutrino dark matter", "authors": ["Latham Boyle", "Kieran Finn", "Neil Turok"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.08930", "doi": "10.1016/j.aop.2022.168767", "topics": ["Physics", "Economics"], "description": "We investigate the idea that the universe before the Big Bang is the $CPT$ reflection of the universe after the bang, both classically and quantum mechanically, so that the universe does {\\it not} spontaneously violate $CPT$.", "description_method": "abstract-excerpt", "abstract": "We investigate the idea that the universe before the Big Bang is the $CPT$ reflection of the universe after the bang, both classically and quantum mechanically, so that the universe does {\\it not} spontaneously violate $CPT$. We show how $CPT$ symmetry selects a preferred vacuum state for quantum fields on a $CPT$-invariant cosmological background spacetime. The universe before the bang and the universe after the bang may be viewed as a universe/anti-universe pair, emerging directly into the hot, radiation-dominated era we observe in our past. This, in turn, leads to a remarkably economical explanation of the cosmological dark matter. With no additional fields beyond Einstein gravity and the standard model of particle physics (including right-handed neutrinos), a $\\mathbb{Z}_{2}$ symmetry stabilizes one of the right-handed neutrinos. We calculate its abundance in detail and show that, in order to match the observed dark matter density, its mass must be $4.8\\times10^{8}~{\\rm GeV}$. We obtain several further predictions, including: (i) that the three light neutrinos are majorana; (ii) that one of these is exactly massless; and (iii) that, in the absence of an epoch of cosmic inflation, there should be no primordial, long-wavelength gravitational waves. We also briefly discuss the natural origin of the matter-antimatter asymmetry within this picture and possibilities for explaining the cosmological perturbations.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-03", "2022-12", "2023-07"], "last_mentioned": "2023-07-26T06:11:47Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1803.10228", "slug": "81670b2d3d6f4928", "title": "Demystifying Differentiable Programming: Shift/Reset the Penultimate Backpropagator", "authors": ["Fei Wang", "Daniel Zheng", "James Decker", "Xilun Wu", "Grégory M. Essertel", "Tiark Rompf"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1803.10228", "doi": null, "topics": ["Machine learning", "Programming languages", "Algorithms"], "description": "Deep learning has seen tremendous success over the past decade in computer vision, machine translation, and gameplay.", "description_method": "abstract-excerpt", "abstract": "Deep learning has seen tremendous success over the past decade in computer vision, machine translation, and gameplay. This success rests in crucial ways on gradient-descent optimization and the ability to learn parameters of a neural network by backpropagating observed errors. However, neural network architectures are growing increasingly sophisticated and diverse, which motivates an emerging quest for even more general forms of differentiable programming, where arbitrary parameterized computations can be trained by gradient descent. In this paper, we take a fresh look at automatic differentiation (AD) techniques, and especially aim to demystify the reverse-mode form of AD that generalizes backpropagation in neural networks. We uncover a tight connection between reverse-mode AD and delimited continuations, which permits implementing reverse-mode AD purely via operator overloading and without any auxiliary data structures. We further show how this formulation of AD can be fruitfully combined with multi-stage programming (staging), leading to a highly efficient implementation that combines the performance benefits of deep learning frameworks based on explicit reified computation graphs (e.g., TensorFlow) with the expressiveness of pure library approaches (e.g., PyTorch).", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-11", "2019-02", "2021-02"], "last_mentioned": "2021-02-13T15:17:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1804.02767", "slug": "0bf6ffb67361f592", "title": "YOLOv3: An Incremental Improvement", "authors": ["Joseph Redmon", "Ali Farhadi"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1804.02767", "doi": null, "topics": ["Other research"], "description": "We present some updates to YOLO!", "description_method": "abstract-excerpt", "abstract": "We present some updates to YOLO! We made a bunch of little design changes to make it better. We also trained this new network that's pretty swell. It's a little bigger than last time but more accurate. It's still fast though, don't worry. At 320x320 YOLOv3 runs in 22 ms at 28.2 mAP, as accurate as SSD but three times faster. When we look at the old .5 IOU mAP detection metric YOLOv3 is quite good. It achieves 57.9 mAP@50 in 51 ms on a Titan X, compared to 57.5 mAP@50 in 198 ms by RetinaNet, similar performance but 3.8x faster. As always, all the code is online at https://pjreddie.com/yolo/", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-01", "2019-10", "2022-11"], "last_mentioned": "2022-11-01T16:09:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1804.08959", "slug": "ae9ddc5e6a73110a", "title": "WhoTracks .Me: Shedding light on the opaque world of online tracking", "authors": ["Arjaldo Karaj", "Sam Macbeth", "Rémi Berson", "Josep M. Pujol"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1804.08959", "doi": null, "topics": ["Security & cryptography", "Psychology"], "description": "Online tracking has become of increasing concern in recent years, however our understanding of its extent to date has been limited to snapshots from web crawls.", "description_method": "abstract-excerpt", "abstract": "Online tracking has become of increasing concern in recent years, however our understanding of its extent to date has been limited to snapshots from web crawls. Previous at-tempts to measure the tracking ecosystem, have been done using instrumented measurement platforms, which are not able to accurately capture how people interact with the web. In this work we present a method for the measurement of tracking in the web through a browser extension, as well as a method for the aggregation and collection of this information which protects the privacy of participants. We deployed this extension to more than 5 million users, enabling measurement across multiple countries, ISPs and browser configurations, to give an accurate picture of real-world tracking. The result is the largest and longest measurement of online tracking to date based on real users, covering 1.5 billion page loads gathered over 12 months. The data, detailing tracking behaviour over a year, is made publicly available to help drive transparency around online tracking practices.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-08", "2019-12", "2023-07"], "last_mentioned": "2023-07-21T19:36:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1805.02544", "slug": "0a2dba3584069f4e", "title": "Need for Sleep: the Impact of a Night of Sleep Deprivation on Novice Developers' Performance", "authors": ["Davide Fucci", "Giuseppe Scanniello", "Simone Romano", "Natalia Juristo"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1805.02544", "doi": null, "topics": ["Programming languages"], "description": "We present a quasi-experiment to investigate whether, and to what extent, sleep deprivation impacts the performance of novice software developers using the agile practice of test-first development (TFD).", "description_method": "abstract-excerpt", "abstract": "We present a quasi-experiment to investigate whether, and to what extent, sleep deprivation impacts the performance of novice software developers using the agile practice of test-first development (TFD). We recruited 45 undergraduates and asked them to tackle a programming task. Among the participants, 23 agreed to stay awake the night before carrying out the task, while 22 slept usually. We analyzed the quality (i.e., the functional correctness) of the implementations delivered by the participants in both groups, their engagement in writing source code (i.e., the amount of activities performed in the IDE while tackling the programming task) and ability to apply TFD (i.e., the extent to which a participant can use this practice). By comparing the two groups of participants, we found that a single night of sleep deprivation leads to a reduction of 50% in the quality of the implementations. There is important evidence that the developers' engagement and their prowess to apply TFD are negatively impacted. Our results also show that sleep-deprived developers make more fixes to syntactic mistakes in the source code. We conclude that sleep deprivation has possibly disruptive effects on software development activities. The results open opportunities for improving developers' performance by integrating the study of sleep with other psycho-physiological factors in which the software engineering research community has recently taken an interest in.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-07", "2020-03", "2022-10"], "last_mentioned": "2022-10-31T19:20:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1806.01261", "slug": "cb979a96dee83ebe", "title": "Relational inductive biases, deep learning, and graph networks", "authors": ["Peter W. Battaglia", "Jessica B. Hamrick", "Victor Bapst", "Alvaro Sanchez-Gonzalez", "Vinicius Zambaldi", "Mateusz Malinowski", "Andrea Tacchetti", "David Raposo", "Adam Santoro", "Ryan Faulkner", "Caglar Gulcehre", "Francis Song", "Andrew Ballard", "Justin Gilmer", "George Dahl", "Ashish Vaswani", "Kelsey Allen", "Charles Nash", "Victoria Langston", "Chris Dyer", "Nicolas Heess", "Daan Wierstra", "Pushmeet Kohli", "Matt Botvinick", "Oriol Vinyals", "Yujia Li", "Razvan Pascanu"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1806.01261", "doi": null, "topics": ["Psychology", "Machine learning", "Databases"], "description": "Artificial intelligence (AI) has undergone a renaissance recently, making major progress in key domains such as vision, language, control, and decision-making.", "description_method": "abstract-excerpt", "abstract": "Artificial intelligence (AI) has undergone a renaissance recently, making major progress in key domains such as vision, language, control, and decision-making. This has been due, in part, to cheap data and cheap compute resources, which have fit the natural strengths of deep learning. However, many defining characteristics of human intelligence, which developed under much different pressures, remain out of reach for current approaches. In particular, generalizing beyond one's experiences--a hallmark of human intelligence from infancy--remains a formidable challenge for modern AI. The following is part position paper, part review, and part unification. We argue that combinatorial generalization must be a top priority for AI to achieve human-like abilities, and that structured representations and computations are key to realizing this objective. Just as biology uses nature and nurture cooperatively, we reject the false choice between \"hand-engineering\" and \"end-to-end\" learning, and instead advocate for an approach which benefits from their complementary strengths. We explore how using relational inductive biases within deep learning architectures can facilitate learning about entities, relations, and rules for composing them. We present a new building block for the AI toolkit with a strong relational inductive bias--the graph network--which generalizes and extends various approaches for neural networks that operate on graphs, and provides a straightforward interface for manipulating structured knowledge and producing structured behaviors. We discuss how graph networks can support relational reasoning and combinatorial generalization, laying the foundation for more sophisticated, interpretable, and flexible patterns of reasoning. As a companion to this paper, we have released an open-source software library for building graph networks, with demonstrations of how to use them in practice.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-06", "2018-08", "2019-08"], "last_mentioned": "2019-08-18T10:17:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1809.02840", "slug": "aa09123f4e3e8651", "title": "Neural Guided Constraint Logic Programming for Program Synthesis", "authors": ["Lisa Zhang", "Gregory Rosenblatt", "Ethan Fetaya", "Renjie Liao", "William E. Byrd", "Matthew Might", "Raquel Urtasun", "Richard Zemel"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1809.02840", "doi": null, "topics": ["Programming languages", "Machine learning", "Psychology"], "description": "Synthesizing programs using example input/outputs is a classic problem in artificial intelligence.", "description_method": "abstract-excerpt", "abstract": "Synthesizing programs using example input/outputs is a classic problem in artificial intelligence. We present a method for solving Programming By Example (PBE) problems by using a neural model to guide the search of a constraint logic programming system called miniKanren. Crucially, the neural model uses miniKanren's internal representation as input; miniKanren represents a PBE problem as recursive constraints imposed by the provided examples. We explore Recurrent Neural Network and Graph Neural Network models. We contribute a modified miniKanren, drivable by an external agent, available at https://github.com/xuexue/neuralkanren. We show that our neural-guided approach using constraints can synthesize programs faster in many cases, and importantly, can generalize to larger problems.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-06", "2023-08", "2025-01"], "last_mentioned": "2025-01-02T15:09:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1809.07430", "slug": "abb795ecc87b1b39", "title": "CRN++: Molecular Programming Language", "authors": ["Marko Vasic", "David Soloveichik", "Sarfraz Khurshid"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1809.07430", "doi": "10.1007/978-3-030-00030-1", "topics": ["Biology & medicine", "Programming languages", "Algorithms"], "description": "Synthetic biology is a rapidly emerging research area, with expected wide-ranging impact in biology, nanofabrication, and medicine.", "description_method": "abstract-excerpt", "abstract": "Synthetic biology is a rapidly emerging research area, with expected wide-ranging impact in biology, nanofabrication, and medicine. A key technical challenge lies in embedding computation in molecular contexts where electronic micro-controllers cannot be inserted. This necessitates effective representation of computation using molecular components. While previous work established the Turing-completeness of chemical reactions, defining representations that are faithful, efficient, and practical remains challenging. This paper introduces CRN++, a new language for programming deterministic (mass-action) chemical kinetics to perform computation. We present its syntax and semantics, and build a compiler translating CRN++ programs into chemical reactions, thereby laying the foundation of a comprehensive framework for molecular programming. Our language addresses the key challenge of embedding familiar imperative constructs into a set of chemical reactions happening simultaneously and manipulating real-valued concentrations. Although some deviation from ideal output value cannot be avoided, we develop methods to minimize the error, and implement error analysis tools. We demonstrate the feasibility of using CRN++ on a suite of well-known algorithms for discrete and real-valued computation. CRN++ can be easily extended to support new commands or chemical reaction implementations, and thus provides a foundation for developing more robust and practical molecular programs.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-11", "2020-04", "2021-05"], "last_mentioned": "2021-05-02T17:14:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1809.08632", "slug": "0aafee91849a1479", "title": "BrainNet: A Multi-Person Brain-to-Brain Interface for Direct Collaboration Between Brains", "authors": ["Linxing Preston Jiang", "Andrea Stocco", "Darby M. Losey", "Justin A. Abernethy", "Chantel S. Prat", "Rajesh P. N. Rao"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1809.08632", "doi": "10.1038/s41598-019-41895-7", "topics": ["Neuroscience", "Economics", "Networks"], "description": "We present BrainNet which, to our knowledge, is the first multi-person non-invasive direct brain-to-brain interface for collaborative problem solving.", "description_method": "abstract-excerpt", "abstract": "We present BrainNet which, to our knowledge, is the first multi-person non-invasive direct brain-to-brain interface for collaborative problem solving. The interface combines electroencephalography (EEG) to record brain signals and transcranial magnetic stimulation (TMS) to deliver information noninvasively to the brain. The interface allows three human subjects to collaborate and solve a task using direct brain-to-brain communication. Two of the three subjects are \"Senders\" whose brain signals are decoded using real-time EEG data analysis to extract decisions about whether to rotate a block in a Tetris-like game before it is dropped to fill a line. The Senders' decisions are transmitted via the Internet to the brain of a third subject, the \"Receiver,\" who cannot see the game screen. The decisions are delivered to the Receiver's brain via magnetic stimulation of the occipital cortex. The Receiver integrates the information received and makes a decision using an EEG interface about either turning the block or keeping it in the same position. A second round of the game gives the Senders one more chance to validate and provide feedback to the Receiver's action. We evaluated the performance of BrainNet in terms of (1) Group-level performance during the game; (2) True/False positive rates of subjects' decisions; (3) Mutual information between subjects. Five groups of three subjects successfully used BrainNet to perform the Tetris task, with an average accuracy of 0.813. Furthermore, by varying the information reliability of the Senders by artificially injecting noise into one Sender's signal, we found that Receivers are able to learn which Sender is more reliable based solely on the information transmitted to their brains. Our results raise the possibility of future brain-to-brain interfaces that enable cooperative problem solving by humans using a \"social network\" of connected brains.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-07", "2021-04", "2022-08"], "last_mentioned": "2022-08-05T03:51:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1811.03116", "slug": "cd4979081a0a3694", "title": "A postquantum theory of classical gravity?", "authors": ["Jonathan Oppenheim"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1811.03116", "doi": null, "topics": ["Physics", "Mathematics", "Statistics & probability"], "description": "The effort to discover a quantum theory of gravity is motivated by the need to reconcile the incompatibility between quantum theory and general relativity.", "description_method": "abstract-excerpt", "abstract": "The effort to discover a quantum theory of gravity is motivated by the need to reconcile the incompatibility between quantum theory and general relativity. Here, we present an alternative approach by constructing a consistent theory of classical gravity coupled to quantum field theory. The dynamics is linear in the density matrix, completely positive and trace preserving, and reduces to Einstein's theory of general relativity in the classical limit. Consequently, the dynamics doesn't suffer from the pathologies of the semiclassical theory based on expectation values. The assumption that general relativity is classical necessarily modifies the dynamical laws of quantum mechanics -- the theory must be fundamentally stochastic in both the metric degrees of freedom and in the quantum matter fields. This allows it to evade several no-go theorems purporting to forbid classical-quantum interactions. The measurement postulate of quantum mechanics is not needed -- the interaction of the quantum degrees of freedom with classical space-time necessarily causes decoherence in the quantum system. We first derive the general form of classical-quantum dynamics and consider realisations which have as its limit deterministic classical Hamiltonian evolution. The formalism is then applied to quantum field theory interacting with the classical space-time metric. One can view the classical-quantum theory as fundamental or as an effective theory useful for computing the back-reaction of quantum fields on geometry. We discuss a number of open questions from the perspective of both viewpoints.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-07", "2023-12", "2024-03"], "last_mentioned": "2024-03-15T23:46:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1812.11118", "slug": "5d5c0c2b60735c59", "title": "Reconciling modern machine learning practice and the bias-variance trade-off", "authors": ["Mikhail Belkin", "Daniel Hsu", "Siyuan Ma", "Soumik Mandal"], "year": 2018, "venue": "arXiv", "url": "https://arxiv.org/abs/1812.11118", "doi": "10.1073/pnas.1903070116", "topics": ["Machine learning", "Psychology", "Mathematics"], "description": "Breakthroughs in machine learning are rapidly changing science and society, yet our fundamental understanding of this technology has lagged far behind.", "description_method": "abstract-excerpt", "abstract": "Breakthroughs in machine learning are rapidly changing science and society, yet our fundamental understanding of this technology has lagged far behind. Indeed, one of the central tenets of the field, the bias-variance trade-off, appears to be at odds with the observed behavior of methods used in the modern machine learning practice. The bias-variance trade-off implies that a model should balance under-fitting and over-fitting: rich enough to express underlying structure in data, simple enough to avoid fitting spurious patterns. However, in the modern practice, very rich models such as neural networks are trained to exactly fit (i.e., interpolate) the data. Classically, such models would be considered over-fit, and yet they often obtain high accuracy on test data. This apparent contradiction has raised questions about the mathematical foundations of machine learning and their relevance to practitioners. In this paper, we reconcile the classical understanding and the modern practice within a unified performance curve. This \"double descent\" curve subsumes the textbook U-shaped bias-variance trade-off curve by showing how increasing model capacity beyond the point of interpolation results in improved performance. We provide evidence for the existence and ubiquity of double descent for a wide spectrum of models and datasets, and we posit a mechanism for its emergence. This connection between the performance and the structure of machine learning models delineates the limits of classical analyses, and has implications for both the theory and practice of machine learning.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-10", "2021-03", "2021-04"], "last_mentioned": "2021-04-05T13:08:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1902.01007", "slug": "d53b8f22f250aa74", "title": "Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inference", "authors": ["R. Thomas McCoy", "Ellie Pavlick", "Tal Linzen"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1902.01007", "doi": null, "topics": ["Statistics & probability", "Machine learning", "Language models"], "description": "A machine learning system can score well on a given test set by relying on heuristics that are effective for frequent example types but break down in more challenging cases.", "description_method": "abstract-excerpt", "abstract": "A machine learning system can score well on a given test set by relying on heuristics that are effective for frequent example types but break down in more challenging cases. We study this issue within natural language inference (NLI), the task of determining whether one sentence entails another. We hypothesize that statistical NLI models may adopt three fallible syntactic heuristics: the lexical overlap heuristic, the subsequence heuristic, and the constituent heuristic. To determine whether models have adopted these heuristics, we introduce a controlled evaluation set called HANS (Heuristic Analysis for NLI Systems), which contains many examples where the heuristics fail. We find that models trained on MNLI, including BERT, a state-of-the-art model, perform very poorly on HANS, suggesting that they have indeed adopted these heuristics. We conclude that there is substantial room for improvement in NLI systems, and that the HANS dataset can motivate and measure progress in this area", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-07", "2020-07", "2021-08"], "last_mentioned": "2021-08-10T17:00:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1902.10935", "slug": "878463c2baa07e1a", "title": "Lower Bounds for Multiplication via Network Coding", "authors": ["Peyman Afshani", "Casper Benjamin Freksen", "Lior Kamma", "Kasper Green Larsen"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1902.10935", "doi": null, "topics": ["Algorithms"], "description": "Multiplication is one of the most fundamental computational problems, yet its true complexity remains elusive.", "description_method": "abstract-excerpt", "abstract": "Multiplication is one of the most fundamental computational problems, yet its true complexity remains elusive. The best known upper bound, by Fürer, shows that two $n$-bit numbers can be multiplied via a boolean circuit of size $O(n \\lg n \\cdot 4^{\\lg^*n})$, where $\\lg^*n$ is the very slowly growing iterated logarithm. In this work, we prove that if a central conjecture in the area of network coding is true, then any constant degree boolean circuit for multiplication must have size $Ω(n \\lg n)$, thus almost completely settling the complexity of multiplication circuits. We additionally revisit classic conjectures in circuit complexity, due to Valiant, and show that the network coding conjecture also implies one of Valiant's conjectures.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-03", "2019-04", "2020-07"], "last_mentioned": "2020-07-22T19:26:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1903.08570", "slug": "32137ea998b12a5b", "title": "Accurate and Infinite Prime Prediction from Novel Quasi-Prime Analytical Methodology", "authors": ["Robert E. Grant", "Talal Ghannam"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1903.08570", "doi": null, "topics": ["Mathematics", "Distributed systems"], "description": "It is known that prime numbers occupy specific geometrical patterns or moduli when numbers from one to infinity are distributed around polygons having sides that are integer multiple of number 6.", "description_method": "abstract-excerpt", "abstract": "It is known that prime numbers occupy specific geometrical patterns or moduli when numbers from one to infinity are distributed around polygons having sides that are integer multiple of number 6. In this paper, we will show that not only prime numbers occupy these moduli, but non-prime numbers sharing these same moduli have unique prime-ness properties. When utilizing digital root methodologies, these non-prime numbers provide a novel method to accurately identify prime numbers and prime factors without trial division or probabilistic-based methods. We will also show that the icositetragon (24-sided regular polygon) is a unique polygon pertaining to prime numbers and their ultimate incidence and distribution.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-08", "2019-11", "2024-07"], "last_mentioned": "2024-07-03T09:16:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1905.05950", "slug": "670ad9160e53e364", "title": "BERT Rediscovers the Classical NLP Pipeline", "authors": ["Ian Tenney", "Dipanjan Das", "Ellie Pavlick"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1905.05950", "doi": null, "topics": ["Language models"], "description": "Pre-trained text encoders have rapidly advanced the state of the art on many NLP tasks.", "description_method": "abstract-excerpt", "abstract": "Pre-trained text encoders have rapidly advanced the state of the art on many NLP tasks. We focus on one such model, BERT, and aim to quantify where linguistic information is captured within the network. We find that the model represents the steps of the traditional NLP pipeline in an interpretable and localizable way, and that the regions responsible for each step appear in the expected sequence: POS tagging, parsing, NER, semantic roles, then coreference. Qualitative analysis reveals that the model can and often does adjust this pipeline dynamically, revising lower-level decisions on the basis of disambiguating information from higher-level representations.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-03", "2023-02", "2023-05"], "last_mentioned": "2023-05-25T20:27:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1906.01563", "slug": "159b051bb58fc957", "title": "Hamiltonian Neural Networks", "authors": ["Sam Greydanus", "Misko Dzamba", "Jason Yosinski"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1906.01563", "doi": null, "topics": ["Machine learning", "Psychology", "Physics"], "description": "Even though neural networks enjoy widespread use, they still struggle to learn the basic laws of physics.", "description_method": "abstract-excerpt", "abstract": "Even though neural networks enjoy widespread use, they still struggle to learn the basic laws of physics. How might we endow them with better inductive biases? In this paper, we draw inspiration from Hamiltonian mechanics to train models that learn and respect exact conservation laws in an unsupervised manner. We evaluate our models on problems where conservation of energy is important, including the two-body problem and pixel observations of a pendulum. Our model trains faster and generalizes better than a regular neural network. An interesting side effect is that our model is perfectly reversible in time.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-02", "2020-06", "2020-07"], "last_mentioned": "2020-07-21T18:57:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1907.05972", "slug": "fc64d5e1619e925f", "title": "Motion Sensor-based Privacy Attack on Smartphones", "authors": ["S Abhishek Anand", "Chen Wang", "Jian Liu", "Nitesh Saxena", "Yingying Chen"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1907.05972", "doi": null, "topics": ["Security & cryptography", "Psychology", "Machine learning"], "description": "In this paper, we build a speech privacy attack that exploits speech reverberations generated from a smartphone's in-built loudspeaker captured via a zero-permission motion sensor (accelerometer).", "description_method": "abstract-excerpt", "abstract": "In this paper, we build a speech privacy attack that exploits speech reverberations generated from a smartphone's in-built loudspeaker captured via a zero-permission motion sensor (accelerometer). We design our attack Spearphone2, and demonstrate that speech reverberations from inbuilt loudspeakers, at an appropriate loudness, can impact the accelerometer, leaking sensitive information about the speech. In particular, we show that by exploiting the affected accelerometer readings and carefully selecting feature sets along with off-the-shelf machine learning techniques, Spearphone can successfully perform gender classification (accuracy over 90%) and speaker identification (accuracy over 80%) for any audio/video playback on the smartphone. Our results with testing the attack on a voice call and voice assistant response were also encouraging, showcasing the impact of the proposed attack. In addition, we perform speech recognition and speech reconstruction to extract more information about the eavesdropped speech to an extent. Our work brings to light a fundamental design vulnerability in many currently-deployed smartphones, which may put people's speech privacy at risk while using the smartphone in the loudspeaker mode during phone calls, media playback or voice assistant interactions.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-09", "2021-11", "2022-11"], "last_mentioned": "2022-11-06T02:10:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1907.08089", "slug": "ff45546c472ab97b", "title": "Comparing the Effects of DNS, DoT, and DoH on Web Performance", "authors": ["Austin Hounsel", "Kevin Borgolte", "Paul Schmitt", "Jordan Holland", "Nick Feamster"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1907.08089", "doi": "10.1145/3366423.3380139", "topics": ["Security & cryptography", "Networks", "Databases"], "description": "Nearly every service on the Internet relies on the Domain Name System (DNS), which translates a human-readable name to an IP address before two endpoints can communicate.", "description_method": "abstract-excerpt", "abstract": "Nearly every service on the Internet relies on the Domain Name System (DNS), which translates a human-readable name to an IP address before two endpoints can communicate. Today, DNS traffic is unencrypted, leaving users vulnerable to eavesdropping and tampering. Past work has demonstrated that DNS queries can reveal a user's browsing history and even what smart devices they are using at home. In response to these privacy concerns, two new protocols have been proposed: DNS-over-HTTPS (DoH) and DNS-over-TLS (DoT). Instead of sending DNS queries and responses in the clear, DoH and DoT establish encrypted connections between users and resolvers. By doing so, these protocols provide privacy and security guarantees that traditional DNS (Do53) lacks. In this paper, we measure the effect of Do53, DoT, and DoH on query response times and page load times from five global vantage points. We find that although DoH and DoT response times are generally higher than Do53, both protocols can perform better than Do53 in terms of page load times. However, as throughput decreases and substantial packet loss and latency are introduced, web pages load fastest with Do53. Additionally, web pages successfully load more often with Do53 and DoT than DoH. Based on these results, we provide several recommendations to improve DNS performance, such as opportunistic partial responses and wire format caching.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-09", "2025-12", "2026-03"], "last_mentioned": "2026-03-16T21:47:01Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1908.09203", "slug": "6d793720cf33b57e", "title": "Release Strategies and the Social Impacts of Language Models", "authors": ["Irene Solaiman", "Miles Brundage", "Jack Clark", "Amanda Askell", "Ariel Herbert-Voss", "Jeff Wu", "Alec Radford", "Gretchen Krueger", "Jong Wook Kim", "Sarah Kreps", "Miles McCain", "Alex Newhouse", "Jason Blazakis", "Kris McGuffie", "Jasmine Wang"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1908.09203", "doi": null, "topics": ["Language models", "Science & research", "Psychology"], "description": "Large language models have a range of beneficial uses: they can assist in prose, poetry, and programming; analyze dataset biases; and more.", "description_method": "abstract-excerpt", "abstract": "Large language models have a range of beneficial uses: they can assist in prose, poetry, and programming; analyze dataset biases; and more. However, their flexibility and generative capabilities also raise misuse concerns. This report discusses OpenAI's work related to the release of its GPT-2 language model. It discusses staged release, which allows time between model releases to conduct risk and benefit analyses as model sizes increased. It also discusses ongoing partnership-based research and provides recommendations for better coordination and responsible publication in AI.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-04", "2023-11", "2024-05"], "last_mentioned": "2024-05-20T04:28:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1910.09433", "slug": "9bdaebff7661f889", "title": "KuroNet: Pre-Modern Japanese Kuzushiji Character Recognition with Deep Learning", "authors": ["Tarin Clanuwat", "Alex Lamb", "Asanobu Kitamoto"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1910.09433", "doi": null, "topics": ["Psychology", "Machine learning", "Mathematics"], "description": "Kuzushiji, a cursive writing style, had been used in Japan for over a thousand years starting from the 8th century.", "description_method": "abstract-excerpt", "abstract": "Kuzushiji, a cursive writing style, had been used in Japan for over a thousand years starting from the 8th century. Over 3 millions books on a diverse array of topics, such as literature, science, mathematics and even cooking are preserved. However, following a change to the Japanese writing system in 1900, Kuzushiji has not been included in regular school curricula. Therefore, most Japanese natives nowadays cannot read books written or printed just 150 years ago. Museums and libraries have invested a great deal of effort into creating digital copies of these historical documents as a safeguard against fires, earthquakes and tsunamis. The result has been datasets with hundreds of millions of photographs of historical documents which can only be read by a small number of specially trained experts. Thus there has been a great deal of interest in using Machine Learning to automatically recognize these historical texts and transcribe them into modern Japanese characters. Nevertheless, several challenges in Kuzushiji recognition have made the performance of existing systems extremely poor. To tackle these challenges, we propose KuroNet, a new end-to-end model which jointly recognizes an entire page of text by using a residual U-Net architecture which predicts the location and identity of all characters given a page of text (without any pre-processing). This allows the model to handle long range context, large vocabularies, and non-standardized character layouts. We demonstrate that our system is able to successfully recognize a large fraction of pre-modern Japanese documents, but also explore areas where our system is limited and suggest directions for future work.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-11", "2025-12", "2026-04"], "last_mentioned": "2026-04-26T10:23:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:1910.09534", "slug": "32fb4da98730eccf", "title": "Leveraging Secondary Storage to Simulate Deep 54-qubit Sycamore Circuits", "authors": ["Edwin Pednault", "John A. Gunnels", "Giacomo Nannicini", "Lior Horesh", "Robert Wisnieff"], "year": 2019, "venue": "arXiv", "url": "https://arxiv.org/abs/1910.09534", "doi": null, "topics": ["Physics", "Economics", "Algorithms"], "description": "In a recent paper, we showed that secondary storage can extend the range of quantum circuits that can be practically simulated with classical algorithms.", "description_method": "abstract-excerpt", "abstract": "In a recent paper, we showed that secondary storage can extend the range of quantum circuits that can be practically simulated with classical algorithms. Here we refine those techniques and apply them to the simulation of Sycamore circuits with 53 and 54 qubits, with the entanglement pattern ABCDCDAB that has proven difficult to classically simulate with other approaches. Our analysis shows that on the Summit supercomputer at Oak Ridge National Laboratories, such circuits can be simulated with high fidelity to arbitrary depth in a matter of days, outputting all the amplitudes.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-10", "2020-12", "2021-07"], "last_mentioned": "2021-07-07T21:27:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2001.07488", "slug": "855b5474d1438f22", "title": "Profunctor Optics, a Categorical Update", "authors": ["Bryce Clarke", "Derek Elkins", "Jeremy Gibbons", "Fosco Loregian", "Bartosz Milewski", "Emily Pillmore", "Mario Román"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2001.07488", "doi": "10.32408/compositionality-6-1", "topics": ["Programming languages"], "description": "Optics are bidirectional data accessors that capture data transformation patterns such as accessing subfields or iterating over containers.", "description_method": "abstract-excerpt", "abstract": "Optics are bidirectional data accessors that capture data transformation patterns such as accessing subfields or iterating over containers. Profunctor optics are a particular choice of representation supporting modularity, meaning that we can construct accessors for complex structures by combining simpler ones. Profunctor optics have previously been studied only in an unenriched and non-mixed setting, in which both directions of access are modelled in the same category. However, functional programming languages are arguably better described by enriched categories; and we have found that some structures in the literature are actually mixed optics, with access directions modelled in different categories. Our work generalizes a classic result by Pastro and Street on Tambara theory and uses it to describe mixed V-enriched profunctor optics and to endow them with V-category structure. We provide some original families of optics and derivations, including an elementary one for traversals. Finally, we discuss a Haskell implementation.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-02", "2022-06", "2022-11"], "last_mentioned": "2022-11-30T19:08:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2001.09977", "slug": "048eaafa9c039901", "title": "Towards a Human-like Open-Domain Chatbot", "authors": ["Daniel Adiwardana", "Minh-Thang Luong", "David R. So", "Jamie Hall", "Noah Fiedel", "Romal Thoppilan", "Zi Yang", "Apoorv Kulshreshtha", "Gaurav Nemade", "Yifeng Lu", "Quoc V. Le"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2001.09977", "doi": null, "topics": ["Machine learning"], "description": "We present Meena, a multi-turn open-domain chatbot trained end-to-end on data mined and filtered from public domain social media conversations.", "description_method": "abstract-excerpt", "abstract": "We present Meena, a multi-turn open-domain chatbot trained end-to-end on data mined and filtered from public domain social media conversations. This 2.6B parameter neural network is simply trained to minimize perplexity of the next token. We also propose a human evaluation metric called Sensibleness and Specificity Average (SSA), which captures key elements of a human-like multi-turn conversation. Our experiments show strong correlation between perplexity and SSA. The fact that the best perplexity end-to-end trained Meena scores high on SSA (72% on multi-turn evaluation) suggests that a human-level SSA of 86% is potentially within reach if we can better optimize perplexity. Additionally, the full version of Meena (with a filtering mechanism and tuned decoding) scores 79% SSA, 23% higher in absolute SSA than the existing chatbots we evaluated.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-04", "2022-04", "2023-03"], "last_mentioned": "2023-03-09T01:39:19Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2002.11871", "slug": "cd37db517cae4520", "title": "Direct Multipixel Imaging and Spectroscopy of an Exoplanet with a Solar Gravity Lens Mission", "authors": ["Slava G. Turyshev", "Michael Shao", "Viktor T. Toth", "Louis D. Friedman", "Leon Alkalai", "Dmitri Mawet", "Janice Shen", "Mark R. Swain", "Hanying Zhou", "Henry Helvajian", "Tom Heinsheimer", "Siegfried Janson", "Zigmond Leszczynski", "John McVey", "Darren Garber", "Artur Davoyan", "Seth Redfield", "Jared R. Males"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2002.11871", "doi": null, "topics": ["Physics", "Algorithms"], "description": "We examined the solar gravitational lens (SGL) as the means to produce direct high-resolution, multipixel images of exoplanets.", "description_method": "abstract-excerpt", "abstract": "We examined the solar gravitational lens (SGL) as the means to produce direct high-resolution, multipixel images of exoplanets. The properties of the SGL are remarkable: it offers maximum light amplification of ~1e11 and angular resolution of ~1e-10 arcsec. A probe with a 1-m telescope in the SGL focal region can image an exoplanet at 30 pc with 10-kilometer resolution on its surface, sufficient to observe seasonal changes, oceans, continents, surface topography. We reached and exceeded all objectives set for our study: We developed a new wave-optical approach to study the imaging of exoplanets while treating them as extended, resolved, faint sources at large but finite distances. We properly accounted for the solar corona brightness. We developed deconvolution algorithms and demonstrated the feasibility of high-quality image reconstruction under realistic conditions. We have proven that multipixel imaging and spectroscopy of exoplanets with the SGL are feasible. We have developed a new mission concept that delivers an array of optical telescopes to the SGL focal region relying on three innovations: i) a new way to enable direct exoplanet imaging, ii) use of smallsats solar sails fast transit through the solar system and beyond, iii) an open architecture to take advantage of swarm technology. This approach enables entirely new missions, providing a great leap in capabilities for NASA and the greater aerospace community. Our results are encouraging as they lead to a realistic design for a mission that will be able to make direct resolved images of exoplanets in our stellar neighborhood. It could allow exploration of exoplanets relying on the SGL capabilities decades, if not centuries, earlier than possible with other extant technologies. The architecture and mission concepts for a mission to the strong interference region of the SGL are promising and should be explored further.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-05", "2024-10", "2025-01"], "last_mentioned": "2025-01-29T16:43:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2003.08505", "slug": "721f993e1543c9fb", "title": "A Metric Learning Reality Check", "authors": ["Kevin Musgrave", "Serge Belongie", "Ser-Nam Lim"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2003.08505", "doi": null, "topics": ["Other research"], "description": "Deep metric learning papers from the past four years have consistently claimed great advances in accuracy, often more than doubling the performance of decade-old methods.", "description_method": "abstract-excerpt", "abstract": "Deep metric learning papers from the past four years have consistently claimed great advances in accuracy, often more than doubling the performance of decade-old methods. In this paper, we take a closer look at the field to see if this is actually true. We find flaws in the experimental methodology of numerous metric learning papers, and show that the actual improvements over time have been marginal at best.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-05", "2020-06", "2022-12"], "last_mentioned": "2022-12-01T22:38:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2005.09535", "slug": "b1fcb0cbf2746a9e", "title": "Backstabber's Knife Collection: A Review of Open Source Software Supply Chain Attacks", "authors": ["Marc Ohm", "Henrik Plate", "Arnold Sykosch", "Michael Meier"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2005.09535", "doi": null, "topics": ["Security & cryptography", "Distributed systems"], "description": "A software supply chain attack is characterized by the injection of malicious code into a software package in order to compromise dependent systems further down the chain.", "description_method": "abstract-excerpt", "abstract": "A software supply chain attack is characterized by the injection of malicious code into a software package in order to compromise dependent systems further down the chain. Recent years saw a number of supply chain attacks that leverage the increasing use of open source during software development, which is facilitated by dependency managers that automatically resolve, download and install hundreds of open source packages throughout the software life cycle. This paper presents a dataset of 174 malicious software packages that were used in real-world attacks on open source software supply chains, and which were distributed via the popular package repositories npm, PyPI, and RubyGems. Those packages, dating from November 2015 to November 2019, were manually collected and analyzed. The paper also presents two general attack trees to provide a structured overview about techniques to inject malicious code into the dependency tree of downstream users, and to execute such code at different times and under different conditions. This work is meant to facilitate the future development of preventive and detective safeguards by open source and research communities.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-05", "2021-04", "2025-09"], "last_mentioned": "2025-09-09T16:44:10Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2006.14805", "slug": "5389212e08fa8d25", "title": "Unexpected Circular Radio Objects at High Galactic Latitude", "authors": ["Ray P. Norris", "Huib T. Intema", "Anna D. Kapinska", "Baerbel S. Koribalski", "Emil Lenc", "L. Rudnick", "Rami Alsaberi", "Craig Anderson", "G. E. Anderson", "E. Crawford", "Roland Crocker", "Jayanne English", "Miroslav D. Filipovic", "Andrew M. Hopkins", "Natasha Hurley-Walker", "Susumu Inoue", "Kieran Luken", "Peter Macgregor", "Pero Manojlovic", "Josh Marvil", "Andrew N. O'Brien", "Wasim Raja", "Devika Shobhana", "Tiziana Venturi", "Jordan D. Collier", "Catherine Hale", "Aidan Hotan", "Vanessa Moss", "Matthew Whiting"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2006.14805", "doi": "10.1017/pasa.2020.52", "topics": ["Other research"], "description": "We have found a class of circular radio objects in the Evolutionary Map of the Universe Pilot Survey, using the Australian Square Kilometre Array Pathfinder telescope.", "description_method": "abstract-excerpt", "abstract": "We have found a class of circular radio objects in the Evolutionary Map of the Universe Pilot Survey, using the Australian Square Kilometre Array Pathfinder telescope. The objects appear in radio images as circular edge-brightened discs, about one arcmin diameter, that are unlike other objects previously reported in the literature. We explore several possible mechanisms that might cause these objects, but none seems to be a compelling explanation.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-07", "2020-12", "2022-03"], "last_mentioned": "2022-03-28T20:30:09Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2006.16668", "slug": "5b5e5fde2c9e3bc4", "title": "GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding", "authors": ["Dmitry Lepikhin", "HyoukJoong Lee", "Yuanzhong Xu", "Dehao Chen", "Orhan Firat", "Yanping Huang", "Maxim Krikun", "Noam Shazeer", "Zhifeng Chen"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2006.16668", "doi": null, "topics": ["Programming languages", "Machine learning", "Language models"], "description": "Neural network scaling has been critical for improving the model quality in many real-world machine learning applications with vast amounts of training data and compute.", "description_method": "abstract-excerpt", "abstract": "Neural network scaling has been critical for improving the model quality in many real-world machine learning applications with vast amounts of training data and compute. Although this trend of scaling is affirmed to be a sure-fire approach for better model quality, there are challenges on the path such as the computation cost, ease of programming, and efficient implementation on parallel devices. GShard is a module composed of a set of lightweight annotation APIs and an extension to the XLA compiler. It provides an elegant way to express a wide range of parallel computation patterns with minimal changes to the existing model code. GShard enabled us to scale up multilingual neural machine translation Transformer model with Sparsely-Gated Mixture-of-Experts beyond 600 billion parameters using automatic sharding. We demonstrate that such a giant model can efficiently be trained on 2048 TPU v3 accelerators in 4 days to achieve far superior quality for translation from 100 languages to English compared to the prior art.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-07", "2021-12", "2023-12"], "last_mentioned": "2023-12-06T21:43:42Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2007.11112", "slug": "d9e52c01ac74c034", "title": "DBOS: A Proposal for a Data-Centric Operating System", "authors": ["Michael Cafarella", "David DeWitt", "Vijay Gadepally", "Jeremy Kepner", "Christos Kozyrakis", "Tim Kraska", "Michael Stonebraker", "Matei Zaharia"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2007.11112", "doi": null, "topics": ["Computer systems", "Security & cryptography", "Databases"], "description": "Current operating systems are complex systems that were designed before today's computing environments.", "description_method": "abstract-excerpt", "abstract": "Current operating systems are complex systems that were designed before today's computing environments. This makes it difficult for them to meet the scalability, heterogeneity, availability, and security challenges in current cloud and parallel computing environments. To address these problems, we propose a radically new OS design based on data-centric architecture: all operating system state should be represented uniformly as database tables, and operations on this state should be made via queries from otherwise stateless tasks. This design makes it easy to scale and evolve the OS without whole-system refactoring, inspect and debug system state, upgrade components without downtime, manage decisions using machine learning, and implement sophisticated security features. We discuss how a database OS (DBOS) can improve the programmability and performance of many of today's most important applications and propose a plan for the development of a DBOS proof of concept.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-12", "2022-09", "2025-10"], "last_mentioned": "2025-10-03T10:09:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2008.06030", "slug": "1a49d6a2dd837064", "title": "On the design of text editors", "authors": ["Nicolas P. Rougier"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2008.06030", "doi": null, "topics": ["Other research"], "description": "Text editors are written by and for developers.", "description_method": "abstract-excerpt", "abstract": "Text editors are written by and for developers. They come with a large set of default and implicit choices in terms of layout, typography, colorization and interaction that hardly change from one editor to the other. It is not clear if these implicit choices derive from the ignorance of alternatives or if they derive from developers' habits, reproducing what they are used to. The goal of this article is to characterize these implicit choices and to illustrate what are some alternatives without prescribing one or the other.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-03", "2023-07", "2025-10"], "last_mentioned": "2025-10-28T23:26:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2009.01325", "slug": "787c1c647d718338", "title": "Learning to summarize from human feedback", "authors": ["Nisan Stiennon", "Long Ouyang", "Jeff Wu", "Daniel M. Ziegler", "Ryan Lowe", "Chelsea Voss", "Alec Radford", "Dario Amodei", "Paul Christiano"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2009.01325", "doi": null, "topics": ["Machine learning", "Psychology", "Language models"], "description": "As language models become more powerful, training and evaluation are increasingly bottlenecked by the data and metrics used for a particular task.", "description_method": "abstract-excerpt", "abstract": "As language models become more powerful, training and evaluation are increasingly bottlenecked by the data and metrics used for a particular task. For example, summarization models are often trained to predict human reference summaries and evaluated using ROUGE, but both of these metrics are rough proxies for what we really care about -- summary quality. In this work, we show that it is possible to significantly improve summary quality by training a model to optimize for human preferences. We collect a large, high-quality dataset of human comparisons between summaries, train a model to predict the human-preferred summary, and use that model as a reward function to fine-tune a summarization policy using reinforcement learning. We apply our method to a version of the TL;DR dataset of Reddit posts and find that our models significantly outperform both human reference summaries and much larger models fine-tuned with supervised learning alone. Our models also transfer to CNN/DM news articles, producing summaries nearly as good as the human reference without any news-specific fine-tuning. We conduct extensive analyses to understand our human feedback dataset and fine-tuned models We establish that our reward model generalizes to new datasets, and that optimizing our reward model results in better summaries than optimizing ROUGE according to humans. We hope the evidence from our paper motivates machine learning researchers to pay closer attention to how their training loss affects the model behavior they actually want.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-01", "2023-03", "2025-08"], "last_mentioned": "2025-08-11T01:11:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2010.08895", "slug": "2de410640add7aa6", "title": "Fourier Neural Operator for Parametric Partial Differential Equations", "authors": ["Zongyi Li", "Nikola Kovachki", "Kamyar Azizzadenesheli", "Burigede Liu", "Kaushik Bhattacharya", "Andrew Stuart", "Anima Anandkumar"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2010.08895", "doi": null, "topics": ["Machine learning", "Computer systems"], "description": "The classical development of neural networks has primarily focused on learning mappings between finite-dimensional Euclidean spaces.", "description_method": "abstract-excerpt", "abstract": "The classical development of neural networks has primarily focused on learning mappings between finite-dimensional Euclidean spaces. Recently, this has been generalized to neural operators that learn mappings between function spaces. For partial differential equations (PDEs), neural operators directly learn the mapping from any functional parametric dependence to the solution. Thus, they learn an entire family of PDEs, in contrast to classical methods which solve one instance of the equation. In this work, we formulate a new neural operator by parameterizing the integral kernel directly in Fourier space, allowing for an expressive and efficient architecture. We perform experiments on Burgers' equation, Darcy flow, and Navier-Stokes equation. The Fourier neural operator is the first ML-based method to successfully model turbulent flows with zero-shot super-resolution. It is up to three orders of magnitude faster compared to traditional PDE solvers. Additionally, it achieves superior accuracy compared to previous learning-based solvers under fixed resolution.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-11", "2024-03", "2025-02"], "last_mentioned": "2025-02-26T15:53:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2011.03395", "slug": "bd4504e9840364d5", "title": "Underspecification Presents Challenges for Credibility in Modern Machine Learning", "authors": ["Alexander D'Amour", "Katherine Heller", "Dan Moldovan", "Ben Adlam", "Babak Alipanahi", "Alex Beutel", "Christina Chen", "Jonathan Deaton", "Jacob Eisenstein", "Matthew D. Hoffman", "Farhad Hormozdiari", "Neil Houlsby", "Shaobo Hou", "Ghassen Jerfel", "Alan Karthikesalingam", "Mario Lucic", "Yian Ma", "Cory McLean", "Diana Mincu", "Akinori Mitani", "Andrea Montanari", "Zachary Nado", "Vivek Natarajan", "Christopher Nielson", "Thomas F. Osborne", "Rajiv Raman", "Kim Ramasamy", "Rory Sayres", "Jessica Schrouff", "Martin Seneviratne", "Shannon Sequeira", "Harini Suresh", "Victor Veitch", "Max Vladymyrov", "Xuezhi Wang", "Kellie Webster", "Steve Yadlowsky", "Taedong Yun", "Xiaohua Zhai", "D. Sculley"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2011.03395", "doi": null, "topics": ["Biology & medicine", "Machine learning", "Psychology"], "description": "ML models often exhibit unexpectedly poor behavior when they are deployed in real-world domains.", "description_method": "abstract-excerpt", "abstract": "ML models often exhibit unexpectedly poor behavior when they are deployed in real-world domains. We identify underspecification as a key reason for these failures. An ML pipeline is underspecified when it can return many predictors with equivalently strong held-out performance in the training domain. Underspecification is common in modern ML pipelines, such as those based on deep learning. Predictors returned by underspecified pipelines are often treated as equivalent based on their training domain performance, but we show here that such predictors can behave very differently in deployment domains. This ambiguity can lead to instability and poor model behavior in practice, and is a distinct failure mode from previously identified issues arising from structural mismatch between training and deployment domains. We show that this problem appears in a wide variety of practical ML pipelines, using examples from computer vision, medical imaging, natural language processing, clinical risk prediction based on electronic health records, and medical genomics. Our results show the need to explicitly account for underspecification in modeling pipelines that are intended for real-world deployment in any domain.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-11", "2021-05", "2023-03"], "last_mentioned": "2023-03-24T01:49:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2011.13456", "slug": "db4c9b7eea9d1472", "title": "Score-Based Generative Modeling through Stochastic Differential Equations", "authors": ["Yang Song", "Jascha Sohl-Dickstein", "Diederik P. Kingma", "Abhishek Kumar", "Stefano Ermon", "Ben Poole"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2011.13456", "doi": null, "topics": ["Machine learning", "Statistics & probability", "Physics"], "description": "Creating noise from data is easy; creating data from noise is generative modeling.", "description_method": "abstract-excerpt", "abstract": "Creating noise from data is easy; creating data from noise is generative modeling. We present a stochastic differential equation (SDE) that smoothly transforms a complex data distribution to a known prior distribution by slowly injecting noise, and a corresponding reverse-time SDE that transforms the prior distribution back into the data distribution by slowly removing the noise. Crucially, the reverse-time SDE depends only on the time-dependent gradient field (\\aka, score) of the perturbed data distribution. By leveraging advances in score-based generative modeling, we can accurately estimate these scores with neural networks, and use numerical SDE solvers to generate samples. We show that this framework encapsulates previous approaches in score-based generative modeling and diffusion probabilistic modeling, allowing for new sampling procedures and new modeling capabilities. In particular, we introduce a predictor-corrector framework to correct errors in the evolution of the discretized reverse-time SDE. We also derive an equivalent neural ODE that samples from the same distribution as the SDE, but additionally enables exact likelihood computation, and improved sampling efficiency. In addition, we provide a new way to solve inverse problems with score-based models, as demonstrated with experiments on class-conditional generation, image inpainting, and colorization. Combined with multiple architectural improvements, we achieve record-breaking performance for unconditional image generation on CIFAR-10 with an Inception score of 9.89 and FID of 2.20, a competitive likelihood of 2.99 bits/dim, and demonstrate high fidelity generation of 1024 x 1024 images for the first time from a score-based generative model.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-05", "2021-10", "2022-12"], "last_mentioned": "2022-12-11T19:22:44Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2012.07805", "slug": "df2c82e06fa68b3d", "title": "Extracting Training Data from Large Language Models", "authors": ["Nicholas Carlini", "Florian Tramer", "Eric Wallace", "Matthew Jagielski", "Ariel Herbert-Voss", "Katherine Lee", "Adam Roberts", "Tom Brown", "Dawn Song", "Ulfar Erlingsson", "Alina Oprea", "Colin Raffel"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2012.07805", "doi": null, "topics": ["Language models", "Security & cryptography", "Networks"], "description": "It has become common to publish large (billion parameter) language models that have been trained on private datasets.", "description_method": "abstract-excerpt", "abstract": "It has become common to publish large (billion parameter) language models that have been trained on private datasets. This paper demonstrates that in such settings, an adversary can perform a training data extraction attack to recover individual training examples by querying the language model. We demonstrate our attack on GPT-2, a language model trained on scrapes of the public Internet, and are able to extract hundreds of verbatim text sequences from the model's training data. These extracted examples include (public) personally identifiable information (names, phone numbers, and email addresses), IRC conversations, code, and 128-bit UUIDs. Our attack is possible even though each of the above sequences are included in just one document in the training data. We comprehensively evaluate our extraction attack to understand the factors that contribute to its success. Worryingly, we find that larger models are more vulnerable than smaller models. We conclude by drawing lessons and discussing possible safeguards for training large language models.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-02", "2021-07", "2023-04"], "last_mentioned": "2023-04-20T08:33:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2012.12501", "slug": "09f4b0acd47a626f", "title": "Learned Indexes for a Google-scale Disk-based Database", "authors": ["Hussam Abu-Libdeh", "Deniz Altınbüken", "Alex Beutel", "Ed H. Chi", "Lyric Doshi", "Tim Kraska", " Xiaozhou", " Li", "Andy Ly", "Christopher Olston"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2012.12501", "doi": null, "topics": ["Databases", "Distributed systems"], "description": "There is great excitement about learned index structures, but understandable skepticism about the practicality of a new method uprooting decades of research on B-Trees.", "description_method": "abstract-excerpt", "abstract": "There is great excitement about learned index structures, but understandable skepticism about the practicality of a new method uprooting decades of research on B-Trees. In this paper, we work to remove some of that uncertainty by demonstrating how a learned index can be integrated in a distributed, disk-based database system: Google's Bigtable. We detail several design decisions we made to integrate learned indexes in Bigtable. Our results show that integrating learned index significantly improves the end-to-end read latency and throughput for Bigtable.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-02", "2022-07", "2025-06"], "last_mentioned": "2025-06-25T17:07:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2012.13966", "slug": "a3f3793204498b1a", "title": "ZX-calculus for the working quantum computer scientist", "authors": ["John van de Wetering"], "year": 2020, "venue": "arXiv", "url": "https://arxiv.org/abs/2012.13966", "doi": null, "topics": ["Mathematics", "Physics"], "description": "The ZX-calculus is a graphical language for reasoning about quantum computation that has recently seen an increased usage in a variety of areas such as quantum circuit optimisation, surface codes and lattice surgery, measurement-based quantum computation, and quantum foundations.", "description_method": "abstract-excerpt", "abstract": "The ZX-calculus is a graphical language for reasoning about quantum computation that has recently seen an increased usage in a variety of areas such as quantum circuit optimisation, surface codes and lattice surgery, measurement-based quantum computation, and quantum foundations. The first half of this review gives a gentle introduction to the ZX-calculus suitable for those familiar with the basics of quantum computing. The aim here is to make the reader comfortable enough with the ZX-calculus that they could use it in their daily work for small computations on quantum circuits and states. The latter sections give a condensed overview of the literature on the ZX-calculus. We discuss Clifford computation and graphically prove the Gottesman-Knill theorem, we discuss a recently introduced extension of the ZX-calculus that allows for convenient reasoning about Toffoli gates, and we discuss the recent completeness theorems for the ZX-calculus that show that, in principle, all reasoning about quantum computation can be done using ZX-diagrams. Additionally, we discuss the categorical and algebraic origins of the ZX-calculus and we discuss several extensions of the language which can represent mixed states, measurement, classical control and higher-dimensional qudits.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-06", "2024-06", "2024-12"], "last_mentioned": "2024-12-05T00:54:03Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2103.03874", "slug": "37fe079f3c66c69f", "title": "Measuring Mathematical Problem Solving With the MATH Dataset", "authors": ["Dan Hendrycks", "Collin Burns", "Saurav Kadavath", "Akul Arora", "Steven Basart", "Eric Tang", "Dawn Song", "Jacob Steinhardt"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2103.03874", "doi": null, "topics": ["Mathematics", "Language models", "Machine learning"], "description": "Many intellectual endeavors require mathematical problem solving, but this skill remains beyond the capabilities of computers.", "description_method": "abstract-excerpt", "abstract": "Many intellectual endeavors require mathematical problem solving, but this skill remains beyond the capabilities of computers. To measure this ability in machine learning models, we introduce MATH, a new dataset of 12,500 challenging competition mathematics problems. Each problem in MATH has a full step-by-step solution which can be used to teach models to generate answer derivations and explanations. To facilitate future research and increase accuracy on MATH, we also contribute a large auxiliary pretraining dataset which helps teach models the fundamentals of mathematics. Even though we are able to increase accuracy on MATH, our results show that accuracy remains relatively low, even with enormous Transformer models. Moreover, we find that simply increasing budgets and model parameter counts will be impractical for achieving strong mathematical reasoning if scaling trends continue. While scaling Transformers is automatically solving most other text-based tasks, scaling is not currently solving MATH. To have more traction on mathematical problem solving we will likely need new algorithmic advancements from the broader research community.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-03", "2023-10", "2025-01"], "last_mentioned": "2025-01-30T20:59:59Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2103.15808", "slug": "33c1e1f45392528c", "title": "CvT: Introducing Convolutions to Vision Transformers", "authors": ["Haiping Wu", "Bin Xiao", "Noel Codella", "Mengchen Liu", "Xiyang Dai", "Lu Yuan", "Lei Zhang"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2103.15808", "doi": null, "topics": ["Computer vision", "Language models", "Machine learning"], "description": "We present in this paper a new architecture, named Convolutional vision Transformer (CvT), that improves Vision Transformer (ViT) in performance and efficiency by introducing convolutions into ViT to yield the best of both designs.", "description_method": "abstract-excerpt", "abstract": "We present in this paper a new architecture, named Convolutional vision Transformer (CvT), that improves Vision Transformer (ViT) in performance and efficiency by introducing convolutions into ViT to yield the best of both designs. This is accomplished through two primary modifications: a hierarchy of Transformers containing a new convolutional token embedding, and a convolutional Transformer block leveraging a convolutional projection. These changes introduce desirable properties of convolutional neural networks (CNNs) to the ViT architecture (\\ie shift, scale, and distortion invariance) while maintaining the merits of Transformers (\\ie dynamic attention, global context, and better generalization). We validate CvT by conducting extensive experiments, showing that this approach achieves state-of-the-art performance over other Vision Transformers and ResNets on ImageNet-1k, with fewer parameters and lower FLOPs. In addition, performance gains are maintained when pretrained on larger datasets (\\eg ImageNet-22k) and fine-tuned to downstream tasks. Pre-trained on ImageNet-22k, our CvT-W24 obtains a top-1 accuracy of 87.7\\% on the ImageNet-1k val set. Finally, our results show that the positional encoding, a crucial component in existing Vision Transformers, can be safely removed in our model, simplifying the design for higher resolution vision tasks. Code will be released at \\url{https://github.com/leoxiaobin/CvT}.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-03", "2022-07", "2025-12"], "last_mentioned": "2025-12-23T15:09:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2104.05704", "slug": "f39655c5945329f4", "title": "Escaping the Big Data Paradigm with Compact Transformers", "authors": ["Ali Hassani", "Steven Walton", "Nikhil Shah", "Abulikemu Abuduweili", "Jiachen Li", "Humphrey Shi"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2104.05704", "doi": null, "topics": ["Language models", "Computer vision", "Science & research"], "description": "With the rise of Transformers as the standard for language processing, and their advancements in computer vision, there has been a corresponding growth in parameter size and amounts of training data.", "description_method": "abstract-excerpt", "abstract": "With the rise of Transformers as the standard for language processing, and their advancements in computer vision, there has been a corresponding growth in parameter size and amounts of training data. Many have come to believe that because of this, transformers are not suitable for small sets of data. This trend leads to concerns such as: limited availability of data in certain scientific domains and the exclusion of those with limited resource from research in the field. In this paper, we aim to present an approach for small-scale learning by introducing Compact Transformers. We show for the first time that with the right size, convolutional tokenization, transformers can avoid overfitting and outperform state-of-the-art CNNs on small datasets. Our models are flexible in terms of model size, and can have as little as 0.28M parameters while achieving competitive results. Our best model can reach 98% accuracy when training from scratch on CIFAR-10 with only 3.7M parameters, which is a significant improvement in data-efficiency over previous Transformer based models being over 10x smaller than other transformers and is 15% the size of ResNet50 while achieving similar performance. CCT also outperforms many modern CNN based approaches, and even some recent NAS-based approaches. Additionally, we obtain a new SOTA result on Flowers-102 with 99.76% top-1 accuracy, and improve upon the existing baseline on ImageNet (82.71% accuracy with 29% as many parameters as ViT), as well as NLP tasks. Our simple and compact design for transformers makes them more feasible to study for those with limited computing resources and/or dealing with small datasets, while extending existing research efforts in data efficient transformers. Our code and pre-trained models are publicly available at https://github.com/SHI-Labs/Compact-Transformers.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-03", "2023-12", "2025-07"], "last_mentioned": "2025-07-24T00:54:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2104.10350", "slug": "273d13153d2a967c", "title": "Carbon Emissions and Large Neural Network Training", "authors": ["David Patterson", "Joseph Gonzalez", "Quoc Le", "Chen Liang", "Lluis-Miquel Munguia", "Daniel Rothchild", "David So", "Maud Texier", "Jeff Dean"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2104.10350", "doi": null, "topics": ["Machine learning", "Language models", "Economics"], "description": "The computation demand for machine learning (ML) has grown rapidly recently, which comes with a number of costs.", "description_method": "abstract-excerpt", "abstract": "The computation demand for machine learning (ML) has grown rapidly recently, which comes with a number of costs. Estimating the energy cost helps measure its environmental impact and finding greener strategies, yet it is challenging without detailed information. We calculate the energy use and carbon footprint of several recent large models-T5, Meena, GShard, Switch Transformer, and GPT-3-and refine earlier estimates for the neural architecture search that found Evolved Transformer. We highlight the following opportunities to improve energy efficiency and CO2 equivalent emissions (CO2e): Large but sparsely activated DNNs can consume <1/10th the energy of large, dense DNNs without sacrificing accuracy despite using as many or even more parameters. Geographic location matters for ML workload scheduling since the fraction of carbon-free energy and resulting CO2e vary ~5X-10X, even within the same country and the same organization. We are now optimizing where and when large models are trained. Specific datacenter infrastructure matters, as Cloud datacenters can be ~1.4-2X more energy efficient than typical datacenters, and the ML-oriented accelerators inside them can be ~2-5X more effective than off-the-shelf systems. Remarkably, the choice of DNN, datacenter, and processor can reduce the carbon footprint up to ~100-1000X. These large factors also make retroactive estimates of energy cost difficult. To avoid miscalculations, we believe ML papers requiring large computational resources should make energy consumption and CO2e explicit when practical. We are working to be more transparent about energy use and CO2e in our future research. To help reduce the carbon footprint of ML, we believe energy usage and CO2e should be a key metric in evaluating models, and we are collaborating with MLPerf developers to include energy usage during training and inference in this industry standard benchmark.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-04", "2021-10", "2022-01"], "last_mentioned": "2022-01-29T17:32:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2105.05241", "slug": "86c1f892a61b3009", "title": "Addressing \"Documentation Debt\" in Machine Learning Research: A Retrospective Datasheet for BookCorpus", "authors": ["Jack Bandy", "Nicholas Vincent"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2105.05241", "doi": null, "topics": ["Machine learning", "Language models"], "description": "Recent literature has underscored the importance of dataset documentation work for machine learning, and part of this work involves addressing \"documentation debt\" for datasets that have been used widely but documented sparsely.", "description_method": "abstract-excerpt", "abstract": "Recent literature has underscored the importance of dataset documentation work for machine learning, and part of this work involves addressing \"documentation debt\" for datasets that have been used widely but documented sparsely. This paper aims to help address documentation debt for BookCorpus, a popular text dataset for training large language models. Notably, researchers have used BookCorpus to train OpenAI's GPT-N models and Google's BERT models, even though little to no documentation exists about the dataset's motivation, composition, collection process, etc. We offer a preliminary datasheet that provides key context and information about BookCorpus, highlighting several notable deficiencies. In particular, we find evidence that (1) BookCorpus likely violates copyright restrictions for many books, (2) BookCorpus contains thousands of duplicated books, and (3) BookCorpus exhibits significant skews in genre representation. We also find hints of other potential deficiencies that call for future research, including problematic content, potential skews in religious representation, and lopsided author contributions. While more work remains, this initial effort to provide a datasheet for BookCorpus adds to growing literature that urges more careful and systematic documentation for machine learning datasets.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-02", "2023-07", "2023-09"], "last_mentioned": "2023-09-20T13:50:25Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2106.05123", "slug": "e19fc74085e0c992", "title": "Pattern-defeating Quicksort", "authors": ["Orson R. L. Peters"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2106.05123", "doi": null, "topics": ["Other research"], "description": "A new solution for the Dutch national flag problem is proposed, requiring no three-way comparisons, which gives quicksort a proper worst-case runtime of $O(nk)$ for inputs with $k$ distinct elements.", "description_method": "abstract-excerpt", "abstract": "A new solution for the Dutch national flag problem is proposed, requiring no three-way comparisons, which gives quicksort a proper worst-case runtime of $O(nk)$ for inputs with $k$ distinct elements. This is used together with other known and novel techniques to construct a hybrid sort that is never significantly slower than regular quicksort while speeding up drastically for many input distributions.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-07", "2023-05", "2023-06"], "last_mentioned": "2023-06-13T12:09:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2106.10207", "slug": "f0a0a29f6e40a84b", "title": "Distributed Deep Learning in Open Collaborations", "authors": ["Michael Diskin", "Alexey Bukhtiyarov", "Max Ryabinin", "Lucile Saulnier", "Quentin Lhoest", "Anton Sinitsin", "Dmitry Popov", "Dmitry Pyrkin", "Maxim Kashirin", "Alexander Borzunov", "Albert Villanova del Moral", "Denis Mazur", "Ilia Kobelev", "Yacine Jernite", "Thomas Wolf", "Gennady Pekhimenko"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2106.10207", "doi": null, "topics": ["Economics", "Machine learning", "Language models"], "description": "Modern deep learning applications require increasingly more compute to train state-of-the-art models.", "description_method": "abstract-excerpt", "abstract": "Modern deep learning applications require increasingly more compute to train state-of-the-art models. To address this demand, large corporations and institutions use dedicated High-Performance Computing clusters, whose construction and maintenance are both environmentally costly and well beyond the budget of most organizations. As a result, some research directions become the exclusive domain of a few large industrial and even fewer academic actors. To alleviate this disparity, smaller groups may pool their computational resources and run collaborative experiments that benefit all participants. This paradigm, known as grid- or volunteer computing, has seen successful applications in numerous scientific areas. However, using this approach for machine learning is difficult due to high latency, asymmetric bandwidth, and several challenges unique to volunteer computing. In this work, we carefully analyze these constraints and propose a novel algorithmic framework designed specifically for collaborative training. We demonstrate the effectiveness of our approach for SwAV and ALBERT pretraining in realistic conditions and achieve performance comparable to traditional setups at a fraction of the cost. Finally, we provide a detailed report of successful collaborative language model pretraining with 40 participants.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-12", "2024-01", "2025-01"], "last_mentioned": "2025-01-20T23:08:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2107.03006", "slug": "bc1787889ef2752b", "title": "Structured Denoising Diffusion Models in Discrete State-Spaces", "authors": ["Jacob Austin", "Daniel D. Johnson", "Jonathan Ho", "Daniel Tarlow", "Rianne van den Berg"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2107.03006", "doi": null, "topics": ["Physics", "Machine learning", "Statistics & probability"], "description": "Denoising diffusion probabilistic models (DDPMs) (Ho et al.", "description_method": "abstract-excerpt", "abstract": "Denoising diffusion probabilistic models (DDPMs) (Ho et al. 2020) have shown impressive results on image and waveform generation in continuous state spaces. Here, we introduce Discrete Denoising Diffusion Probabilistic Models (D3PMs), diffusion-like generative models for discrete data that generalize the multinomial diffusion model of Hoogeboom et al. 2021, by going beyond corruption processes with uniform transition probabilities. This includes corruption with transition matrices that mimic Gaussian kernels in continuous space, matrices based on nearest neighbors in embedding space, and matrices that introduce absorbing states. The third allows us to draw a connection between diffusion models and autoregressive and mask-based generative models. We show that the choice of transition matrix is an important design decision that leads to improved results in image and text domains. We also introduce a new loss function that combines the variational lower bound with an auxiliary cross entropy loss. For text, this model class achieves strong results on character-level text generation while scaling to large vocabularies on LM1B. On the image dataset CIFAR-10, our models approach the sample quality and exceed the log-likelihood of the continuous-space DDPM model.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-12", "2025-05", "2025-10"], "last_mentioned": "2025-10-20T15:19:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2109.08203", "slug": "1aeaebd2645b11ee", "title": "Torch.manual_seed(3407) is all you need: On the influence of random seeds in deep learning architectures for computer vision", "authors": ["David Picard"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2109.08203", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "In this paper I investigate the effect of random seed selection on the accuracy when using popular deep learning architectures for computer vision.", "description_method": "abstract-excerpt", "abstract": "In this paper I investigate the effect of random seed selection on the accuracy when using popular deep learning architectures for computer vision. I scan a large amount of seeds (up to $10^4$) on CIFAR 10 and I also scan fewer seeds on Imagenet using pre-trained models to investigate large scale datasets. The conclusions are that even if the variance is not very large, it is surprisingly easy to find an outlier that performs much better or much worse than the average.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-05", "2024-07", "2025-01"], "last_mentioned": "2025-01-28T10:41:53Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2112.01488", "slug": "d00ad0ca30a13bf2", "title": "ColBERTv2: Effective and Efficient Retrieval via Lightweight Late Interaction", "authors": ["Keshav Santhanam", "Omar Khattab", "Jon Saad-Falcon", "Christopher Potts", "Matei Zaharia"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2112.01488", "doi": null, "topics": ["Other research"], "description": "Neural information retrieval (IR) has greatly advanced search and other knowledge-intensive language tasks.", "description_method": "abstract-excerpt", "abstract": "Neural information retrieval (IR) has greatly advanced search and other knowledge-intensive language tasks. While many neural IR methods encode queries and documents into single-vector representations, late interaction models produce multi-vector representations at the granularity of each token and decompose relevance modeling into scalable token-level computations. This decomposition has been shown to make late interaction more effective, but it inflates the space footprint of these models by an order of magnitude. In this work, we introduce ColBERTv2, a retriever that couples an aggressive residual compression mechanism with a denoised supervision strategy to simultaneously improve the quality and space footprint of late interaction. We evaluate ColBERTv2 across a wide range of benchmarks, establishing state-of-the-art quality within and outside the training domain while reducing the space footprint of late interaction models by 6--10$\\times$.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-11", "2024-04", "2025-12"], "last_mentioned": "2025-12-03T14:55:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2112.03178", "slug": "73d569ee4c6acc26", "title": "Student of Games: A unified learning algorithm for both perfect and imperfect information games", "authors": ["Martin Schmid", "Matej Moravcik", "Neil Burch", "Rudolf Kadlec", "Josh Davidson", "Kevin Waugh", "Nolan Bard", "Finbarr Timbers", "Marc Lanctot", "G. Zacharias Holland", "Elnaz Davoodi", "Alden Christianson", "Michael Bowling"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2112.03178", "doi": "10.1126/sciadv.adg3256", "topics": ["Algorithms", "Machine learning", "Psychology"], "description": "Games have a long history as benchmarks for progress in artificial intelligence.", "description_method": "abstract-excerpt", "abstract": "Games have a long history as benchmarks for progress in artificial intelligence. Approaches using search and learning produced strong performance across many perfect information games, and approaches using game-theoretic reasoning and learning demonstrated strong performance for specific imperfect information poker variants. We introduce Student of Games, a general-purpose algorithm that unifies previous approaches, combining guided search, self-play learning, and game-theoretic reasoning. Student of Games achieves strong empirical performance in large perfect and imperfect information games -- an important step towards truly general algorithms for arbitrary environments. We prove that Student of Games is sound, converging to perfect play as available computation and approximation capacity increases. Student of Games reaches strong performance in chess and Go, beats the strongest openly available agent in heads-up no-limit Texas hold'em poker, and defeats the state-of-the-art agent in Scotland Yard, an imperfect information game that illustrates the value of guided search, learning, and game-theoretic reasoning.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-01", "2022-07", "2023-12"], "last_mentioned": "2023-12-20T21:11:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2201.00652", "slug": "e28b56f0393fe876", "title": "Hyper-Fast Positive Energy Warp Drives", "authors": ["Erik W. Lentz"], "year": 2021, "venue": "arXiv", "url": "https://arxiv.org/abs/2201.00652", "doi": null, "topics": ["Physics", "Robotics"], "description": "Solitons in space-time capable of transporting time-like observers at superluminal speeds have long been tied to violations of the weak, strong, and dominant energy conditions of general relativity.", "description_method": "abstract-excerpt", "abstract": "Solitons in space-time capable of transporting time-like observers at superluminal speeds have long been tied to violations of the weak, strong, and dominant energy conditions of general relativity. This trend was recently broken by a new approach that identified soliton solutions capable of superluminal travel while being sourced by purely positive energy densities. This is the first example of hyper-fast solitons satisfying the weak energy condition, reopening the discussion of superluminal mechanisms rooted in conventional physics. This article summarizes the recent finding and its context in the literature. Remaining challenges to autonomous superluminal travel, such as the dominant energy condition, horizons, and the identification of a creation mechanism are also discussed.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-11", "2023-10", "2024-06"], "last_mentioned": "2024-06-23T13:01:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2201.01174", "slug": "ba4dd7d8b556f28f", "title": "Binary Fuse Filters: Fast and Smaller Than Xor Filters", "authors": ["Thomas Mueller Graf", "Daniel Lemire"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2201.01174", "doi": "10.1145/3510449", "topics": ["Databases"], "description": "Bloom and cuckoo filters provide fast approximate set membership while using little memory.", "description_method": "abstract-excerpt", "abstract": "Bloom and cuckoo filters provide fast approximate set membership while using little memory. Engineers use them to avoid expensive disk and network accesses. The recently introduced xor filters can be faster and smaller than Bloom and cuckoo filters. The xor filters are within 23% of the theoretical lower bound in storage as opposed to 44% for Bloom filters. Inspired by Dietzfelbinger and Walzer, we build probabilistic filters -- called binary fuse filters -- that are within 13% of the storage lower bound -- without sacrificing query speed. As an additional benefit, the construction of the new binary fuse filters can be more than twice as fast as the construction of xor filters. By slightly sacrificing query speed, we further reduce storage to within 8% of the lower bound. We compare the performance against a wide range of competitive alternatives such as Bloom filters, blocked Bloom filters, vector quotient filters, cuckoo filters, and the recent ribbon filters. Our experiments suggest that binary fuse filters are superior to xor filters.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-10", "2024-09", "2025-05"], "last_mentioned": "2025-05-02T08:21:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2202.00666", "slug": "2802fc6500b69fae", "title": "Locally Typical Sampling", "authors": ["Clara Meister", "Tiago Pimentel", "Gian Wiher", "Ryan Cotterell"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2202.00666", "doi": null, "topics": ["Statistics & probability", "Psychology"], "description": "Today's probabilistic language generators fall short when it comes to producing coherent and fluent text despite the fact that the underlying models perform well under standard metrics, e.g., perplexity.", "description_method": "abstract-excerpt", "abstract": "Today's probabilistic language generators fall short when it comes to producing coherent and fluent text despite the fact that the underlying models perform well under standard metrics, e.g., perplexity. This discrepancy has puzzled the language generation community for the last few years. In this work, we posit that the abstraction of natural language generation as a discrete stochastic process--which allows for an information-theoretic analysis--can provide new insights into the behavior of probabilistic language generators, e.g., why high-probability texts can be dull or repetitive. Humans use language as a means of communicating information, aiming to do so in a simultaneously efficient and error-minimizing manner; in fact, psycholinguistics research suggests humans choose each word in a string with this subconscious goal in mind. We formally define the set of strings that meet this criterion: those for which each word has an information content close to the expected information content, i.e., the conditional entropy of our model. We then propose a simple and efficient procedure for enforcing this criterion when generating from probabilistic models, which we call locally typical sampling. Automatic and human evaluations show that, in comparison to nucleus and top-k sampling, locally typical sampling offers competitive performance (in both abstractive summarization and story generation) in terms of quality while consistently reducing degenerate repetitions.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-03", "2023-06", "2023-10"], "last_mentioned": "2023-10-31T23:16:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2202.11214", "slug": "55723d372d8c17aa", "title": "FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators", "authors": ["Jaideep Pathak", "Shashank Subramanian", "Peter Harrington", "Sanjeev Raja", "Ashesh Chattopadhyay", "Morteza Mardani", "Thorsten Kurth", "David Hall", "Zongyi Li", "Kamyar Azizzadenesheli", "Pedram Hassanzadeh", "Karthik Kashinath", "Animashree Anandkumar"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2202.11214", "doi": null, "topics": ["Machine learning"], "description": "FourCastNet, short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at $0.25^{\\circ}$ resolution.", "description_method": "abstract-excerpt", "abstract": "FourCastNet, short for Fourier Forecasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at $0.25^{\\circ}$ resolution. FourCastNet accurately forecasts high-resolution, fast-timescale variables such as the surface wind speed, precipitation, and atmospheric water vapor. It has important implications for planning wind energy resources, predicting extreme weather events such as tropical cyclones, extra-tropical cyclones, and atmospheric rivers. FourCastNet matches the forecasting accuracy of the ECMWF Integrated Forecasting System (IFS), a state-of-the-art Numerical Weather Prediction (NWP) model, at short lead times for large-scale variables, while outperforming IFS for variables with complex fine-scale structure, including precipitation. FourCastNet generates a week-long forecast in less than 2 seconds, orders of magnitude faster than IFS. The speed of FourCastNet enables the creation of rapid and inexpensive large-ensemble forecasts with thousands of ensemble-members for improving probabilistic forecasting. We discuss how data-driven deep learning models such as FourCastNet are a valuable addition to the meteorology toolkit to aid and augment NWP models.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-02", "2024-03", "2025-02"], "last_mentioned": "2025-02-26T15:53:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2203.00671", "slug": "7bd0948b1ea4e847", "title": "Maximum Flow and Minimum-Cost Flow in Almost-Linear Time", "authors": ["Li Chen", "Rasmus Kyng", "Yang P. Liu", "Richard Peng", "Maximilian Probst Gutenberg", "Sushant Sachdeva"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2203.00671", "doi": null, "topics": ["Algorithms", "Statistics & probability"], "description": "We give an algorithm that computes exact maximum flows and minimum-cost flows on directed graphs with $m$ edges and polynomially bounded integral demands, costs, and capacities in $m^{1+o(1)}$ time.", "description_method": "abstract-excerpt", "abstract": "We give an algorithm that computes exact maximum flows and minimum-cost flows on directed graphs with $m$ edges and polynomially bounded integral demands, costs, and capacities in $m^{1+o(1)}$ time. Our algorithm builds the flow through a sequence of $m^{1+o(1)}$ approximate undirected minimum-ratio cycles, each of which is computed and processed in amortized $m^{o(1)}$ time using a new dynamic graph data structure. Our framework extends to algorithms running in $m^{1+o(1)}$ time for computing flows that minimize general edge-separable convex functions to high accuracy. This gives almost-linear time algorithms for several problems including entropy-regularized optimal transport, matrix scaling, $p$-norm flows, and $p$-norm isotonic regression on arbitrary directed acyclic graphs.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-06", "2023-02", "2024-06"], "last_mentioned": "2024-06-30T01:04:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2203.03466", "slug": "15e362370b840ec9", "title": "Tensor Programs V: Tuning Large Neural Networks via Zero-Shot Hyperparameter Transfer", "authors": ["Greg Yang", "Edward J. Hu", "Igor Babuschkin", "Szymon Sidor", "Xiaodong Liu", "David Farhi", "Nick Ryder", "Jakub Pachocki", "Weizhu Chen", "Jianfeng Gao"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2203.03466", "doi": null, "topics": ["Machine learning", "Language models"], "description": "Hyperparameter (HP) tuning in deep learning is an expensive process, prohibitively so for neural networks (NNs) with billions of parameters.", "description_method": "abstract-excerpt", "abstract": "Hyperparameter (HP) tuning in deep learning is an expensive process, prohibitively so for neural networks (NNs) with billions of parameters. We show that, in the recently discovered Maximal Update Parametrization (muP), many optimal HPs remain stable even as model size changes. This leads to a new HP tuning paradigm we call muTransfer: parametrize the target model in muP, tune the HP indirectly on a smaller model, and zero-shot transfer them to the full-sized model, i.e., without directly tuning the latter at all. We verify muTransfer on Transformer and ResNet. For example, 1) by transferring pretraining HPs from a model of 13M parameters, we outperform published numbers of BERT-large (350M parameters), with a total tuning cost equivalent to pretraining BERT-large once; 2) by transferring from 40M parameters, we outperform published numbers of the 6.7B GPT-3 model, with tuning cost only 7% of total pretraining cost. A Pytorch implementation of our technique can be found at github.com/microsoft/mup and installable via `pip install mup`.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-04", "2023-03", "2024-12"], "last_mentioned": "2024-12-23T06:10:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2203.11171", "slug": "02a15b098d96a94d", "title": "Self-Consistency Improves Chain of Thought Reasoning in Language Models", "authors": ["Xuezhi Wang", "Jason Wei", "Dale Schuurmans", "Quoc Le", "Ed Chi", "Sharan Narang", "Aakanksha Chowdhery", "Denny Zhou"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2203.11171", "doi": null, "topics": ["Distributed systems", "Language models"], "description": "Chain-of-thought prompting combined with pre-trained large language models has achieved encouraging results on complex reasoning tasks.", "description_method": "abstract-excerpt", "abstract": "Chain-of-thought prompting combined with pre-trained large language models has achieved encouraging results on complex reasoning tasks. In this paper, we propose a new decoding strategy, self-consistency, to replace the naive greedy decoding used in chain-of-thought prompting. It first samples a diverse set of reasoning paths instead of only taking the greedy one, and then selects the most consistent answer by marginalizing out the sampled reasoning paths. Self-consistency leverages the intuition that a complex reasoning problem typically admits multiple different ways of thinking leading to its unique correct answer. Our extensive empirical evaluation shows that self-consistency boosts the performance of chain-of-thought prompting with a striking margin on a range of popular arithmetic and commonsense reasoning benchmarks, including GSM8K (+17.9%), SVAMP (+11.0%), AQuA (+12.2%), StrategyQA (+6.4%) and ARC-challenge (+3.9%).", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-03", "2024-02", "2026-02"], "last_mentioned": "2026-02-22T07:54:08Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2204.05862", "slug": "45c6e20f553c02ad", "title": "Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback", "authors": ["Yuntao Bai", "Andy Jones", "Kamal Ndousse", "Amanda Askell", "Anna Chen", "Nova DasSarma", "Dawn Drain", "Stanislav Fort", "Deep Ganguli", "Tom Henighan", "Nicholas Joseph", "Saurav Kadavath", "Jackson Kernion", "Tom Conerly", "Sheer El-Showk", "Nelson Elhage", "Zac Hatfield-Dodds", "Danny Hernandez", "Tristan Hume", "Scott Johnston", "Shauna Kravec", "Liane Lovitt", "Neel Nanda", "Catherine Olsson", "Dario Amodei", "Tom Brown", "Jack Clark", "Sam McCandlish", "Chris Olah", "Ben Mann", "Jared Kaplan"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2204.05862", "doi": null, "topics": ["Machine learning", "Language models"], "description": "We apply preference modeling and reinforcement learning from human feedback (RLHF) to finetune language models to act as helpful and harmless assistants.", "description_method": "abstract-excerpt", "abstract": "We apply preference modeling and reinforcement learning from human feedback (RLHF) to finetune language models to act as helpful and harmless assistants. We find this alignment training improves performance on almost all NLP evaluations, and is fully compatible with training for specialized skills such as python coding and summarization. We explore an iterated online mode of training, where preference models and RL policies are updated on a weekly cadence with fresh human feedback data, efficiently improving our datasets and models. Finally, we investigate the robustness of RLHF training, and identify a roughly linear relation between the RL reward and the square root of the KL divergence between the policy and its initialization. Alongside our main results, we perform peripheral analyses on calibration, competing objectives, and the use of OOD detection, compare our models with human writers, and provide samples from our models using prompts appearing in recent related work.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-03", "2023-08", "2023-09"], "last_mentioned": "2023-09-12T05:22:11Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2204.14198", "slug": "1744b91a1fb55e75", "title": "Flamingo: a Visual Language Model for Few-Shot Learning", "authors": ["Jean-Baptiste Alayrac", "Jeff Donahue", "Pauline Luc", "Antoine Miech", "Iain Barr", "Yana Hasson", "Karel Lenc", "Arthur Mensch", "Katie Millican", "Malcolm Reynolds", "Roman Ring", "Eliza Rutherford", "Serkan Cabi", "Tengda Han", "Zhitao Gong", "Sina Samangooei", "Marianne Monteiro", "Jacob Menick", "Sebastian Borgeaud", "Andrew Brock", "Aida Nematzadeh", "Sahand Sharifzadeh", "Mikolaj Binkowski", "Ricardo Barreira", "Oriol Vinyals", "Andrew Zisserman", "Karen Simonyan"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2204.14198", "doi": null, "topics": ["Language models", "Machine learning"], "description": "Building models that can be rapidly adapted to novel tasks using only a handful of annotated examples is an open challenge for multimodal machine learning research.", "description_method": "abstract-excerpt", "abstract": "Building models that can be rapidly adapted to novel tasks using only a handful of annotated examples is an open challenge for multimodal machine learning research. We introduce Flamingo, a family of Visual Language Models (VLM) with this ability. We propose key architectural innovations to: (i) bridge powerful pretrained vision-only and language-only models, (ii) handle sequences of arbitrarily interleaved visual and textual data, and (iii) seamlessly ingest images or videos as inputs. Thanks to their flexibility, Flamingo models can be trained on large-scale multimodal web corpora containing arbitrarily interleaved text and images, which is key to endow them with in-context few-shot learning capabilities. We perform a thorough evaluation of our models, exploring and measuring their ability to rapidly adapt to a variety of image and video tasks. These include open-ended tasks such as visual question-answering, where the model is prompted with a question which it has to answer; captioning tasks, which evaluate the ability to describe a scene or an event; and close-ended tasks such as multiple-choice visual question-answering. For tasks lying anywhere on this spectrum, a single Flamingo model can achieve a new state of the art with few-shot learning, simply by prompting the model with task-specific examples. On numerous benchmarks, Flamingo outperforms models fine-tuned on thousands of times more task-specific data.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-05", "2024-02", "2025-10"], "last_mentioned": "2025-10-01T19:28:43Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2206.01288", "slug": "64c123cc276942a0", "title": "Decentralized Training of Foundation Models in Heterogeneous Environments", "authors": ["Binhang Yuan", "Yongjun He", "Jared Quincy Davis", "Tianyi Zhang", "Tri Dao", "Beidi Chen", "Percy Liang", "Christopher Re", "Ce Zhang"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2206.01288", "doi": null, "topics": ["Algorithms", "Language models", "Distributed systems"], "description": "Training foundation models, such as GPT-3 and PaLM, can be extremely expensive, often involving tens of thousands of GPUs running continuously for months.", "description_method": "abstract-excerpt", "abstract": "Training foundation models, such as GPT-3 and PaLM, can be extremely expensive, often involving tens of thousands of GPUs running continuously for months. These models are typically trained in specialized clusters featuring fast, homogeneous interconnects and using carefully designed software systems that support both data parallelism and model/pipeline parallelism. Such dedicated clusters can be costly and difficult to obtain. Can we instead leverage the much greater amount of decentralized, heterogeneous, and lower-bandwidth interconnected compute? Previous works examining the heterogeneous, decentralized setting focus on relatively small models that can be trained in a purely data parallel manner. State-of-the-art schemes for model parallel foundation model training, such as Megatron, only consider the homogeneous data center setting. In this paper, we present the first study of training large foundation models with model parallelism in a decentralized regime over a heterogeneous network. Our key technical contribution is a scheduling algorithm that allocates different computational \"tasklets\" in the training of foundation models to a group of decentralized GPU devices connected by a slow heterogeneous network. We provide a formal cost model and further propose an efficient evolutionary algorithm to find the optimal allocation strategy. We conduct extensive experiments that represent different scenarios for learning over geo-distributed devices simulated using real-world network measurements. In the most extreme case, across 8 different cities spanning 3 continents, our approach is 4.8X faster than prior state-of-the-art training systems (Megatron).", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-11", "2023-02", "2024-08"], "last_mentioned": "2024-08-28T18:19:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2206.14486", "slug": "13c05ef45b3b7ae8", "title": "Beyond neural scaling laws: beating power law scaling via data pruning", "authors": ["Ben Sorscher", "Robert Geirhos", "Shashank Shekhar", "Surya Ganguli", "Ari S. Morcos"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2206.14486", "doi": null, "topics": ["Machine learning"], "description": "Widely observed neural scaling laws, in which error falls off as a power of the training set size, model size, or both, have driven substantial performance improvements in deep learning.", "description_method": "abstract-excerpt", "abstract": "Widely observed neural scaling laws, in which error falls off as a power of the training set size, model size, or both, have driven substantial performance improvements in deep learning. However, these improvements through scaling alone require considerable costs in compute and energy. Here we focus on the scaling of error with dataset size and show how in theory we can break beyond power law scaling and potentially even reduce it to exponential scaling instead if we have access to a high-quality data pruning metric that ranks the order in which training examples should be discarded to achieve any pruned dataset size. We then test this improved scaling prediction with pruned dataset size empirically, and indeed observe better than power law scaling in practice on ResNets trained on CIFAR-10, SVHN, and ImageNet. Next, given the importance of finding high-quality pruning metrics, we perform the first large-scale benchmarking study of ten different data pruning metrics on ImageNet. We find most existing high performing metrics scale poorly to ImageNet, while the best are computationally intensive and require labels for every image. We therefore developed a new simple, cheap and scalable self-supervised pruning metric that demonstrates comparable performance to the best supervised metrics. Overall, our work suggests that the discovery of good data-pruning metrics may provide a viable path forward to substantially improved neural scaling laws, thereby reducing the resource costs of modern deep learning.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-07", "2022-12", "2026-03"], "last_mentioned": "2026-03-01T12:51:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2207.04859", "slug": "8e9ca65ae4fe3623", "title": "An Overview on IEEE 802.11bf: WLAN Sensing", "authors": ["Rui Du", "Hailiang Xie", "Mengshi Hu", " Narengerile", "Yan Xin", "Stephen McCann", "Michael Montemurro", "Tony Xiao Han", "Jie Xu"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2207.04859", "doi": null, "topics": ["Security & cryptography", "Psychology"], "description": "With recent advancements, the wireless local area network (WLAN) or wireless fidelity (Wi-Fi) technology has been successfully utilized to realize sensing functionalities such as detection, localization, and recognition.", "description_method": "abstract-excerpt", "abstract": "With recent advancements, the wireless local area network (WLAN) or wireless fidelity (Wi-Fi) technology has been successfully utilized to realize sensing functionalities such as detection, localization, and recognition. However, the WLANs standards are developed mainly for the purpose of communication, and thus may not be able to meet the stringent sensing requirements in emerging applications. To resolve this issue, a new Task Group (TG), namely IEEE 802.11bf, has been established by the IEEE 802.11 working group, with the objective of creating a new amendment to the WLAN standard to provide advanced sensing requirements while minimizing the effect on communications. This paper provides a comprehensive overview on the up-to-date efforts in the IEEE 802.11bf TG. First, we introduce the definition of the 802.11bf amendment and its standardization timeline. Then, we discuss the WLAN sensing procedure and framework used for measurement acquisition, by considering both conventional sensing at sub-7 GHz and directional multi-gigabit (DMG) sensing at 60 GHz, respectively. Next, we present various candidate technical features for IEEE 802.11bf, including waveform/sequence design, feedback types, quantization, as well as security and privacy. Finally, we describe the methodologies used by the IEEE 802.11bf TG to evaluate the alternative performance. It is desired that this overview paper provide useful insights on IEEE 802.11 WLAN sensing to people with great interests and promote the IEEE 802.11bf standard to be widely deployed.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-11", "2024-05", "2024-09"], "last_mentioned": "2024-09-28T08:59:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2208.09392", "slug": "a05c2ab43fecb09a", "title": "Cold Diffusion: Inverting Arbitrary Image Transforms Without Noise", "authors": ["Arpit Bansal", "Eitan Borgnia", "Hong-Min Chu", "Jie S. Li", "Hamid Kazemi", "Furong Huang", "Micah Goldblum", "Jonas Geiping", "Tom Goldstein"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2208.09392", "doi": null, "topics": ["Physics", "Machine learning", "Psychology"], "description": "Standard diffusion models involve an image transform -- adding Gaussian noise -- and an image restoration operator that inverts this degradation.", "description_method": "abstract-excerpt", "abstract": "Standard diffusion models involve an image transform -- adding Gaussian noise -- and an image restoration operator that inverts this degradation. We observe that the generative behavior of diffusion models is not strongly dependent on the choice of image degradation, and in fact an entire family of generative models can be constructed by varying this choice. Even when using completely deterministic degradations (e.g., blur, masking, and more), the training and test-time update rules that underlie diffusion models can be easily generalized to create generative models. The success of these fully deterministic models calls into question the community's understanding of diffusion models, which relies on noise in either gradient Langevin dynamics or variational inference, and paves the way for generalized diffusion models that invert arbitrary processes. Our code is available at https://github.com/arpitbansal297/Cold-Diffusion-Models", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-10", "2023-01", "2023-03"], "last_mentioned": "2023-03-14T12:25:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2209.06930", "slug": "e6f33e557c62d544", "title": "How Much Structure Is Needed for Huge Quantum Speedups?", "authors": ["Scott Aaronson"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2209.06930", "doi": null, "topics": ["Physics", "Algorithms", "Science & research"], "description": "I survey, for a general scientific audience, three decades of research into which sorts of problems admit exponential speedups via quantum computers -- from the classics (like the algorithms of Simon and Shor), to the breakthrough of Yamakawa and Zhandry from April 2022.", "description_method": "abstract-excerpt", "abstract": "I survey, for a general scientific audience, three decades of research into which sorts of problems admit exponential speedups via quantum computers -- from the classics (like the algorithms of Simon and Shor), to the breakthrough of Yamakawa and Zhandry from April 2022. I discuss both the quantum circuit model, which is what we ultimately care about in practice but where our knowledge is radically incomplete, and the so-called oracle or black-box or query complexity model, where we've managed to achieve a much more thorough understanding that then informs our conjectures about the circuit model. I discuss the strengths and weaknesses of switching attention to sampling tasks, as was done in the recent quantum supremacy experiments. I make some skeptical remarks about widely-repeated claims of exponential quantum speedups for practical machine learning and optimization problems. Through many examples, I try to convey the \"law of conservation of weirdness,\" according to which every problem admitting an exponential quantum speedup must have some unusual property to allow the amplitude to be concentrated on the unknown right answer(s).", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-10", "2022-11", "2025-10"], "last_mentioned": "2025-10-22T19:23:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2211.04325", "slug": "048c1d0388b25d75", "title": "Will we run out of data? Limits of LLM scaling based on human-generated data", "authors": ["Pablo Villalobos", "Anson Ho", "Jaime Sevilla", "Tamay Besiroglu", "Lennart Heim", "Marius Hobbhahn"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2211.04325", "doi": null, "topics": ["Language models"], "description": "We investigate the potential constraints on LLM scaling posed by the availability of public human-generated text data.", "description_method": "abstract-excerpt", "abstract": "We investigate the potential constraints on LLM scaling posed by the availability of public human-generated text data. We forecast the growing demand for training data based on current trends and estimate the total stock of public human text data. Our findings indicate that if current LLM development trends continue, models will be trained on datasets roughly equal in size to the available stock of public human text data between 2026 and 2032, or slightly earlier if models are overtrained. We explore how progress in language modeling can continue when human-generated text datasets cannot be scaled any further. We argue that synthetic data generation, transfer learning from data-rich domains, and data efficiency improvements might support further progress.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-12", "2023-12", "2025-09"], "last_mentioned": "2025-09-28T16:09:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2211.05102", "slug": "c6f3ac1af2b9c2b4", "title": "Efficiently Scaling Transformer Inference", "authors": ["Reiner Pope", "Sholto Douglas", "Aakanksha Chowdhery", "Jacob Devlin", "James Bradbury", "Anselm Levskaya", "Jonathan Heek", "Kefan Xiao", "Shivani Agrawal", "Jeff Dean"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2211.05102", "doi": null, "topics": ["Language models", "Databases", "Algorithms"], "description": "We study the problem of efficient generative inference for Transformer models, in one of its most challenging settings: large deep models, with tight latency targets and long sequence lengths.", "description_method": "abstract-excerpt", "abstract": "We study the problem of efficient generative inference for Transformer models, in one of its most challenging settings: large deep models, with tight latency targets and long sequence lengths. Better understanding of the engineering tradeoffs for inference for large Transformer-based models is important as use cases of these models are growing rapidly throughout application areas. We develop a simple analytical model for inference efficiency to select the best multi-dimensional partitioning techniques optimized for TPU v4 slices based on the application requirements. We combine these with a suite of low-level optimizations to achieve a new Pareto frontier on the latency and model FLOPS utilization (MFU) tradeoffs on 500B+ parameter models that outperforms the FasterTransformer suite of benchmarks. We further show that with appropriate partitioning, the lower memory requirements of multiquery attention (i.e. multiple query heads share single key/value head) enables scaling up to 32x larger context lengths. Finally, we achieve a low-batch-size latency of 29ms per token during generation (using int8 weight quantization) and a 76% MFU during large-batch-size processing of input tokens, while supporting a long 2048-token context length on the PaLM 540B parameter model.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-06", "2023-07", "2024-02"], "last_mentioned": "2024-02-25T17:55:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2211.06956", "slug": "0bafb2e497d5a3b2", "title": "Seeing Beyond the Brain: Conditional Diffusion Model with Sparse Masked Modeling for Vision Decoding", "authors": ["Zijiao Chen", "Jiaxin Qing", "Tiange Xiang", "Wan Lin Yue", "Juan Helen Zhou"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2211.06956", "doi": null, "topics": ["Neuroscience", "Physics", "Machine learning"], "description": "Decoding visual stimuli from brain recordings aims to deepen our understanding of the human visual system and build a solid foundation for bridging human and computer vision through the Brain-Computer Interface.", "description_method": "abstract-excerpt", "abstract": "Decoding visual stimuli from brain recordings aims to deepen our understanding of the human visual system and build a solid foundation for bridging human and computer vision through the Brain-Computer Interface. However, reconstructing high-quality images with correct semantics from brain recordings is a challenging problem due to the complex underlying representations of brain signals and the scarcity of data annotations. In this work, we present MinD-Vis: Sparse Masked Brain Modeling with Double-Conditioned Latent Diffusion Model for Human Vision Decoding. Firstly, we learn an effective self-supervised representation of fMRI data using mask modeling in a large latent space inspired by the sparse coding of information in the primary visual cortex. Then by augmenting a latent diffusion model with double-conditioning, we show that MinD-Vis can reconstruct highly plausible images with semantically matching details from brain recordings using very few paired annotations. We benchmarked our model qualitatively and quantitatively; the experimental results indicate that our method outperformed state-of-the-art in both semantic mapping (100-way semantic classification) and generation quality (FID) by 66% and 41% respectively. An exhaustive ablation study was also conducted to analyze our framework.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-11", "2023-02", "2023-05"], "last_mentioned": "2023-05-02T04:07:29Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2211.10435", "slug": "437bf9ed9fd0e927", "title": "PAL: Program-aided Language Models", "authors": ["Luyu Gao", "Aman Madaan", "Shuyan Zhou", "Uri Alon", "Pengfei Liu", "Yiming Yang", "Jamie Callan", "Graham Neubig"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2211.10435", "doi": null, "topics": ["Language models", "Mathematics", "Algorithms"], "description": "Large language models (LLMs) have recently demonstrated an impressive ability to perform arithmetic and symbolic reasoning tasks, when provided with a few examples at test time (\"few-shot prompting\").", "description_method": "abstract-excerpt", "abstract": "Large language models (LLMs) have recently demonstrated an impressive ability to perform arithmetic and symbolic reasoning tasks, when provided with a few examples at test time (\"few-shot prompting\"). Much of this success can be attributed to prompting methods such as \"chain-of-thought'', which employ LLMs for both understanding the problem description by decomposing it into steps, as well as solving each step of the problem. While LLMs seem to be adept at this sort of step-by-step decomposition, LLMs often make logical and arithmetic mistakes in the solution part, even when the problem is decomposed correctly. In this paper, we present Program-Aided Language models (PAL): a novel approach that uses the LLM to read natural language problems and generate programs as the intermediate reasoning steps, but offloads the solution step to a runtime such as a Python interpreter. With PAL, decomposing the natural language problem into runnable steps remains the only learning task for the LLM, while solving is delegated to the interpreter. We demonstrate this synergy between a neural LLM and a symbolic interpreter across 13 mathematical, symbolic, and algorithmic reasoning tasks from BIG-Bench Hard and other benchmarks. In all these natural language reasoning tasks, generating code using an LLM and reasoning using a Python interpreter leads to more accurate results than much larger models. For example, PAL using Codex achieves state-of-the-art few-shot accuracy on the GSM8K benchmark of math word problems, surpassing PaLM-540B which uses chain-of-thought by absolute 15% top-1. Our code and data are publicly available at http://reasonwithpal.com/ .", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-01", "2023-03", "2025-11"], "last_mentioned": "2025-11-30T20:25:49Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.11279", "slug": "5436d8543002e7f4", "title": "Annotated History of Modern AI and Deep Learning", "authors": ["Juergen Schmidhuber"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.11279", "doi": null, "topics": ["Machine learning", "Mathematics", "Psychology"], "description": "Machine learning (ML) is the science of credit assignment.", "description_method": "abstract-excerpt", "abstract": "Machine learning (ML) is the science of credit assignment. It seeks to find patterns in observations that explain and predict the consequences of events and actions. This then helps to improve future performance. Minsky's so-called \"fundamental credit assignment problem\" (1963) surfaces in all sciences including physics (why is the world the way it is?) and history (which persons/ideas/actions have shaped society and civilisation?). Here I focus on the history of ML itself. Modern artificial intelligence (AI) is dominated by artificial neural networks (NNs) and deep learning, both of which are conceptually closer to the old field of cybernetics than what was traditionally called AI (e.g., expert systems and logic programming). A modern history of AI & ML must emphasize breakthroughs outside the scope of shallow AI text books. In particular, it must cover the mathematical foundations of today's NNs such as the chain rule (1676), the first NNs (circa 1800), the first practical AI (1914), the theory of AI and its limitations (1931-34), and the first working deep learning algorithms (1965-). From the perspective of 2025, I provide a timeline of the most significant events in the history of NNs, ML, deep learning, AI, computer science, and mathematics in general, crediting the individuals who laid the field's foundations. The text contains numerous hyperlinks to relevant overview sites. With a ten-year delay, it supplements my 2015 award-winning deep learning survey which provides hundreds of additional references. Finally, I will put things in a broader historical context, spanning from the Big Bang to when the universe will be many times older than it is now.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-06", "2024-09", "2025-04"], "last_mentioned": "2025-04-30T10:19:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2212.12372", "slug": "6ef776c43ce79c31", "title": "Factoring integers with sublinear resources on a superconducting quantum processor", "authors": ["Bao Yan", "Ziqi Tan", "Shijie Wei", "Haocong Jiang", "Weilong Wang", "Hong Wang", "Lan Luo", "Qianheng Duan", "Yiting Liu", "Wenhao Shi", "Yangyang Fei", "Xiangdong Meng", "Yu Han", "Zheng Shan", "Jiachen Chen", "Xuhao Zhu", "Chuanyu Zhang", "Feitong Jin", "Hekang Li", "Chao Song", "Zhen Wang", "Zhi Ma", "H. Wang", "Gui-Lu Long"], "year": 2022, "venue": "arXiv", "url": "https://arxiv.org/abs/2212.12372", "doi": null, "topics": ["Physics", "Algorithms", "Security & cryptography"], "description": "Shor's algorithm has seriously challenged information security based on public key cryptosystems.", "description_method": "abstract-excerpt", "abstract": "Shor's algorithm has seriously challenged information security based on public key cryptosystems. However, to break the widely used RSA-2048 scheme, one needs millions of physical qubits, which is far beyond current technical capabilities. Here, we report a universal quantum algorithm for integer factorization by combining the classical lattice reduction with a quantum approximate optimization algorithm (QAOA). The number of qubits required is O(logN/loglog N), which is sublinear in the bit length of the integer $N$, making it the most qubit-saving factorization algorithm to date. We demonstrate the algorithm experimentally by factoring integers up to 48 bits with 10 superconducting qubits, the largest integer factored on a quantum device. We estimate that a quantum circuit with 372 physical qubits and a depth of thousands is necessary to challenge RSA-2048 using our algorithm. Our study shows great promise in expediting the application of current noisy quantum computers, and paves the way to factor large integers of realistic cryptographic significance.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-01", "2023-08", "2024-05"], "last_mentioned": "2024-05-24T17:19:35Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.02111", "slug": "eb219f8c52a781eb", "title": "Neural Codec Language Models are Zero-Shot Text to Speech Synthesizers", "authors": ["Chengyi Wang", "Sanyuan Chen", "Yu Wu", "Ziqiang Zhang", "Long Zhou", "Shujie Liu", "Zhuo Chen", "Yanqing Liu", "Huaming Wang", "Jinyu Li", "Lei He", "Sheng Zhao", "Furu Wei"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.02111", "doi": null, "topics": ["Language models", "Statistics & probability"], "description": "We introduce a language modeling approach for text to speech synthesis (TTS).", "description_method": "abstract-excerpt", "abstract": "We introduce a language modeling approach for text to speech synthesis (TTS). Specifically, we train a neural codec language model (called Vall-E) using discrete codes derived from an off-the-shelf neural audio codec model, and regard TTS as a conditional language modeling task rather than continuous signal regression as in previous work. During the pre-training stage, we scale up the TTS training data to 60K hours of English speech which is hundreds of times larger than existing systems. Vall-E emerges in-context learning capabilities and can be used to synthesize high-quality personalized speech with only a 3-second enrolled recording of an unseen speaker as an acoustic prompt. Experiment results show that Vall-E significantly outperforms the state-of-the-art zero-shot TTS system in terms of speech naturalness and speaker similarity. In addition, we find Vall-E could preserve the speaker's emotion and acoustic environment of the acoustic prompt in synthesis. See https://aka.ms/valle for demos of our work.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-01", "2023-02", "2024-09"], "last_mentioned": "2024-09-17T03:07:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.04104", "slug": "266d59e63d2f705e", "title": "Mastering Diverse Domains through World Models", "authors": ["Danijar Hafner", "Jurgis Pasukonis", "Jimmy Ba", "Timothy Lillicrap"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.04104", "doi": null, "topics": ["Algorithms", "Psychology", "Machine learning"], "description": "Developing a general algorithm that learns to solve tasks across a wide range of applications has been a fundamental challenge in artificial intelligence.", "description_method": "abstract-excerpt", "abstract": "Developing a general algorithm that learns to solve tasks across a wide range of applications has been a fundamental challenge in artificial intelligence. Although current reinforcement learning algorithms can be readily applied to tasks similar to what they have been developed for, configuring them for new application domains requires significant human expertise and experimentation. We present DreamerV3, a general algorithm that outperforms specialized methods across over 150 diverse tasks, with a single configuration. Dreamer learns a model of the environment and improves its behavior by imagining future scenarios. Robustness techniques based on normalization, balancing, and transformations enable stable learning across domains. Applied out of the box, Dreamer is the first algorithm to collect diamonds in Minecraft from scratch without human data or curricula. This achievement has been posed as a significant challenge in artificial intelligence that requires exploring farsighted strategies from pixels and sparse rewards in an open world. Our work allows solving challenging control problems without extensive experimentation, making reinforcement learning broadly applicable.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-01", "2023-07", "2025-04"], "last_mentioned": "2025-04-07T15:55:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.05062", "slug": "3e56faf8bfb6beac", "title": "Tracr: Compiled Transformers as a Laboratory for Interpretability", "authors": ["David Lindner", "János Kramár", "Sebastian Farquhar", "Matthew Rahtz", "Thomas McGrath", "Vladimir Mikulik"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.05062", "doi": null, "topics": ["Language models", "Economics", "Algorithms"], "description": "We show how to \"compile\" human-readable programs into standard decoder-only transformer models.", "description_method": "abstract-excerpt", "abstract": "We show how to \"compile\" human-readable programs into standard decoder-only transformer models. Our compiler, Tracr, generates models with known structure. This structure can be used to design experiments. For example, we use it to study \"superposition\" in transformers that execute multi-step algorithms. Additionally, the known structure of Tracr-compiled models can serve as ground-truth for evaluating interpretability methods. Commonly, because the \"programs\" learned by transformers are unknown it is unclear whether an interpretation succeeded. We demonstrate our approach by implementing and examining programs including computing token frequencies, sorting, and parenthesis checking. We provide an open-source implementation of Tracr at https://github.com/google-deepmind/tracr.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-02", "2023-09", "2023-12"], "last_mentioned": "2023-12-13T14:44:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.11325", "slug": "4ff38613ee9da1ae", "title": "MusicLM: Generating Music From Text", "authors": ["Andrea Agostinelli", "Timo I. Denk", "Zalán Borsos", "Jesse Engel", "Mauro Verzetti", "Antoine Caillon", "Qingqing Huang", "Aren Jansen", "Adam Roberts", "Marco Tagliasacchi", "Matt Sharifi", "Neil Zeghidour", "Christian Frank"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.11325", "doi": null, "topics": ["Other research"], "description": "We introduce MusicLM, a model generating high-fidelity music from text descriptions such as \"a calming violin melody backed by a distorted guitar riff\".", "description_method": "abstract-excerpt", "abstract": "We introduce MusicLM, a model generating high-fidelity music from text descriptions such as \"a calming violin melody backed by a distorted guitar riff\". MusicLM casts the process of conditional music generation as a hierarchical sequence-to-sequence modeling task, and it generates music at 24 kHz that remains consistent over several minutes. Our experiments show that MusicLM outperforms previous systems both in audio quality and adherence to the text description. Moreover, we demonstrate that MusicLM can be conditioned on both text and a melody in that it can transform whistled and hummed melodies according to the style described in a text caption. To support future research, we publicly release MusicCaps, a dataset composed of 5.5k music-text pairs, with rich text descriptions provided by human experts.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-01", "2023-05", "2025-07"], "last_mentioned": "2025-07-31T14:18:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2301.12597", "slug": "bfa86ff73462125e", "title": "BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models", "authors": ["Junnan Li", "Dongxu Li", "Silvio Savarese", "Steven Hoi"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2301.12597", "doi": null, "topics": ["Language models", "Databases"], "description": "The cost of vision-and-language pre-training has become increasingly prohibitive due to end-to-end training of large-scale models.", "description_method": "abstract-excerpt", "abstract": "The cost of vision-and-language pre-training has become increasingly prohibitive due to end-to-end training of large-scale models. This paper proposes BLIP-2, a generic and efficient pre-training strategy that bootstraps vision-language pre-training from off-the-shelf frozen pre-trained image encoders and frozen large language models. BLIP-2 bridges the modality gap with a lightweight Querying Transformer, which is pre-trained in two stages. The first stage bootstraps vision-language representation learning from a frozen image encoder. The second stage bootstraps vision-to-language generative learning from a frozen language model. BLIP-2 achieves state-of-the-art performance on various vision-language tasks, despite having significantly fewer trainable parameters than existing methods. For example, our model outperforms Flamingo80B by 8.7% on zero-shot VQAv2 with 54x fewer trainable parameters. We also demonstrate the model's emerging capabilities of zero-shot image-to-text generation that can follow natural language instructions.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-03", "2023-04", "2023-05"], "last_mentioned": "2023-05-05T11:44:55Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2302.12173", "slug": "89d4954c8eba18c4", "title": "Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection", "authors": ["Kai Greshake", "Sahar Abdelnabi", "Shailesh Mishra", "Christoph Endres", "Thorsten Holz", "Mario Fritz"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2302.12173", "doi": null, "topics": ["Language models", "Security & cryptography"], "description": "Large Language Models (LLMs) are increasingly being integrated into various applications.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) are increasingly being integrated into various applications. The functionalities of recent LLMs can be flexibly modulated via natural language prompts. This renders them susceptible to targeted adversarial prompting, e.g., Prompt Injection (PI) attacks enable attackers to override original instructions and employed controls. So far, it was assumed that the user is directly prompting the LLM. But, what if it is not the user prompting? We argue that LLM-Integrated Applications blur the line between data and instructions. We reveal new attack vectors, using Indirect Prompt Injection, that enable adversaries to remotely (without a direct interface) exploit LLM-integrated applications by strategically injecting prompts into data likely to be retrieved. We derive a comprehensive taxonomy from a computer security perspective to systematically investigate impacts and vulnerabilities, including data theft, worming, information ecosystem contamination, and other novel security risks. We demonstrate our attacks' practical viability against both real-world systems, such as Bing's GPT-4 powered Chat and code-completion engines, and synthetic applications built on GPT-4. We show how processing retrieved prompts can act as arbitrary code execution, manipulate the application's functionality, and control how and if other APIs are called. Despite the increasing integration and reliance on LLMs, effective mitigations of these emerging threats are currently lacking. By raising awareness of these vulnerabilities and providing key insights into their implications, we aim to promote the safe and responsible deployment of these powerful models and the development of robust defenses that protect users and systems from potential attacks.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-12", "2024-03", "2025-10"], "last_mentioned": "2025-10-06T13:15:18Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2303.10798", "slug": "0d85aec64d07255d", "title": "An aperiodic monotile", "authors": ["David Smith", "Joseph Samuel Myers", "Craig S. Kaplan", "Chaim Goodman-Strauss"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2303.10798", "doi": "10.5070/C64163843", "topics": ["Algorithms"], "description": "A longstanding open problem asks for an aperiodic monotile, also known as an \"einstein\": a shape that admits tilings of the plane, but never periodic tilings.", "description_method": "abstract-excerpt", "abstract": "A longstanding open problem asks for an aperiodic monotile, also known as an \"einstein\": a shape that admits tilings of the plane, but never periodic tilings. We answer this problem for topological disk tiles by exhibiting a continuum of combinatorially equivalent aperiodic polygons. We first show that a representative example, the \"hat\" polykite, can form clusters called \"metatiles\", for which substitution rules can be defined. Because the metatiles admit tilings of the plane, so too does the hat. We then prove that generic members of our continuum of polygons are aperiodic, through a new kind of geometric incommensurability argument. Separately, we give a combinatorial, computer-assisted proof that the hat must form hierarchical -- and hence aperiodic -- tilings.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-11", "2023-12", "2024-03"], "last_mentioned": "2024-03-20T08:39:40Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.05284", "slug": "8c788587f53c393d", "title": "Simple and Controllable Music Generation", "authors": ["Jade Copet", "Felix Kreuk", "Itai Gat", "Tal Remez", "David Kant", "Gabriel Synnaeve", "Yossi Adi", "Alexandre Défossez"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.05284", "doi": null, "topics": ["Language models"], "description": "We tackle the task of conditional music generation.", "description_method": "abstract-excerpt", "abstract": "We tackle the task of conditional music generation. We introduce MusicGen, a single Language Model (LM) that operates over several streams of compressed discrete music representation, i.e., tokens. Unlike prior work, MusicGen is comprised of a single-stage transformer LM together with efficient token interleaving patterns, which eliminates the need for cascading several models, e.g., hierarchically or upsampling. Following this approach, we demonstrate how MusicGen can generate high-quality samples, both mono and stereo, while being conditioned on textual description or melodic features, allowing better controls over the generated output. We conduct extensive empirical evaluation, considering both automatic and human studies, showing the proposed approach is superior to the evaluated baselines on a standard text-to-music benchmark. Through ablation studies, we shed light over the importance of each of the components comprising MusicGen. Music samples, code, and models are available at https://github.com/facebookresearch/audiocraft", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-06", "2025-03", "2025-10"], "last_mentioned": "2025-10-28T03:17:45Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.06391", "slug": "5b9447e9621c319c", "title": "The Design Principles of the Elixir Type System", "authors": ["Giuseppe Castagna", "Guillaume Duboc", "José Valim"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.06391", "doi": "10.22152/programming-journal.org/2024/8/4", "topics": ["Programming languages", "Computer systems"], "description": "Elixir is a dynamically-typed functional language running on the Erlang Virtual Machine, designed for building scalable and maintainable applications.", "description_method": "abstract-excerpt", "abstract": "Elixir is a dynamically-typed functional language running on the Erlang Virtual Machine, designed for building scalable and maintainable applications. Its characteristics have earned it a surging adoption by hundreds of industrial actors and tens of thousands of developers. Static typing seems nowadays to be the most important request coming from the Elixir community. We present a gradual type system we plan to include in the Elixir compiler, outline its characteristics and design principles, and show by some short examples how to use it in practice. Developing a static type system suitable for Erlang's family of languages has been an open research problem for almost two decades. Our system transposes to this family of languages a polymorphic type system with set-theoretic types and semantic subtyping. To do that, we had to improve and extend both semantic subtyping and the typing techniques thereof, to account for several characteristics of these languages -- and of Elixir in particular -- such as the arity of functions, the use of guards, a uniform treatment of records and dictionaries, the need for a new sound gradual typing discipline that does not rely on the insertion at compile time of specific run-time type-tests but, rather, takes into account both the type tests performed by the virtual machine and those explicitly added by the programmer. The system presented here is \"gradually\" being implemented and integrated in Elixir, but a prototype implementation is already available. The aim of this work is to serve as a longstanding reference that will be used to introduce types to Elixir programmers, as well as to hint at some future directions and possible evolutions of the Elixir language.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-11", "2024-05", "2025-11"], "last_mentioned": "2025-11-04T01:07:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2306.06657", "slug": "ad5912e371577d12", "title": "Starlink Generation 2 Mini Satellites: Photometric Characterization", "authors": ["Anthony Mallama", "Richard E. Cole", "Scott Harrington", "Andreas Hornig", "Jay Respler", "Aaron Worley", "Ron Lee"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2306.06657", "doi": null, "topics": ["Other research"], "description": "Starlink Generation 2 Mini satellites are fainter than Gen 1 spacecraft despite their larger size.", "description_method": "abstract-excerpt", "abstract": "Starlink Generation 2 Mini satellites are fainter than Gen 1 spacecraft despite their larger size. The mean of apparent magnitudes for satellites in brightness mitigation mode is 7.06 +/- 0.10. When these magnitudes are adjusted to a uniform distance of 1,000 km that mean is 7.87 +/- 0.09. The brightness mitigation mode reduces distance-adjusted satellite luminosity by a factor of 12 relative to spacecraft that are not mitigated.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-09", "2024-04", "2025-04"], "last_mentioned": "2025-04-19T21:51:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2307.01850", "slug": "3a32ad4e1eff6075", "title": "Self-Consuming Generative Models Go MAD", "authors": ["Sina Alemohammad", "Josue Casco-Rodriguez", "Lorenzo Luzi", "Ahmed Imtiaz Humayun", "Hossein Babaei", "Daniel LeJeune", "Ali Siahkoohi", "Richard G. Baraniuk"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2307.01850", "doi": null, "topics": ["Machine learning", "Psychology", "Biology & medicine"], "description": "Seismic advances in generative AI algorithms for imagery, text, and other data types has led to the temptation to use synthetic data to train next-generation models.", "description_method": "abstract-excerpt", "abstract": "Seismic advances in generative AI algorithms for imagery, text, and other data types has led to the temptation to use synthetic data to train next-generation models. Repeating this process creates an autophagous (self-consuming) loop whose properties are poorly understood. We conduct a thorough analytical and empirical analysis using state-of-the-art generative image models of three families of autophagous loops that differ in how fixed or fresh real training data is available through the generations of training and in whether the samples from previous generation models have been biased to trade off data quality versus diversity. Our primary conclusion across all scenarios is that without enough fresh real data in each generation of an autophagous loop, future generative models are doomed to have their quality (precision) or diversity (recall) progressively decrease. We term this condition Model Autophagy Disorder (MAD), making analogy to mad cow disease.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-07", "2023-10", "2024-07"], "last_mentioned": "2024-07-25T05:01:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2307.09702", "slug": "5a5588a3ff5b8167", "title": "Efficient Guided Generation for Large Language Models", "authors": ["Brandon T. Willard", "Rémi Louf"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2307.09702", "doi": null, "topics": ["Language models"], "description": "In this article we show how the problem of neural text generation can be constructively reformulated in terms of transitions between the states of a finite-state machine.", "description_method": "abstract-excerpt", "abstract": "In this article we show how the problem of neural text generation can be constructively reformulated in terms of transitions between the states of a finite-state machine. This framework leads to an efficient approach to guiding text generation with regular expressions and context-free grammars by allowing the construction of an index over a language model's vocabulary. The approach is model agnostic, allows one to enforce domain-specific knowledge and constraints, and enables the construction of reliable interfaces by guaranteeing the structure of the generated text. It adds little overhead to the token sequence generation process and significantly outperforms existing solutions. An implementation is provided in the open source Python library Outlines", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-08", "2024-05", "2026-01"], "last_mentioned": "2026-01-17T00:33:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.13548", "slug": "d6fb2b001042aa3d", "title": "Towards Understanding Sycophancy in Language Models", "authors": ["Mrinank Sharma", "Meg Tong", "Tomasz Korbak", "David Duvenaud", "Amanda Askell", "Samuel R. Bowman", "Newton Cheng", "Esin Durmus", "Zac Hatfield-Dodds", "Scott R. Johnston", "Shauna Kravec", "Timothy Maxwell", "Sam McCandlish", "Kamal Ndousse", "Oliver Rausch", "Nicholas Schiefer", "Da Yan", "Miranda Zhang", "Ethan Perez"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.13548", "doi": null, "topics": ["Psychology", "Language models"], "description": "Human feedback is commonly utilized to finetune AI assistants.", "description_method": "abstract-excerpt", "abstract": "Human feedback is commonly utilized to finetune AI assistants. But human feedback may also encourage model responses that match user beliefs over truthful ones, a behaviour known as sycophancy. We investigate the prevalence of sycophancy in models whose finetuning procedure made use of human feedback, and the potential role of human preference judgments in such behavior. We first demonstrate that five state-of-the-art AI assistants consistently exhibit sycophancy across four varied free-form text-generation tasks. To understand if human preferences drive this broadly observed behavior, we analyze existing human preference data. We find that when a response matches a user's views, it is more likely to be preferred. Moreover, both humans and preference models (PMs) prefer convincingly-written sycophantic responses over correct ones a non-negligible fraction of the time. Optimizing model outputs against PMs also sometimes sacrifices truthfulness in favor of sycophancy. Overall, our results indicate that sycophancy is a general behavior of state-of-the-art AI assistants, likely driven in part by human preference judgments favoring sycophantic responses.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-03", "2026-05", "2026-06"], "last_mentioned": "2026-06-27T13:21:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.15916", "slug": "e0e293767d1b461a", "title": "In-Context Learning Creates Task Vectors", "authors": ["Roee Hendel", "Mor Geva", "Amir Globerson"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.15916", "doi": null, "topics": ["Language models", "Statistics & probability", "Machine learning"], "description": "In-context learning (ICL) in Large Language Models (LLMs) has emerged as a powerful new learning paradigm.", "description_method": "abstract-excerpt", "abstract": "In-context learning (ICL) in Large Language Models (LLMs) has emerged as a powerful new learning paradigm. However, its underlying mechanism is still not well understood. In particular, it is challenging to map it to the \"standard\" machine learning framework, where one uses a training set $S$ to find a best-fitting function $f(x)$ in some hypothesis class. Here we make progress on this problem by showing that the functions learned by ICL often have a very simple structure: they correspond to the transformer LLM whose only inputs are the query $x$ and a single \"task vector\" calculated from the training set. Thus, ICL can be seen as compressing $S$ into a single task vector $\\boldsymbolθ(S)$ and then using this task vector to modulate the transformer to produce the output. We support the above claim via comprehensive experiments across a range of models and tasks.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-10", "2024-11", "2025-12"], "last_mentioned": "2025-12-14T07:46:36Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2310.17680", "slug": "8c5c67ae3af9b2af", "title": "CodeFusion: A Pre-trained Diffusion Model for Code Generation", "authors": ["Mukul Singh", "José Cambronero", "Sumit Gulwani", "Vu Le", "Carina Negreanu", "Gust Verbruggen"], "year": 2023, "venue": "arXiv", "url": "https://arxiv.org/abs/2310.17680", "doi": null, "topics": ["Physics", "Machine learning"], "description": "Imagine a developer who can only change their last line of code, how often would they have to start writing a function from scratch before it is correct?", "description_method": "abstract-excerpt", "abstract": "Imagine a developer who can only change their last line of code, how often would they have to start writing a function from scratch before it is correct? Auto-regressive models for code generation from natural language have a similar limitation: they do not easily allow reconsidering earlier tokens generated. We introduce CodeFusion, a pre-trained diffusion code generation model that addresses this limitation by iteratively denoising a complete program conditioned on the encoded natural language. We evaluate CodeFusion on the task of natural language to code generation for Bash, Python, and Microsoft Excel conditional formatting (CF) rules. Experiments show that CodeFusion (75M parameters) performs on par with state-of-the-art auto-regressive systems (350M-175B parameters) in top-1 accuracy and outperforms them in top-3 and top-5 accuracy due to its better balance in diversity versus quality.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-10", "2023-11", "2024-03"], "last_mentioned": "2024-03-23T06:52:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2401.05618", "slug": "c0e9e6ee06f93f63", "title": "The Benefits of a Concise Chain of Thought on Problem-Solving in Large Language Models", "authors": ["Matthew Renze", "Erhan Guven"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2401.05618", "doi": "10.1109/FLLM63129.2024.10852493", "topics": ["Language models"], "description": "In this paper, we introduce Concise Chain-of-Thought (CCoT) prompting.", "description_method": "abstract-excerpt", "abstract": "In this paper, we introduce Concise Chain-of-Thought (CCoT) prompting. We compared standard CoT and CCoT prompts to see how conciseness impacts response length and correct-answer accuracy. We evaluated this using GPT-3.5 and GPT-4 with a multiple-choice question-and-answer (MCQA) benchmark. CCoT reduced average response length by 48.70% for both GPT-3.5 and GPT-4 while having a negligible impact on problem-solving performance. However, on math problems, GPT-3.5 with CCoT incurs a performance penalty of 27.69%. Overall, CCoT leads to an average per-token cost reduction of 22.67%. All code, data, and supplemental materials are available on GitHub at https://github.com/matthewrenze/jhu-concise-cot", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-03", "2026-04", "2026-07"], "last_mentioned": "2026-07-30T09:13:28Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.01030", "slug": "db45d0b3b04844bc", "title": "Executable Code Actions Elicit Better LLM Agents", "authors": ["Xingyao Wang", "Yangyi Chen", "Lifan Yuan", "Yizhe Zhang", "Yunzhu Li", "Hao Peng", "Heng Ji"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.01030", "doi": null, "topics": ["Language models", "Robotics", "Economics"], "description": "Large Language Model (LLM) agents, capable of performing a broad range of actions, such as invoking tools and controlling robots, show great potential in tackling real-world challenges.", "description_method": "abstract-excerpt", "abstract": "Large Language Model (LLM) agents, capable of performing a broad range of actions, such as invoking tools and controlling robots, show great potential in tackling real-world challenges. LLM agents are typically prompted to produce actions by generating JSON or text in a pre-defined format, which is usually limited by constrained action space (e.g., the scope of pre-defined tools) and restricted flexibility (e.g., inability to compose multiple tools). This work proposes to use executable Python code to consolidate LLM agents' actions into a unified action space (CodeAct). Integrated with a Python interpreter, CodeAct can execute code actions and dynamically revise prior actions or emit new actions upon new observations through multi-turn interactions. Our extensive analysis of 17 LLMs on API-Bank and a newly curated benchmark shows that CodeAct outperforms widely used alternatives (up to 20% higher success rate). The encouraging performance of CodeAct motivates us to build an open-source LLM agent that interacts with environments by executing interpretable code and collaborates with users using natural language. To this end, we collect an instruction-tuning dataset CodeActInstruct that consists of 7k multi-turn interactions using CodeAct. We show that it can be used with existing data to improve models in agent-oriented tasks without compromising their general capability. CodeActAgent, finetuned from Llama2 and Mistral, is integrated with Python interpreter and uniquely tailored to perform sophisticated tasks (e.g., model training) using existing libraries and autonomously self-debug.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-08", "2025-12", "2026-08"], "last_mentioned": "2026-08-11T01:41:04Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.01817", "slug": "d4b5a3d7cb6b8ad1", "title": "LLMs Can't Plan, But Can Help Planning in LLM-Modulo Frameworks", "authors": ["Subbarao Kambhampati", "Karthik Valmeekam", "Lin Guan", "Mudit Verma", "Kaya Stechly", "Siddhant Bhambri", "Lucas Saldyt", "Anil Murthy"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.01817", "doi": null, "topics": ["Language models", "Physics"], "description": "There is considerable confusion about the role of Large Language Models (LLMs) in planning and reasoning tasks.", "description_method": "abstract-excerpt", "abstract": "There is considerable confusion about the role of Large Language Models (LLMs) in planning and reasoning tasks. On one side are over-optimistic claims that LLMs can indeed do these tasks with just the right prompting or self-verification strategies. On the other side are perhaps over-pessimistic claims that all that LLMs are good for in planning/reasoning tasks are as mere translators of the problem specification from one syntactic format to another, and ship the problem off to external symbolic solvers. In this position paper, we take the view that both these extremes are misguided. We argue that auto-regressive LLMs cannot, by themselves, do planning or self-verification (which is after all a form of reasoning), and shed some light on the reasons for misunderstandings in the literature. We will also argue that LLMs should be viewed as universal approximate knowledge sources that have much more meaningful roles to play in planning/reasoning tasks beyond simple front-end/back-end format translators. We present a vision of {\\bf LLM-Modulo Frameworks} that combine the strengths of LLMs with external model-based verifiers in a tighter bi-directional interaction regime. We will show how the models driving the external verifiers themselves can be acquired with the help of LLMs. We will also argue that rather than simply pipelining LLMs and symbolic components, this LLM-Modulo Framework provides a better neuro-symbolic approach that offers tighter integration between LLMs and symbolic components, and allows extending the scope of model-based planning/reasoning regimes towards more flexible knowledge, problem and preference specifications.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-10", "2025-04", "2025-12"], "last_mentioned": "2025-12-09T23:46:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2402.15391", "slug": "2affc9abf435d836", "title": "Genie: Generative Interactive Environments", "authors": ["Jake Bruce", "Michael Dennis", "Ashley Edwards", "Jack Parker-Holder", "Yuge Shi", "Edward Hughes", "Matthew Lai", "Aditi Mavalankar", "Richie Steigerwald", "Chris Apps", "Yusuf Aytar", "Sarah Bechtle", "Feryal Behbahani", "Stephanie Chan", "Nicolas Heess", "Lucy Gonzalez", "Simon Osindero", "Sherjil Ozair", "Scott Reed", "Jingwei Zhang", "Konrad Zolna", "Jeff Clune", "Nando de Freitas", "Satinder Singh", "Tim Rocktäschel"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2402.15391", "doi": null, "topics": ["Psychology", "Networks"], "description": "We introduce Genie, the first generative interactive environment trained in an unsupervised manner from unlabelled Internet videos.", "description_method": "abstract-excerpt", "abstract": "We introduce Genie, the first generative interactive environment trained in an unsupervised manner from unlabelled Internet videos. The model can be prompted to generate an endless variety of action-controllable virtual worlds described through text, synthetic images, photographs, and even sketches. At 11B parameters, Genie can be considered a foundation world model. It is comprised of a spatiotemporal video tokenizer, an autoregressive dynamics model, and a simple and scalable latent action model. Genie enables users to act in the generated environments on a frame-by-frame basis despite training without any ground-truth action labels or other domain-specific requirements typically found in the world model literature. Further the resulting learned latent action space facilitates training agents to imitate behaviors from unseen videos, opening the path for training generalist agents of the future.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-02", "2024-12", "2026-01"], "last_mentioned": "2026-01-29T17:45:52Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2404.13208", "slug": "e85e0eab52621982", "title": "The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions", "authors": ["Eric Wallace", "Kai Xiao", "Reimar Leike", "Lilian Weng", "Johannes Heidecke", "Alex Beutel"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2404.13208", "doi": null, "topics": ["Security & cryptography", "Psychology", "Language models"], "description": "Today's LLMs are susceptible to prompt injections, jailbreaks, and other attacks that allow adversaries to overwrite a model's original instructions with their own malicious prompts.", "description_method": "abstract-excerpt", "abstract": "Today's LLMs are susceptible to prompt injections, jailbreaks, and other attacks that allow adversaries to overwrite a model's original instructions with their own malicious prompts. In this work, we argue that one of the primary vulnerabilities underlying these attacks is that LLMs often consider system prompts (e.g., text from an application developer) to be the same priority as text from untrusted users and third parties. To address this, we propose an instruction hierarchy that explicitly defines how models should behave when instructions of different priorities conflict. We then propose a data generation method to demonstrate this hierarchical instruction following behavior, which teaches LLMs to selectively ignore lower-privileged instructions. We apply this method to GPT-3.5, showing that it drastically increases robustness -- even for attack types not seen during training -- while imposing minimal degradations on standard capabilities.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-08", "2025-05", "2026-06"], "last_mentioned": "2026-06-22T20:01:13Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2405.14831", "slug": "8038d9a28660c926", "title": "HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models", "authors": ["Bernal Jiménez Gutiérrez", "Yiheng Shu", "Yu Gu", "Michihiro Yasunaga", "Yu Su"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2405.14831", "doi": null, "topics": ["Language models", "Neuroscience", "Biology & medicine"], "description": "In order to thrive in hostile and ever-changing natural environments, mammalian brains evolved to store large amounts of knowledge about the world and continually integrate new information while avoiding catastrophic forgetting.", "description_method": "abstract-excerpt", "abstract": "In order to thrive in hostile and ever-changing natural environments, mammalian brains evolved to store large amounts of knowledge about the world and continually integrate new information while avoiding catastrophic forgetting. Despite the impressive accomplishments, large language models (LLMs), even with retrieval-augmented generation (RAG), still struggle to efficiently and effectively integrate a large amount of new experiences after pre-training. In this work, we introduce HippoRAG, a novel retrieval framework inspired by the hippocampal indexing theory of human long-term memory to enable deeper and more efficient knowledge integration over new experiences. HippoRAG synergistically orchestrates LLMs, knowledge graphs, and the Personalized PageRank algorithm to mimic the different roles of neocortex and hippocampus in human memory. We compare HippoRAG with existing RAG methods on multi-hop question answering and show that our method outperforms the state-of-the-art methods remarkably, by up to 20%. Single-step retrieval with HippoRAG achieves comparable or better performance than iterative retrieval like IRCoT while being 10-30 times cheaper and 6-13 times faster, and integrating HippoRAG into IRCoT brings further substantial gains. Finally, we show that our method can tackle new types of scenarios that are out of reach of existing methods. Code and data are available at https://github.com/OSU-NLP-Group/HippoRAG.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-01", "2025-02", "2026-04"], "last_mentioned": "2026-04-06T23:24:17Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2406.07016", "slug": "3d58c956b7da4516", "title": "Delving into LLM-assisted writing in biomedical publications through excess vocabulary", "authors": ["Dmitry Kobak", "Rita González-Márquez", "Emőke-Ágnes Horvát", "Jan Lause"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2406.07016", "doi": "10.1126/sciadv.adt3813", "topics": ["Biology & medicine", "Language models", "Science & research"], "description": "Large language models (LLMs) like ChatGPT can generate and revise text with human-level performance.", "description_method": "abstract-excerpt", "abstract": "Large language models (LLMs) like ChatGPT can generate and revise text with human-level performance. These models come with clear limitations: they can produce inaccurate information, reinforce existing biases, and be easily misused. Yet, many scientists use them for their scholarly writing. But how wide-spread is such LLM usage in the academic literature? To answer this question for the field of biomedical research, we present an unbiased, large-scale approach: we study vocabulary changes in over 15 million biomedical abstracts from 2010--2024 indexed by PubMed, and show how the appearance of LLMs led to an abrupt increase in the frequency of certain style words. This excess word analysis suggests that at least 13.5% of 2024 abstracts were processed with LLMs. This lower bound differed across disciplines, countries, and journals, reaching 40% for some subcorpora. We show that LLMs have had an unprecedented impact on scientific writing in biomedical research, surpassing the effect of major world events such as the Covid pandemic.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-07", "2024-08", "2026-01"], "last_mentioned": "2026-01-16T14:57:51Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2407.01449", "slug": "2f87b0ea2fa1cf6f", "title": "ColPali: Efficient Document Retrieval with Vision Language Models", "authors": ["Manuel Faysse", "Hugues Sibille", "Tony Wu", "Bilel Omrani", "Gautier Viaud", "Céline Hudelot", "Pierre Colombo"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2407.01449", "doi": null, "topics": ["Language models", "Algorithms"], "description": "Documents are visually rich structures that convey information through text, but also figures, page layouts, tables, or even fonts.", "description_method": "abstract-excerpt", "abstract": "Documents are visually rich structures that convey information through text, but also figures, page layouts, tables, or even fonts. Since modern retrieval systems mainly rely on the textual information they extract from document pages to index documents -often through lengthy and brittle processes-, they struggle to exploit key visual cues efficiently. This limits their capabilities in many practical document retrieval applications such as Retrieval Augmented Generation (RAG). To benchmark current systems on visually rich document retrieval, we introduce the Visual Document Retrieval Benchmark ViDoRe, composed of various page-level retrieval tasks spanning multiple domains, languages, and practical settings. The inherent complexity and performance shortcomings of modern systems motivate a new concept; doing document retrieval by directly embedding the images of the document pages. We release ColPali, a Vision Language Model trained to produce high-quality multi-vector embeddings from images of document pages. Combined with a late interaction matching mechanism, ColPali largely outperforms modern document retrieval pipelines while being drastically simpler, faster and end-to-end trainable. We release models, data, code and benchmarks under open licenses at https://hf.co/vidore.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-07", "2024-10", "2025-01"], "last_mentioned": "2025-01-14T17:37:06Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2409.13373", "slug": "377b806e283f875a", "title": "LLMs Still Can't Plan; Can LRMs? A Preliminary Evaluation of OpenAI's o1 on PlanBench", "authors": ["Karthik Valmeekam", "Kaya Stechly", "Subbarao Kambhampati"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2409.13373", "doi": null, "topics": ["Physics", "Language models"], "description": "The ability to plan a course of action that achieves a desired state of affairs has long been considered a core competence of intelligent agents and has been an integral part of AI research since its inception.", "description_method": "abstract-excerpt", "abstract": "The ability to plan a course of action that achieves a desired state of affairs has long been considered a core competence of intelligent agents and has been an integral part of AI research since its inception. With the advent of large language models (LLMs), there has been considerable interest in the question of whether or not they possess such planning abilities. PlanBench, an extensible benchmark we developed in 2022, soon after the release of GPT3, has remained an important tool for evaluating the planning abilities of LLMs. Despite the slew of new private and open source LLMs since GPT3, progress on this benchmark has been surprisingly slow. OpenAI claims that their recent o1 (Strawberry) model has been specifically constructed and trained to escape the normal limitations of autoregressive LLMs--making it a new kind of model: a Large Reasoning Model (LRM). Using this development as a catalyst, this paper takes a comprehensive look at how well current LLMs and new LRMs do on PlanBench. As we shall see, while o1's performance is a quantum improvement on the benchmark, outpacing the competition, it is still far from saturating it. This improvement also brings to the fore questions about accuracy, efficiency, and guarantees which must be considered before deploying such systems.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-09", "2024-10", "2025-01"], "last_mentioned": "2025-01-01T02:28:31Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2412.08821", "slug": "37ce7af1181949f7", "title": "Large Concept Models: Language Modeling in a Sentence Representation Space", "authors": [" LCM team", "Loïc Barrault", "Paul-Ambroise Duquenne", "Maha Elbayad", "Artyom Kozhevnikov", "Belen Alastruey", "Pierre Andrews", "Mariano Coria", "Guillaume Couairon", "Marta R. Costa-jussà", "David Dale", "Hady Elsahar", "Kevin Heffernan", "João Maria Janeiro", "Tuan Tran", "Christophe Ropers", "Eduardo Sánchez", "Robin San Roman", "Alexandre Mourachko", "Safiyyah Saleem", "Holger Schwenk"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2412.08821", "doi": null, "topics": ["Language models", "Statistics & probability", "Psychology"], "description": "LLMs have revolutionized the field of artificial intelligence and have emerged as the de-facto tool for many tasks.", "description_method": "abstract-excerpt", "abstract": "LLMs have revolutionized the field of artificial intelligence and have emerged as the de-facto tool for many tasks. The current established technology of LLMs is to process input and generate output at the token level. This is in sharp contrast to humans who operate at multiple levels of abstraction, well beyond single words, to analyze information and to generate creative content. In this paper, we present an attempt at an architecture which operates on an explicit higher-level semantic representation, which we name a concept. Concepts are language- and modality-agnostic and represent a higher level idea or action in a flow. Hence, we build a \"Large Concept Model\". In this study, as proof of feasibility, we assume that a concept corresponds to a sentence, and use an existing sentence embedding space, SONAR, which supports up to 200 languages in both text and speech modalities. The Large Concept Model is trained to perform autoregressive sentence prediction in an embedding space. We explore multiple approaches, namely MSE regression, variants of diffusion-based generation, and models operating in a quantized SONAR space. These explorations are performed using 1.6B parameter models and training data in the order of 1.3T tokens. We then scale one architecture to a model size of 7B parameters and training data of about 2.7T tokens. We perform an experimental evaluation on several generative tasks, namely summarization and a new task of summary expansion. Finally, we show that our model exhibits impressive zero-shot generalization performance to many languages, outperforming existing LLMs of the same size. The training code of our models is freely available.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-03", "2025-06", "2026-01"], "last_mentioned": "2026-01-09T17:55:32Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2412.16339", "slug": "c72b10fb73fccbf4", "title": "Deliberative Alignment: Reasoning Enables Safer Language Models", "authors": ["Melody Y. Guan", "Manas Joglekar", "Eric Wallace", "Saachi Jain", "Boaz Barak", "Alec Helyar", "Rachel Dias", "Andrea Vallone", "Hongyu Ren", "Jason Wei", "Hyung Won Chung", "Sam Toyer", "Johannes Heidecke", "Alex Beutel", "Amelia Glaese"], "year": 2024, "venue": "arXiv", "url": "https://arxiv.org/abs/2412.16339", "doi": null, "topics": ["Language models", "Machine learning"], "description": "As large-scale language models increasingly impact safety-critical domains, ensuring their reliable adherence to well-defined principles remains a fundamental challenge.", "description_method": "abstract-excerpt", "abstract": "As large-scale language models increasingly impact safety-critical domains, ensuring their reliable adherence to well-defined principles remains a fundamental challenge. We introduce Deliberative Alignment, a new paradigm that directly teaches the model safety specifications and trains it to explicitly recall and accurately reason over the specifications before answering. We used this approach to align OpenAI's o-series models, and achieved highly precise adherence to OpenAI's safety policies, without requiring human-written chain-of-thoughts or answers. Deliberative Alignment pushes the Pareto frontier by simultaneously increasing robustness to jailbreaks while decreasing overrefusal rates, and also improves out-of-distribution generalization. We demonstrate that reasoning over explicitly specified policies enables more scalable, trustworthy, and interpretable alignment.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-05", "2026-05", "2026-06"], "last_mentioned": "2026-06-27T13:21:14Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2502.18600", "slug": "09b0331c2a41091e", "title": "Chain of Draft: Thinking Faster by Writing Less", "authors": ["Silei Xu", "Wenhao Xie", "Lingxiao Zhao", "Pengcheng He"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2502.18600", "doi": null, "topics": ["Neuroscience", "Language models"], "description": "Large Language Models (LLMs) have demonstrated remarkable performance in solving complex reasoning tasks through mechanisms like Chain-of-Thought (CoT) prompting, which emphasizes verbose, step-by-step reasoning.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) have demonstrated remarkable performance in solving complex reasoning tasks through mechanisms like Chain-of-Thought (CoT) prompting, which emphasizes verbose, step-by-step reasoning. However, humans typically employ a more efficient strategy: drafting concise intermediate thoughts that capture only essential information. In this work, we propose Chain of Draft (CoD), a novel paradigm inspired by human cognitive processes, where LLMs generate minimalistic yet informative intermediate reasoning outputs while solving tasks. By reducing verbosity and focusing on critical insights, CoD matches or surpasses CoT in accuracy while using as little as only 7.6% of the tokens, significantly reducing cost and latency across various reasoning tasks. Our code and data are available at https://github.com/sileix/chain-of-draft.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-03", "2025-05", "2025-12"], "last_mentioned": "2025-12-22T01:40:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2504.13837", "slug": "b90f70269aeb0231", "title": "Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?", "authors": ["Yang Yue", "Zhiqi Chen", "Rui Lu", "Andrew Zhao", "Zhaokai Wang", "Yang Yue", "Shiji Song", "Gao Huang"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2504.13837", "doi": null, "topics": ["Machine learning", "Algorithms", "Programming languages"], "description": "Reinforcement Learning with Verifiable Rewards (RLVR) has recently demonstrated notable success in enhancing the reasoning performance of large language models (LLMs), particularly on mathematics and programming tasks.", "description_method": "abstract-excerpt", "abstract": "Reinforcement Learning with Verifiable Rewards (RLVR) has recently demonstrated notable success in enhancing the reasoning performance of large language models (LLMs), particularly on mathematics and programming tasks. Similar to how traditional RL helps agents explore and learn new strategies, RLVR is believed to enable LLMs to continuously self-improve, thus acquiring novel reasoning abilities beyond those of the corresponding base models. In this study we critically examine the current state of RLVR by systematically probing the reasoning capability boundaries of RLVR-trained LLMs across various model families, RL algorithms, and math, coding, and visual reasoning benchmarks, using pass@k at large k values as the evaluation metric. Surprisingly, we find that the current training setup does not elicit fundamentally new reasoning patterns. While RLVR-trained models outperform their base models at small k (e.g., k = 1), the base models achieve a higher pass@k score when k is large. Coverage and perplexity analyses show that the observed reasoning abilities originate from and are bounded by the base model. Treating the base model as an upper bound, our quantitative analysis shows that six popular RLVR algorithms perform similarly and remain far from optimal in leveraging the potential of the base model. By contrast, we find that distillation can introduce new reasoning patterns from the teacher and genuinely expand the model's reasoning capabilities. Overall, our findings suggest that current RLVR methods have not yet realized the potential of RL to elicit truly novel reasoning abilities in LLMs. This highlights the need for improved RL paradigms, such as continual scaling and multi-turn agent-environment interaction, to unlock this potential.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-05", "2026-03", "2026-07"], "last_mentioned": "2026-07-29T19:57:12Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2505.06120", "slug": "aa07ebbcc70892bd", "title": "LLMs Get Lost In Multi-Turn Conversation", "authors": ["Philippe Laban", "Hiroaki Hayashi", "Yingbo Zhou", "Jennifer Neville"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2505.06120", "doi": null, "topics": ["Language models"], "description": "Large Language Models (LLMs) are conversational interfaces.", "description_method": "abstract-excerpt", "abstract": "Large Language Models (LLMs) are conversational interfaces. As such, LLMs have the potential to assist their users not only when they can fully specify the task at hand, but also to help them define, explore, and refine what they need through multi-turn conversational exchange. Although analysis of LLM conversation logs has confirmed that underspecification occurs frequently in user instructions, LLM evaluation has predominantly focused on the single-turn, fully-specified instruction setting. In this work, we perform large-scale simulation experiments to compare LLM performance in single- and multi-turn settings. Our experiments confirm that all the top open- and closed-weight LLMs we test exhibit significantly lower performance in multi-turn conversations than single-turn, with an average drop of 39% across six generation tasks. Analysis of 200,000+ simulated conversations decomposes the performance degradation into two components: a minor loss in aptitude and a significant increase in unreliability. We find that LLMs often make assumptions in early turns and prematurely attempt to generate final solutions, on which they overly rely. In simpler terms, we discover that *when LLMs take a wrong turn in a conversation, they get lost and do not recover*.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-05", "2025-07", "2025-12"], "last_mentioned": "2025-12-08T16:48:20Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2505.09388", "slug": "9f8e189534ce30a5", "title": "Qwen3 Technical Report", "authors": ["An Yang", "Anfeng Li", "Baosong Yang", "Beichen Zhang", "Binyuan Hui", "Bo Zheng", "Bowen Yu", "Chang Gao", "Chengen Huang", "Chenxu Lv", "Chujie Zheng", "Dayiheng Liu", "Fan Zhou", "Fei Huang", "Feng Hu", "Hao Ge", "Haoran Wei", "Huan Lin", "Jialong Tang", "Jian Yang", "Jianhong Tu", "Jianwei Zhang", "Jianxin Yang", "Jiaxi Yang", "Jing Zhou", "Jingren Zhou", "Junyang Lin", "Kai Dang", "Keqin Bao", "Kexin Yang", "Le Yu", "Lianghao Deng", "Mei Li", "Mingfeng Xue", "Mingze Li", "Pei Zhang", "Peng Wang", "Qin Zhu", "Rui Men", "Ruize Gao", "Shixuan Liu", "Shuang Luo", "Tianhao Li", "Tianyi Tang", "Wenbiao Yin", "Xingzhang Ren", "Xinyu Wang", "Xinyu Zhang", "Xuancheng Ren", "Yang Fan", "Yang Su", "Yichang Zhang", "Yinger Zhang", "Yu Wan", "Yuqiong Liu", "Zekun Wang", "Zeyu Cui", "Zhenru Zhang", "Zhipeng Zhou", "Zihan Qiu"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2505.09388", "doi": null, "topics": ["Language models", "Science & research", "Mathematics"], "description": "In this work, we present Qwen3, the latest version of the Qwen model family.", "description_method": "abstract-excerpt", "abstract": "In this work, we present Qwen3, the latest version of the Qwen model family. Qwen3 comprises a series of large language models (LLMs) designed to advance performance, efficiency, and multilingual capabilities. The Qwen3 series includes models of both dense and Mixture-of-Expert (MoE) architectures, with parameter scales ranging from 0.6 to 235 billion. A key innovation in Qwen3 is the integration of thinking mode (for complex, multi-step reasoning) and non-thinking mode (for rapid, context-driven responses) into a unified framework. This eliminates the need to switch between different models--such as chat-optimized models (e.g., GPT-4o) and dedicated reasoning models (e.g., QwQ-32B)--and enables dynamic mode switching based on user queries or chat templates. Meanwhile, Qwen3 introduces a thinking budget mechanism, allowing users to allocate computational resources adaptively during inference, thereby balancing latency and performance based on task complexity. Moreover, by leveraging the knowledge from the flagship models, we significantly reduce the computational resources required to build smaller-scale models, while ensuring their highly competitive performance. Empirical evaluations demonstrate that Qwen3 achieves state-of-the-art results across diverse benchmarks, including tasks in code generation, mathematical reasoning, agent tasks, etc., competitive against larger MoE models and proprietary models. Compared to its predecessor Qwen2.5, Qwen3 expands multilingual support from 29 to 119 languages and dialects, enhancing global accessibility through improved cross-lingual understanding and generation capabilities. To facilitate reproducibility and community-driven research and development, all Qwen3 models are publicly accessible under Apache 2.0.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2026-02", "2026-05", "2026-07"], "last_mentioned": "2026-07-25T18:27:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2505.11831", "slug": "3f3d1a5d96309193", "title": "ARC-AGI-2: A New Challenge for Frontier AI Reasoning Systems", "authors": ["Francois Chollet", "Mike Knoop", "Gregory Kamradt", "Bryan Landers", "Henry Pinkard"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2505.11831", "doi": null, "topics": ["Psychology", "Neuroscience", "Algorithms"], "description": "The Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI), introduced in 2019, established a challenging benchmark for evaluating the general fluid intelligence of artificial systems via a set of unique, novel tasks only requiring minimal prior knowledge.", "description_method": "abstract-excerpt", "abstract": "The Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI), introduced in 2019, established a challenging benchmark for evaluating the general fluid intelligence of artificial systems via a set of unique, novel tasks only requiring minimal prior knowledge. While ARC-AGI has spurred significant research activity over the past five years, recent AI progress calls for benchmarks capable of finer-grained evaluation at higher levels of cognitive complexity. We introduce ARC-AGI-2, an upgraded version of the benchmark. ARC-AGI-2 preserves the input-output pair task format of its predecessor, ensuring continuity for researchers. It incorporates a newly curated and expanded set of tasks specifically designed to provide a more granular signal to assess abstract reasoning and problem-solving abilities at higher levels of fluid intelligence. To contextualize the difficulty and characteristics of ARC-AGI-2, we present extensive results from human testing, providing a robust baseline that highlights the benchmark's accessibility to human intelligence, yet difficulty for current AI systems. ARC-AGI-2 aims to serve as a next-generation tool for rigorously measuring progress towards more general and human-like AI capabilities.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-06", "2025-07", "2026-07"], "last_mentioned": "2026-07-09T02:19:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2508.18475", "slug": "68360efa3f2f6459", "title": "A convex polyhedron without Rupert's property", "authors": ["Jakob Steininger", "Sergey Yurkevich"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2508.18475", "doi": null, "topics": ["Other research"], "description": "A three-dimensional convex body is said to have Rupert's property if its copy can be passed through a straight hole inside that body.", "description_method": "abstract-excerpt", "abstract": "A three-dimensional convex body is said to have Rupert's property if its copy can be passed through a straight hole inside that body. In this work we construct a polyhedron which is provably not Rupert, thus we disprove a conjecture from 2017. We also find a polyhedron that is Rupert but not locally Rupert.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-10", "2025-11", "2025-12"], "last_mentioned": "2025-12-07T07:38:34Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2510.04871", "slug": "4345eebe9dc7e67f", "title": "Less is More: Recursive Reasoning with Tiny Networks", "authors": ["Alexia Jolicoeur-Martineau"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2510.04871", "doi": null, "topics": ["Machine learning", "Language models", "Biology & medicine"], "description": "Hierarchical Reasoning Model (HRM) is a novel approach using two small neural networks recursing at different frequencies.", "description_method": "abstract-excerpt", "abstract": "Hierarchical Reasoning Model (HRM) is a novel approach using two small neural networks recursing at different frequencies. This biologically inspired method beats Large Language models (LLMs) on hard puzzle tasks such as Sudoku, Maze, and ARC-AGI while trained with small models (27M parameters) on small data (around 1000 examples). HRM holds great promise for solving hard problems with small networks, but it is not yet well understood and may be suboptimal. We propose Tiny Recursive Model (TRM), a much simpler recursive reasoning approach that achieves significantly higher generalization than HRM, while using a single tiny network with only 2 layers. With only 7M parameters, TRM obtains 45% test-accuracy on ARC-AGI-1 and 8% on ARC-AGI-2, higher than most LLMs (e.g., Deepseek R1, o3-mini, Gemini 2.5 Pro) with less than 0.01% of the parameters.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-11", "2026-06", "2026-07"], "last_mentioned": "2026-07-23T20:50:27Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2510.08338", "slug": "0a3300ee50ea95ca", "title": "LLMs Reproduce Human Purchase Intent via Semantic Similarity Elicitation of Likert Ratings", "authors": ["Benjamin F. Maier", "Ulf Aslak", "Luca Fiaschi", "Nina Rismal", "Kemble Fletcher", "Christian C. Luhmann", "Robbie Dow", "Kli Pappas", "Thomas V. Wiecki"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2510.08338", "doi": null, "topics": ["Psychology", "Language models", "Economics"], "description": "Consumer research costs companies billions annually yet suffers from panel biases and limited scale.", "description_method": "abstract-excerpt", "abstract": "Consumer research costs companies billions annually yet suffers from panel biases and limited scale. Large language models (LLMs) offer an alternative by simulating synthetic consumers, but produce unrealistic response distributions when asked directly for numerical ratings. We present semantic similarity rating (SSR), a method that elicits textual responses from LLMs and maps these to Likert distributions using embedding similarity to reference statements. Testing on an extensive dataset comprising 57 personal care product surveys conducted by a leading corporation in that market (9,300 human responses), SSR achieves 90% of human test-retest reliability while maintaining realistic response distributions (KS similarity > 0.85). Additionally, these synthetic respondents provide rich qualitative feedback explaining their ratings. This framework enables scalable consumer research simulations while preserving traditional survey metrics and interpretability.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-10", "2025-11", "2026-07"], "last_mentioned": "2026-07-08T13:19:48Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2510.12269", "slug": "6376f4e63527d6a9", "title": "Tensor Logic: The Language of AI", "authors": ["Pedro Domingos"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2510.12269", "doi": null, "topics": ["Statistics & probability", "Programming languages", "Machine learning"], "description": "Progress in AI is hindered by the lack of a programming language with all the requisite features.", "description_method": "abstract-excerpt", "abstract": "Progress in AI is hindered by the lack of a programming language with all the requisite features. Libraries like PyTorch and TensorFlow provide automatic differentiation and efficient GPU implementation, but are additions to Python, which was never intended for AI. Their lack of support for automated reasoning and knowledge acquisition has led to a long and costly series of hacky attempts to tack them on. On the other hand, AI languages like LISP and Prolog lack scalability and support for learning. This paper proposes tensor logic, a language that solves these problems by unifying neural and symbolic AI at a fundamental level. The sole construct in tensor logic is the tensor equation, based on the observation that logical rules and Einstein summation are essentially the same operation, and all else can be reduced to them. I show how to elegantly implement key forms of neural, symbolic and statistical AI in tensor logic, including transformers, formal reasoning, kernel machines and graphical models. Most importantly, tensor logic makes new directions possible, such as sound reasoning in embedding space. This combines the scalability and learnability of neural networks with the reliability and transparency of symbolic reasoning, and is potentially a basis for the wider adoption of AI.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2026-01", "2026-04", "2026-07"], "last_mentioned": "2026-07-14T04:51:30Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2511.09030", "slug": "b9a32aa071d9c8fc", "title": "Solving a Million-Step LLM Task with Zero Errors", "authors": ["Elliot Meyerson", "Giuseppe Paolo", "Roberto Dailey", "Hormoz Shahrzad", "Olivier Francon", "Conor F. Hayes", "Xin Qiu", "Babak Hodjat", "Risto Miikkulainen"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2511.09030", "doi": null, "topics": ["Language models"], "description": "LLMs have achieved remarkable breakthroughs in reasoning, insights, and tool use, but chaining these abilities into extended processes at the scale of those routinely executed by humans, organizations, and societies has remained out of reach.", "description_method": "abstract-excerpt", "abstract": "LLMs have achieved remarkable breakthroughs in reasoning, insights, and tool use, but chaining these abilities into extended processes at the scale of those routinely executed by humans, organizations, and societies has remained out of reach. The models have a persistent error rate that prevents scale-up: for instance, recent experiments in the Towers of Hanoi benchmark domain showed that the process inevitably becomes derailed after at most a few hundred steps. Thus, although LLM research is often still benchmarked on tasks with relatively few dependent logical steps, there is increasing attention on the ability (or inability) of LLMs to perform long range tasks. This paper describes MAKER, the first system that successfully solves a task with over one million LLM steps with zero errors, and, in principle, scales far beyond this level. The approach relies on an extreme decomposition of a task into subtasks, each of which can be tackled by focused microagents. The high level of modularity resulting from the decomposition allows error correction to be applied at each step through an efficient multi-agent voting scheme. This combination of extreme decomposition and error correction makes scaling possible. Thus, the results suggest that instead of relying on continual improvement of current LLMs, massively decomposed agentic processes (MDAPs) may provide a way to efficiently solve problems at the level of organizations and societies.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-11", "2025-12", "2026-02"], "last_mentioned": "2026-02-05T19:10:54Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2512.02556", "slug": "cb7afc75b4f9d392", "title": "DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models", "authors": [" DeepSeek-AI", "Aixin Liu", "Aoxue Mei", "Bangcai Lin", "Bing Xue", "Bingxuan Wang", "Bingzheng Xu", "Bochao Wu", "Bowei Zhang", "Chaofan Lin", "Chen Dong", "Chengda Lu", "Chenggang Zhao", "Chengqi Deng", "Chenhao Xu", "Chong Ruan", "Damai Dai", "Daya Guo", "Dejian Yang", "Deli Chen", "Erhang Li", "Fangqi Zhou", "Fangyun Lin", "Fucong Dai", "Guangbo Hao", "Guanting Chen", "Guowei Li", "H. Zhang", "Hanwei Xu", "Hao Li", "Haofen Liang", "Haoran Wei", "Haowei Zhang", "Haowen Luo", "Haozhe Ji", "Honghui Ding", "Hongxuan Tang", "Huanqi Cao", "Huazuo Gao", "Hui Qu", "Hui Zeng", "Jialiang Huang", "Jiashi Li", "Jiaxin Xu", "Jiewen Hu", "Jingchang Chen", "Jingting Xiang", "Jingyang Yuan", "Jingyuan Cheng", "Jinhua Zhu", "Jun Ran", "Junguang Jiang", "Junjie Qiu", "Junlong Li", "Junxiao Song", "Kai Dong", "Kaige Gao", "Kang Guan", "Kexin Huang", "Kexing Zhou", "Kezhao Huang", "Kuai Yu", "Lean Wang", "Lecong Zhang", "Lei Wang", "Liang Zhao", "Liangsheng Yin", "Lihua Guo", "Lingxiao Luo", "Linwang Ma", "Litong Wang", "Liyue Zhang", "M. S. Di", "M. Y Xu", "Mingchuan Zhang", "Minghua Zhang", "Minghui Tang", "Mingxu Zhou", "Panpan Huang", "Peixin Cong", "Peiyi Wang", "Qiancheng Wang", "Qihao Zhu", "Qingyang Li", "Qinyu Chen", "Qiushi Du", "Ruiling Xu", "Ruiqi Ge", "Ruisong Zhang", "Ruizhe Pan", "Runji Wang", "Runqiu Yin", "Runxin Xu", "Ruomeng Shen", "Ruoyu Zhang", "S. H. Liu", "Shanghao Lu", "Shangyan Zhou", "Shanhuang Chen", "Shaofei Cai", "Shaoyuan Chen", "Shengding Hu", "Shengyu Liu", "Shiqiang Hu", "Shirong Ma", "Shiyu Wang", "Shuiping Yu", "Shunfeng Zhou", "Shuting Pan", "Songyang Zhou", "Tao Ni", "Tao Yun", "Tian Pei", "Tian Ye", "Tianyuan Yue", "Wangding Zeng", "Wen Liu", "Wenfeng Liang", "Wenjie Pang", "Wenjing Luo", "Wenjun Gao", "Wentao Zhang", "Xi Gao", "Xiangwen Wang", "Xiao Bi", "Xiaodong Liu", "Xiaohan Wang", "Xiaokang Chen", "Xiaokang Zhang", "Xiaotao Nie", "Xin Cheng", "Xin Liu", "Xin Xie", "Xingchao Liu", "Xingkai Yu", "Xingyou Li", "Xinyu Yang", "Xinyuan Li", "Xu Chen", "Xuecheng Su", "Xuehai Pan", "Xuheng Lin", "Xuwei Fu", "Y. Q. Wang", "Yang Zhang", "Yanhong Xu", "Yanru Ma", "Yao Li", "Yao Li", "Yao Zhao", "Yaofeng Sun", "Yaohui Wang", "Yi Qian", "Yi Yu", "Yichao Zhang", "Yifan Ding", "Yifan Shi", "Yiliang Xiong", "Ying He", "Ying Zhou", "Yinmin Zhong", "Yishi Piao", "Yisong Wang", "Yixiao Chen", "Yixuan Tan", "Yixuan Wei", "Yiyang Ma", "Yiyuan Liu", "Yonglun Yang", "Yongqiang Guo", "Yongtong Wu", "Yu Wu", "Yuan Cheng", "Yuan Ou", "Yuanfan Xu", "Yuduan Wang", "Yue Gong", "Yuhan Wu", "Yuheng Zou", "Yukun Li", "Yunfan Xiong", "Yuxiang Luo", "Yuxiang You", "Yuxuan Liu", "Yuyang Zhou", "Z. F. Wu", "Z. Z. Ren", "Zehua Zhao", "Zehui Ren", "Zhangli Sha", "Zhe Fu", "Zhean Xu", "Zhenda Xie", "Zhengyan Zhang", "Zhewen Hao", "Zhibin Gou", "Zhicheng Ma", "Zhigang Yan", "Zhihong Shao", "Zhixian Huang", "Zhiyu Wu", "Zhuoshu Li", "Zhuping Zhang", "Zian Xu", "Zihao Wang", "Zihui Gu", "Zijia Zhu", "Zilin Li", "Zipeng Zhang", "Ziwei Xie", "Ziyi Gao", "Zizheng Pan", "Zongqing Yao", "Bei Feng", "Hui Li", "J. L. Cai", "Jiaqi Ni", "Lei Xu", "Meng Li", "Ning Tian", "R. J. Chen", "R. L. Jin", "S. S. Li", "Shuang Zhou", "Tianyu Sun", "X. Q. Li", "Xiangyue Jin", "Xiaojin Shen", "Xiaosha Chen", "Xinnan Song", "Xinyi Zhou", "Y. X. Zhu", "Yanping Huang", "Yaohui Li", "Yi Zheng", "Yuchen Zhu", "Yunxian Ma", "Zhen Huang", "Zhipeng Xu", "Zhongyu Zhang", "Dongjie Ji", "Jian Liang", "Jianzhong Guo", "Jin Chen", "Leyi Xia", "Miaojun Wang", "Mingming Li", "Peng Zhang", "Ruyi Chen", "Shangmian Sun", "Shaoqing Wu", "Shengfeng Ye", "T. Wang", "W. L. Xiao", "Wei An", "Xianzu Wang", "Xiaowen Sun", "Xiaoxiang Wang", "Ying Tang", "Yukun Zha", "Zekai Zhang", "Zhe Ju", "Zhen Zhang", "Zihua Qu"], "year": 2025, "venue": "arXiv", "url": "https://arxiv.org/abs/2512.02556", "doi": null, "topics": ["Language models", "Machine learning", "Mathematics"], "description": "We introduce DeepSeek-V3.2, a model that harmonizes high computational efficiency with superior reasoning and agent performance.", "description_method": "abstract-excerpt", "abstract": "We introduce DeepSeek-V3.2, a model that harmonizes high computational efficiency with superior reasoning and agent performance. The key technical breakthroughs of DeepSeek-V3.2 are as follows: (1) DeepSeek Sparse Attention (DSA): We introduce DSA, an efficient attention mechanism that substantially reduces computational complexity while preserving model performance in long-context scenarios. (2) Scalable Reinforcement Learning Framework: By implementing a robust reinforcement learning protocol and scaling post-training compute, DeepSeek-V3.2 performs comparably to GPT-5. Notably, our high-compute variant, DeepSeek-V3.2-Speciale, surpasses GPT-5 and exhibits reasoning proficiency on par with Gemini-3.0-Pro, achieving gold-medal performance in both the 2025 International Mathematical Olympiad (IMO) and the International Olympiad in Informatics (IOI). (3) Large-Scale Agentic Task Synthesis Pipeline: To integrate reasoning into tool-use scenarios, we developed a novel synthesis pipeline that systematically generates training data at scale. This methodology facilitates scalable agentic post-training, yielding substantial improvements in generalization and instruction-following robustness within complex, interactive environments.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2026-02", "2026-05", "2026-06"], "last_mentioned": "2026-06-06T06:55:00Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2603.03823", "slug": "fc91613deec3ebc0", "title": "SWE-CI: Evaluating Agent Capabilities in Maintaining Codebases via Continuous Integration", "authors": ["Jialong Chen", "Xander Xu", "Hu Wei", "Chuan Chen", "Bing Zhao"], "year": 2026, "venue": "arXiv", "url": "https://arxiv.org/abs/2603.03823", "doi": null, "topics": ["Language models", "Programming languages"], "description": "Large language model (LLM)-powered agents have demonstrated strong capabilities in automating software engineering tasks such as static bug fixing.", "description_method": "abstract-excerpt", "abstract": "Large language model (LLM)-powered agents have demonstrated strong capabilities in automating software engineering tasks such as static bug fixing. However, in the real world, the development of mature software is typically predicated on complex requirement changes and long-term feature iterations -- a process that static, one-shot repair paradigms fail to capture. To bridge this gap, we propose SWE-CI, the first repository-level benchmark built upon the Continuous Integration loop, aiming to shift the evaluation paradigm for code generation from static, short-term functional correctness toward dynamic, long-term maintainability. The key insight is simple: Maintainability can be revealed by tracking how functional correctness changes over time. The benchmark comprises 100 tasks, each deriving from a real-world code repository with a development history spanning an average of 233 days and 71 consecutive commits. SWE-CI requires agents to systematically resolve these tasks through dozens of rounds of analysis and coding iterations. SWE-CI provides valuable insights into how well agents can sustain code quality throughout long-term evolution.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2026-03", "2026-05", "2026-06"], "last_mentioned": "2026-06-22T11:41:56Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:2603.21687", "slug": "f9dfb4aabb765dcc", "title": "MIRAGE: The Illusion of Visual Understanding", "authors": ["Mohammad Asadi", "Jack W. O'Sullivan", "Fang Cao", "Tahoura Nedaee", "Kamyar Rajabalifardi", "Fei-Fei Li", "Ehsan Adeli", "Euan Ashley"], "year": 2026, "venue": "arXiv", "url": "https://arxiv.org/abs/2603.21687", "doi": null, "topics": ["Biology & medicine", "Psychology", "Language models"], "description": "Multimodal AI systems have achieved remarkable performance across a broad range of real-world tasks, yet the mechanisms underlying visual-language reasoning remain surprisingly poorly understood.", "description_method": "abstract-excerpt", "abstract": "Multimodal AI systems have achieved remarkable performance across a broad range of real-world tasks, yet the mechanisms underlying visual-language reasoning remain surprisingly poorly understood. We report three findings that challenge prevailing assumptions about how these systems process and integrate visual information. First, Frontier models readily generate detailed image descriptions and elaborate reasoning traces, including pathology-biased clinical findings, for images never provided; we term this phenomenon mirage reasoning. Second, without any image input, models also attain strikingly high scores across general and medical multimodal benchmarks, bringing into question their utility and design. In the most extreme case, our model achieved the top rank on a standard chest X-ray question-answering benchmark without access to any images. Third, when models were explicitly instructed to guess answers without image access, rather than being implicitly prompted to assume images were present, performance declined markedly. Explicit guessing appears to engage a more conservative response regime, in contrast to the mirage regime in which models behave as though images have been provided. These findings expose fundamental vulnerabilities in how visual-language models reason and are evaluated, pointing to an urgent need for private benchmarks that eliminate textual cues enabling non-visual inference, particularly in medical contexts where miscalibrated AI carries the greatest consequence. We introduce B-Clean as a principled solution for fair, vision-grounded evaluation of multimodal AI systems.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2026-03", "2026-05", "2026-06"], "last_mentioned": "2026-06-29T13:22:07Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:cs/0004001", "slug": "87dd85e3387c76a9", "title": "A Theory of Universal Artificial Intelligence based on Algorithmic Complexity", "authors": ["Marcus Hutter"], "year": 2000, "venue": "arXiv", "url": "https://arxiv.org/abs/cs/0004001", "doi": null, "topics": ["Psychology", "Algorithms", "Statistics & probability"], "description": "Decision theory formally solves the problem of rational agents in uncertain worlds if the true environmental prior probability distribution is known.", "description_method": "abstract-excerpt", "abstract": "Decision theory formally solves the problem of rational agents in uncertain worlds if the true environmental prior probability distribution is known. Solomonoff's theory of universal induction formally solves the problem of sequence prediction for unknown prior distribution. We combine both ideas and get a parameterless theory of universal Artificial Intelligence. We give strong arguments that the resulting AIXI model is the most intelligent unbiased agent possible. We outline for a number of problem classes, including sequence prediction, strategic games, function minimization, reinforcement and supervised learning, how the AIXI model can formally solve them. The major drawback of the AIXI model is that it is uncomputable. To overcome this problem, we construct a modified algorithm AIXI-tl, which is still effectively more intelligent than any other time t and space l bounded agent. The computation time of AIXI-tl is of the order tx2^l. Other discussed topics are formal definitions of intelligence order relations, the horizon problem and relations of the AIXI theory to other AI approaches.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-01", "2021-06", "2022-08"], "last_mentioned": "2022-08-21T03:22:24Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:gr-qc/9704009", "slug": "fd9655c79f678624", "title": "Is \"the theory of everything'' merely the ultimate ensemble theory?", "authors": ["Max Tegmark"], "year": 1997, "venue": "arXiv", "url": "https://arxiv.org/abs/gr-qc/9704009", "doi": "10.1006/aphy.1998.5855", "topics": ["Statistics & probability", "Physics", "Mathematics"], "description": "We discuss some physical consequences of what might be called ``the ultimate ensemble theory'', where not only worlds corresponding to say different sets of initial data or different physical constants are considered equally real, but also worlds ruled by altogether different equations.", "description_method": "abstract-excerpt", "abstract": "We discuss some physical consequences of what might be called ``the ultimate ensemble theory'', where not only worlds corresponding to say different sets of initial data or different physical constants are considered equally real, but also worlds ruled by altogether different equations. The only postulate in this theory is that all structures that exist mathematically exist also physically, by which we mean that in those complex enough to contain self-aware substructures (SASs), these SASs will subjectively perceive themselves as existing in a physically ``real'' world. We find that it is far from clear that this simple theory, which has no free parameters whatsoever, is observationally ruled out. The predictions of the theory take the form of probability distributions for the outcome of experiments, which makes it testable. In addition, it may be possible to rule it out by comparing its a priori predictions for the observable attributes of nature (the particle masses, the dimensionality of spacetime, etc) with what is observed.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2012-12", "2020-02", "2020-04"], "last_mentioned": "2020-04-15T05:10:15Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:math/0303352", "slug": "87141ca048b1b8c0", "title": "From Philosophy to Program Size", "authors": ["G. J. Chaitin"], "year": 2003, "venue": "arXiv", "url": "https://arxiv.org/abs/math/0303352", "doi": null, "topics": ["Mathematics", "Algorithms", "Statistics & probability"], "description": "Most work on computational complexity is concerned with time.", "description_method": "abstract-excerpt", "abstract": "Most work on computational complexity is concerned with time. However this course will try to show that program-size complexity, which measures algorithmic information, is of much greater philosophical significance. I'll discuss how one can use this complexity measure to study what can and cannot be achieved by formal axiomatic mathematical theories. In particular, I'll show (a) that there are natural information-theoretic constraints on formal axiomatic theories, and that program-size complexity provides an alternative path to incompleteness from the one originally used by Kurt Godel. Furthermore, I'll show (b) that in pure mathematics there are mathematical facts that are true for no reason, that are true by accident. These have to do with determining the successive binary digits of the precise numerical value of the halting probability Omega for a \"self-delimiting\" universal Turing machine. I believe that these meta-theorems (a,b) showing (a) that the complexity of axiomatic theories can be characterized information-theoretically and (b) that God plays dice in pure mathematics, both strongly suggest a quasi-empirical view of mathematics. I.e., math is different from physics, but perhaps not as different as people usually think. I'll also discuss the convergence of theoretical computer science with theoretical physics, Leibniz's ideas on complexity, Stephen Wolfram's book A New Kind of Science, and how to attempt to use information theory to define what a living being is.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-03", "2017-09", "2026-07"], "last_mentioned": "2026-07-27T22:54:41Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:math/0406077", "slug": "14bb84b51b682974", "title": "A tutorial introduction to the minimum description length principle", "authors": ["Peter Grunwald"], "year": 2004, "venue": "arXiv", "url": "https://arxiv.org/abs/math/0406077", "doi": null, "topics": ["Mathematics"], "description": "This tutorial provides an overview of and introduction to Rissanen's Minimum Description Length (MDL) Principle.", "description_method": "abstract-excerpt", "abstract": "This tutorial provides an overview of and introduction to Rissanen's Minimum Description Length (MDL) Principle. The first chapter provides a conceptual, entirely non-technical introduction to the subject. It serves as a basis for the technical introduction given in the second chapter, in which all the ideas of the first chapter are made mathematically precise. The main ideas are discussed in great conceptual and technical detail. This tutorial is an extended version of the first two chapters of the collection \"Advances in Minimum Description Length: Theory and Application\" (edited by P.Grunwald, I.J. Myung and M. Pitt, to be published by the MIT Press, Spring 2005).", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-05", "2026-02", "2026-07"], "last_mentioned": "2026-07-07T19:03:23Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "arxiv:quant-ph/9508027", "slug": "115cb112fdeb5a88", "title": "Polynomial-Time Algorithms for Prime Factorization and Discrete Logarithms on a Quantum Computer", "authors": ["Peter W. Shor"], "year": 1995, "venue": "arXiv", "url": "https://arxiv.org/abs/quant-ph/9508027", "doi": "10.1137/S0097539795293172", "topics": ["Physics", "Algorithms"], "description": "A digital computer is generally believed to be an efficient universal computing device; that is, it is believed able to simulate any physical computing device with an increase in computation time of at most a polynomial factor.", "description_method": "abstract-excerpt", "abstract": "A digital computer is generally believed to be an efficient universal computing device; that is, it is believed able to simulate any physical computing device with an increase in computation time of at most a polynomial factor. This may not be true when quantum mechanics is taken into consideration. This paper considers factoring integers and finding discrete logarithms, two problems which are generally thought to be hard on a classical computer and have been used as the basis of several proposed cryptosystems. Efficient randomized algorithms are given for these two problems on a hypothetical quantum computer. These algorithms take a number of steps polynomial in the input size, e.g., the number of digits of the integer to be factored.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2016-08", "2018-02", "2023-01"], "last_mentioned": "2023-01-03T20:54:16Z", "metadata_source": "arxiv-api", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/35057062", "slug": "14eae3799bd3f2e7", "title": "Initial sequencing and analysis of the human genome", "authors": ["International Human Genome Sequencing Consortium", "Whitehead Institute for Biomedical Research, Center for Genome Research:", "Eric S. Lander", "Lauren M. Linton", "Bruce Birren", "Chad Nusbaum", "Michael C. Zody", "Jennifer Baldwin", "Keri Devon", "Ken Dewar", "Michael Doyle", "William FitzHugh", "Roel Funke", "Diane Gage", "Katrina Harris", "Andrew Heaford", "John Howland", "Lisa Kann", "Jessica Lehoczky", "Rosie LeVine", "Paul McEwan", "Kevin McKernan", "James Meldrim", "Jill P. Mesirov", "Cher Miranda", "William Morris", "Jerome Naylor", "Christina Raymond", "Mark Rosetti", "Ralph Santos", "Andrew Sheridan", "Carrie Sougnez", "Nicole Stange-Thomann", "Nikola Stojanovic", "Aravind Subramanian", "Dudley Wyman", "The Sanger Centre:", "Jane Rogers", "John Sulston", "Rachael Ainscough", "Stephan Beck", "David Bentley", "John Burton", "Christopher Clee", "Nigel Carter", "Alan Coulson", "Rebecca Deadman", "Panos Deloukas", "Andrew Dunham", "Ian Dunham", "Richard Durbin", "Lisa French", "Darren Grafham", "Simon Gregory", "Tim Hubbard", "Sean Humphray", "Adrienne Hunt", "Matthew Jones", "Christine Lloyd", "Amanda McMurray", "Lucy Matthews", "Simon Mercer", "Sarah Milne", "James C. Mullikin", "Andrew Mungall", "Robert Plumb", "Mark Ross", "Ratna Shownkeen", "Sarah Sims", "Washington University Genome Sequencing Center", "Robert H. Waterston", "Richard K. Wilson", "LaDeana W. Hillier", "John D. McPherson", "Marco A. Marra", "Elaine R. Mardis", "Lucinda A. Fulton", "Asif T. Chinwalla", "Kymberlie H. Pepin", "Warren R. Gish", "Stephanie L. Chissoe", "Michael C. Wendl", "Kim D. Delehaunty", "Tracie L. Miner", "Andrew Delehaunty", "Jason B. Kramer", "Lisa L. Cook", "Robert S. Fulton", "Douglas L. Johnson", "Patrick J. Minx", "Sandra W. Clifton", "US DOE Joint Genome Institute:", "Trevor Hawkins", "Elbert Branscomb", "Paul Predki", "Paul Richardson", "Sarah Wenning", "Tom Slezak", "Norman Doggett", "Jan-Fang Cheng", "Anne Olsen", "Susan Lucas", "Christopher Elkin", "Edward Uberbacher", "Marvin Frazier", "Baylor College of Medicine Human Genome Sequencing Center:", "Richard A. Gibbs", "Donna M. Muzny", "Steven E. Scherer", "John B. Bouck", "Erica J. Sodergren", "Kim C. Worley", "Catherine M. Rives", "James H. Gorrell", "Michael L. Metzker", "Susan L. Naylor", "Raju S. Kucherlapati", "David L. Nelson", "George M. Weinstock", "RIKEN Genomic Sciences Center:", "Yoshiyuki Sakaki", "Asao Fujiyama", "Masahira Hattori", "Tetsushi Yada", "Atsushi Toyoda", "Takehiko Itoh", "Chiharu Kawagoe", "Hidemi Watanabe", "Yasushi Totoki", "Todd Taylor", "Genoscope and CNRS UMR-8030:", "Jean Weissenbach", "Roland Heilig", "William Saurin", "Francois Artiguenave", "Philippe Brottier", "Thomas Bruls", "Eric Pelletier", "Catherine Robert", "Patrick Wincker", "Department of Genome Analysis, Institute of Molecular Biotechnology:", "André Rosenthal", "Matthias Platzer", "Gerald Nyakatura", "Stefan Taudien", "Andreas Rump", "GTC Sequencing Center:", "Douglas R. Smith", "Lynn Doucette-Stamm", "Marc Rubenfield", "Keith Weinstock", "Hong Mei Lee", "JoAnn Dubois", "Beijing Genomics Institute/Human Genome Center:", "Huanming Yang", "Jun Yu", "Jian Wang", "Guyang Huang", "Jun Gu", "Multimegabase Sequencing Center, The Institute for Systems Biology:", "Leroy Hood", "Lee Rowen", "Anup Madan", "Shizen Qin", "Stanford Genome Technology Center:", "Ronald W. Davis", "Nancy A. Federspiel", "A. Pia Abola", "Michael J. Proctor", "University of Oklahoma's Advanced Center for Genome Technology:", "Bruce A. Roe", "Feng Chen", "Huaqin Pan", "Max Planck Institute for Molecular Genetics:", "Juliane Ramser", "Hans Lehrach", "Richard Reinhardt", "Cold Spring Harbor Laboratory, Lita Annenberg Hazen Genome Center:", "W. Richard McCombie", "Melissa de la Bastide", "Neilay Dedhia", "GBF—German Research Centre for Biotechnology:", "Helmut Blöcker", "Klaus Hornischer", "Gabriele Nordsiek", "*Genome Analysis Group (listed in alphabetical order, also includes individuals listed under other headings):", "Richa Agarwala", "L. Aravind", "Jeffrey A. Bailey", "Alex Bateman", "Serafim Batzoglou", "Ewan Birney", "Peer Bork", "Daniel G. Brown", "Christopher B. Burge", "Lorenzo Cerutti", "Hsiu-Chuan Chen", "Deanna Church", "Michele Clamp", "Richard R. Copley", "Tobias Doerks", "Sean R. Eddy", "Evan E. Eichler", "Terrence S. Furey", "James Galagan", "James G. R. Gilbert", "Cyrus Harmon", "Yoshihide Hayashizaki", "David Haussler", "Henning Hermjakob", "Karsten Hokamp", "Wonhee Jang", "L. Steven Johnson", "Thomas A. Jones", "Simon Kasif", "Arek Kaspryzk", "Scot Kennedy", "W. James Kent", "Paul Kitts", "Eugene V. Koonin", "Ian Korf", "David Kulp", "Doron Lancet", "Todd M. Lowe", "Aoife McLysaght", "Tarjei Mikkelsen", "John V. Moran", "Nicola Mulder", "Victor J. Pollara", "Chris P. Ponting", "Greg Schuler", "Jörg Schultz", "Guy Slater", "Arian F. A. Smit", "Elia Stupka", "Joseph Szustakowki", "Danielle Thierry-Mieg", "Jean Thierry-Mieg", "Lukas Wagner", "John Wallis", "Raymond Wheeler", "Alan Williams", "Yuri I. Wolf", "Kenneth H. Wolfe", "Shiaw-Pyng Yang", "Ru-Fang Yeh", "Scientific management: National Human Genome Research Institute, US National Institutes of Health:", "Francis Collins", "Mark S. Guyer", "Jane Peterson", "Adam Felsenfeld", "Kris A. Wetterstrand", "Stanford Human Genome Center:", "Richard M. Myers", "Jeremy Schmutz", "Mark Dickson", "Jane Grimwood", "David R. Cox", "University of Washington Genome Center:", "Maynard V. Olson", "Rajinder Kaul", "Christopher Raymond", "Department of Molecular Biology, Keio University School of Medicine:", "Nobuyoshi Shimizu", "Kazuhiko Kawasaki", "Shinsei Minoshima", "University of Texas Southwestern Medical Center at Dallas:", "Glen A. Evans", "Maria Athanasiou", "Roger Schultz", "Office of Science, US Department of Energy:", "Aristides Patrinos", "The Wellcome Trust:", "Michael J. Morgan"], "year": 2001, "venue": "Nature", "url": "https://doi.org/10.1038/35057062", "doi": "10.1038/35057062", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-07", "2020-10", "2026-04"], "last_mentioned": "2026-04-30T08:20:37Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/454034a", "slug": "91d55407022694a5", "title": "Arise 'cliodynamics'", "authors": ["Peter Turchin"], "year": 2008, "venue": "Nature", "url": "https://doi.org/10.1038/454034a", "doi": "10.1038/454034a", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-05", "2021-03", "2021-07"], "last_mentioned": "2021-07-06T15:09:47Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature.2012.11811", "slug": "d84d237a793a9407", "title": "How brainless slime molds redefine intelligence", "authors": ["Ferris Jabr"], "year": 2012, "venue": "Nature", "url": "https://doi.org/10.1038/nature.2012.11811", "doi": "10.1038/nature.2012.11811", "topics": ["Psychology", "Neuroscience"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-11", "2025-07", "2026-03"], "last_mentioned": "2026-03-24T23:17:32Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature15393", "slug": "34ac586fd4185c2c", "title": "A global reference for human genetic variation", "authors": ["The 1000 Genomes Project Consortium", "Corresponding authors", "Adam Auton", "Gonçalo R. Abecasis", "Steering committee", "David M. Altshuler", "Richard M. Durbin", "Gonçalo R. Abecasis", "David R. Bentley", "Aravinda Chakravarti", "Andrew G. Clark", "Peter Donnelly", "Evan E. Eichler", "Paul Flicek", "Stacey B. Gabriel", "Richard A. Gibbs", "Eric D. Green", "Matthew E. Hurles", "Bartha M. Knoppers", "Jan O. Korbel", "Eric S. Lander", "Charles Lee", "Hans Lehrach", "Elaine R. Mardis", "Gabor T. Marth", "Gil A. McVean", "Deborah A. Nickerson", "Jeanette P. Schmidt", "Stephen T. Sherry", "Jun Wang", "Richard K. Wilson", "Production group", "Baylor College of Medicine", "Richard A. Gibbs", "Eric Boerwinkle", "Harsha Doddapaneni", "Yi Han", "Viktoriya Korchina", "Christie Kovar", "Sandra Lee", "Donna Muzny", "Jeffrey G. Reid", "Yiming Zhu", "BGI-Shenzhen", "Jun Wang", "Yuqi Chang", "Qiang Feng", "Xiaodong Fang", "Xiaosen Guo", "Min Jian", "Hui Jiang", "Xin Jin", "Tianming Lan", "Guoqing Li", "Jingxiang Li", "Yingrui Li", "Shengmao Liu", "Xiao Liu", "Yao Lu", "Xuedi Ma", "Meifang Tang", "Bo Wang", "Guangbiao Wang", "Honglong Wu", "Renhua Wu", "Xun Xu", "Ye Yin", "Dandan Zhang", "Wenwei Zhang", "Jiao Zhao", "Meiru Zhao", "Xiaole Zheng", "Broad Institute of MIT and Harvard", "Eric S. Lander", "David M. Altshuler", "Stacey B. Gabriel", "Namrata Gupta", "Coriell Institute for Medical Research", "Neda Gharani", "Lorraine H. Toji", "Norman P. Gerry", "Alissa M. Resch", "European Molecular Biology Laboratory, European Bioinformatics Institute", "Paul Flicek", "Jonathan Barker", "Laura Clarke", "Laurent Gil", "Sarah E. Hunt", "Gavin Kelman", "Eugene Kulesha", "Rasko Leinonen", "William M. McLaren", "Rajesh Radhakrishnan", "Asier Roa", "Dmitriy Smirnov", "Richard E. Smith", "Ian Streeter", "Anja Thormann", "Iliana Toneva", "Brendan Vaughan", "Xiangqun Zheng-Bradley", "Illumina", "David R. Bentley", "Russell Grocock", "Sean Humphray", "Terena James", "Zoya Kingsbury", "Max Planck Institute for Molecular Genetics", "Hans Lehrach", "Ralf Sudbrak", "Marcus W. Albrecht", "Vyacheslav S. Amstislavskiy", "Tatiana A. Borodina", "Matthias Lienhard", "Florian Mertes", "Marc Sultan", "Bernd Timmermann", "Marie-Laure Yaspo", "McDonnell Genome Institute at Washington University", "Elaine R. Mardis", "Richard K. Wilson", "Lucinda Fulton", "Robert Fulton", "US National Institutes of Health", "Stephen T. Sherry", "Victor Ananiev", "Zinaida Belaia", "Dimitriy Beloslyudtsev", "Nathan Bouk", "Chao Chen", "Deanna Church", "Robert Cohen", "Charles Cook", "John Garner", "Timothy Hefferon", "Mikhail Kimelman", "Chunlei Liu", "John Lopez", "Peter Meric", "Chris O’Sullivan", "Yuri Ostapchuk", "Lon Phan", "Sergiy Ponomarov", "Valerie Schneider", "Eugene Shekhtman", "Karl Sirotkin", "Douglas Slotta", "Hua Zhang", "University of Oxford", "Gil A. McVean", "Wellcome Trust Sanger Institute", "Richard M. Durbin", "Senduran Balasubramaniam", "John Burton", "Petr Danecek", "Thomas M. Keane", "Anja Kolb-Kokocinski", "Shane McCarthy", "James Stalker", "Michael Quail", "Analysis group", "Affymetrix", "Jeanette P. Schmidt", "Christopher J. Davies", "Jeremy Gollub", "Teresa Webster", "Brant Wong", "Yiping Zhan", "Albert Einstein College of Medicine", "Adam Auton", "Christopher L. Campbell", "Yu Kong", "Anthony Marcketta", "Baylor College of Medicine", "Richard A. Gibbs", "Fuli Yu", "Lilian Antunes", "Matthew Bainbridge", "Donna Muzny", "Aniko Sabo", "Zhuoyi Huang", "BGI-Shenzhen", "Jun Wang", "Lachlan J. M. Coin", "Lin Fang", "Xiaosen Guo", "Xin Jin", "Guoqing Li", "Qibin Li", "Yingrui Li", "Zhenyu Li", "Haoxiang Lin", "Binghang Liu", "Ruibang Luo", "Haojing Shao", "Yinlong Xie", "Chen Ye", "Chang Yu", "Fan Zhang", "Hancheng Zheng", "Hongmei Zhu", "Bilkent University", "Can Alkan", "Elif Dal", "Fatma Kahveci", "Boston College", "Gabor T. Marth", "Erik P. Garrison", "Deniz Kural", "Wan-Ping Lee", "Wen Fung Leong", "Michael Stromberg", "Alistair N. Ward", "Jiantao Wu", "Mengyao Zhang", "Broad Institute of MIT and Harvard", "Mark J. Daly", "Mark A. DePristo", "Robert E. Handsaker", "David M. Altshuler", "Eric Banks", "Gaurav Bhatia", "Guillermo del Angel", "Stacey B. Gabriel", "Giulio Genovese", "Namrata Gupta", "Heng Li", "Seva Kashin", "Eric S. Lander", "Steven A. McCarroll", "James C. Nemesh", "Ryan E. Poplin", "Cold Spring Harbor Laboratory", "Seungtai C. Yoon", "Jayon Lihm", "Vladimir Makarov", "Cornell University", "Andrew G. Clark", "Srikanth Gottipati", "Alon Keinan", "Juan L. Rodriguez-Flores", "European Molecular Biology Laboratory", "Jan O. Korbel", "Tobias Rausch", "Markus H. Fritz", "Adrian M. Stütz", "European Molecular Biology Laboratory, European Bioinformatics Institute", "Paul Flicek", "Kathryn Beal", "Laura Clarke", "Avik Datta", "Javier Herrero", "William M. McLaren", "Graham R. S. Ritchie", "Richard E. Smith", "Daniel Zerbino", "Xiangqun Zheng-Bradley", "Harvard University", "Pardis C. Sabeti", "Ilya Shlyakhter", "Stephen F. Schaffner", "Joseph Vitti", "Human Gene Mutation Database", "David N. Cooper", "Edward V. Ball", "Peter D. Stenson", "Illumina", "David R. Bentley", "Bret Barnes", "Markus Bauer", "R. Keira Cheetham", "Anthony Cox", "Michael Eberle", "Sean Humphray", "Scott Kahn", "Lisa Murray", "John Peden", "Richard Shaw", "Icahn School of Medicine at Mount Sinai", "Eimear E. Kenny", "Louisiana State University", "Mark A. Batzer", "Miriam K. Konkel", "Jerilyn A. Walker", "Massachusetts General Hospital", "Daniel G. MacArthur", "Monkol Lek", "Max Planck Institute for Molecular Genetics", "Ralf Sudbrak", "Vyacheslav S. Amstislavskiy", "Ralf Herwig", "McDonnell Genome Institute at Washington University", "Elaine R. Mardis", "Li Ding", "Daniel C. Koboldt", "David Larson", "Kai Ye", "McGill University", "Simon Gravel", "National Eye Institute, NIH", "Anand Swaroop", "Emily Chew", "New York Genome Center", "Tuuli Lappalainen", "Yaniv Erlich", "Melissa Gymrek", "Thomas Frederick Willems", "Ontario Institute for Cancer Research", "Jared T. Simpson", "Pennsylvania State University", "Mark D. Shriver", "Rutgers Cancer Institute of New Jersey", "Jeffrey A. Rosenfeld", "Stanford University", "Carlos D. Bustamante", "Stephen B. Montgomery", "Francisco M. De La Vega", "Jake K. Byrnes", "Andrew W. Carroll", "Marianne K. DeGorter", "Phil Lacroute", "Brian K. Maples", "Alicia R. Martin", "Andres Moreno-Estrada", "Suyash S. Shringarpure", "Fouad Zakharia", "Tel-Aviv University", "Eran Halperin", "Yael Baran", "The Jackson Laboratory for Genomic Medicine", "Charles Lee", "Eliza Cerveira", "Jaeho Hwang", "Ankit Malhotra", "Dariusz Plewczynski", "Kamen Radew", "Mallory Romanovitch", "Chengsheng Zhang", "Thermo Fisher Scientific", "Fiona C. L. Hyland", "Translational Genomics Research Institute", "David W. Craig", "Alexis Christoforides", "Nils Homer", "Tyler Izatt", "Ahmet A. Kurdoglu", "Shripad A. Sinari", "Kevin Squire", "US National Institutes of Health", "Stephen T. Sherry", "Chunlin Xiao", "University of California, San Diego", "Jonathan Sebat", "Danny Antaki", "Madhusudan Gujral", "Amina Noor", "Kenny Ye", "University of California, San Francisco", "Esteban G. Burchard", "Ryan D. Hernandez", "Christopher R. Gignoux", "University of California, Santa Cruz", "David Haussler", "Sol J. Katzman", "W. James Kent", "University of Chicago", "Bryan Howie", "University College London", "Andres Ruiz-Linares", "University of Geneva", "Emmanouil T. Dermitzakis", "University of Maryland School of Medicine", "Scott E. Devine", "University of Michigan", "Gonçalo R. Abecasis", "Hyun Min Kang", "Jeffrey M. Kidd", "Tom Blackwell", "Sean Caron", "Wei Chen", "Sarah Emery", "Lars Fritsche", "Christian Fuchsberger", "Goo Jun", "Bingshan Li", "Robert Lyons", "Chris Scheller", "Carlo Sidore", "Shiya Song", "Elzbieta Sliwerska", "Daniel Taliun", "Adrian Tan", "Ryan Welch", "Mary Kate Wing", "Xiaowei Zhan", "University of Montréal", "Philip Awadalla", "Alan Hodgkinson", "University of North Carolina at Chapel Hill", "Yun Li", "University of North Carolina at Charlotte", "Xinghua Shi", "Andrew Quitadamo", "University of Oxford", "Gerton Lunter", "Gil A. McVean", "Jonathan L. Marchini", "Simon Myers", "Claire Churchhouse", "Olivier Delaneau", "Anjali Gupta-Hinch", "Warren Kretzschmar", "Zamin Iqbal", "Iain Mathieson", "Androniki Menelaou", "Andy Rimmer", "Dionysia K. Xifara", "University of Puerto Rico", "Taras K. Oleksyk", "University of Texas Health Sciences Center at Houston", "Yunxin Fu", "Xiaoming Liu", "Momiao Xiong", "University of Utah", "Lynn Jorde", "David Witherspoon", "Jinchuan Xing", "University of Washington", "Evan E. Eichler", "Brian L. Browning", "Sharon R. Browning", "Fereydoun Hormozdiari", "Peter H. Sudmant", "Weill Cornell Medical College", "Ekta Khurana", "Wellcome Trust Sanger Institute", "Richard M. Durbin", "Matthew E. Hurles", "Chris Tyler-Smith", "Cornelis A. Albers", "Qasim Ayub", "Senduran Balasubramaniam", "Yuan Chen", "Vincenza Colonna", "Petr Danecek", "Luke Jostins", "Thomas M. Keane", "Shane McCarthy", "Klaudia Walter", "Yali Xue", "Yale University", "Mark B. Gerstein", "Alexej Abyzov", "Suganthi Balasubramanian", "Jieming Chen", "Declan Clarke", "Yao Fu", "Arif O. Harmanci", "Mike Jin", "Donghoon Lee", "Jeremy Liu", "Xinmeng Jasmine Mu", "Jing Zhang", "Yan Zhang", "Structural variation group", "BGI-Shenzhen", "Yingrui Li", "Ruibang Luo", "Hongmei Zhu", "Bilkent University", "Can Alkan", "Elif Dal", "Fatma Kahveci", "Boston College", "Gabor T. Marth", "Erik P. Garrison", "Deniz Kural", "Wan-Ping Lee", "Alistair N. Ward", "Jiantao Wu", "Mengyao Zhang", "Broad Institute of MIT and Harvard", "Steven A. McCarroll", "Robert E. Handsaker", "David M. Altshuler", "Eric Banks", "Guillermo del Angel", "Giulio Genovese", "Chris Hartl", "Heng Li", "Seva Kashin", "James C. Nemesh", "Khalid Shakir", "Cold Spring Harbor Laboratory", "Seungtai C. Yoon", "Jayon Lihm", "Vladimir Makarov", "Cornell University", "Jeremiah Degenhardt", "European Molecular Biology Laboratory", "Jan O. Korbel", "Markus H. Fritz", "Sascha Meiers", "Benjamin Raeder", "Tobias Rausch", "Adrian M. Stütz", "European Molecular Biology Laboratory, European Bioinformatics Institute", "Paul Flicek", "Francesco Paolo Casale", "Laura Clarke", "Richard E. Smith", "Oliver Stegle", "Xiangqun Zheng-Bradley", "Illumina", "David R. Bentley", "Bret Barnes", "R. Keira Cheetham", "Michael Eberle", "Sean Humphray", "Scott Kahn", "Lisa Murray", "Richard Shaw", "Leiden University Medical Center", "Eric-Wubbo Lameijer", "Louisiana State University", "Mark A. Batzer", "Miriam K. Konkel", "Jerilyn A. Walker", "McDonnell Genome Institute at Washington University", "Li Ding", "Ira Hall", "Kai Ye", "Stanford University", "Phil Lacroute", "The Jackson Laboratory for Genomic Medicine", "Charles Lee", "Eliza Cerveira", "Ankit Malhotra", "Jaeho Hwang", "Dariusz Plewczynski", "Kamen Radew", "Mallory Romanovitch", "Chengsheng Zhang", "Translational Genomics Research Institute", "David W. Craig", "Nils Homer", "US National Institutes of Health", "Deanna Church", "Chunlin Xiao", "University of California, San Diego", "Jonathan Sebat", "Danny Antaki", "Vineet Bafna", "Jacob Michaelson", "Kenny Ye", "University of Maryland School of Medicine", "Scott E. Devine", "Eugene J. Gardner", "University of Michigan", "Gonçalo R. Abecasis", "Jeffrey M. Kidd", "Ryan E. Mills", "Gargi Dayama", "Sarah Emery", "Goo Jun", "University of North Carolina at Charlotte", "Xinghua Shi", "Andrew Quitadamo", "University of Oxford", "Gerton Lunter", "Gil A. McVean", "University of Texas MD Anderson Cancer Center", "Ken Chen", "Xian Fan", "Zechen Chong", "Tenghui Chen", "University of Utah", "David Witherspoon", "Jinchuan Xing", "University of Washington", "Evan E. Eichler", "Mark J. Chaisson", "Fereydoun Hormozdiari", "John Huddleston", "Maika Malig", "Bradley J. Nelson", "Peter H. Sudmant", "Vanderbilt University School of Medicine", "Nicholas F. Parrish", "Weill Cornell Medical College", "Ekta Khurana", "Wellcome Trust Sanger Institute", "Matthew E. Hurles", "Ben Blackburne", "Sarah J. Lindsay", "Zemin Ning", "Klaudia Walter", "Yujun Zhang", "Yale University", "Mark B. Gerstein", "Alexej Abyzov", "Jieming Chen", "Declan Clarke", "Hugo Lam", "Xinmeng Jasmine Mu", "Cristina Sisu", "Jing Zhang", "Yan Zhang", "Exome group", "Baylor College of Medicine", "Richard A. Gibbs", "Fuli Yu", "Matthew Bainbridge", "Danny Challis", "Uday S. Evani", "Christie Kovar", "James Lu", "Donna Muzny", "Uma Nagaswamy", "Jeffrey G. Reid", "Aniko Sabo", "Jin Yu", "BGI-Shenzhen", "Xiaosen Guo", "Wangshen Li", "Yingrui Li", "Renhua Wu", "Boston College", "Gabor T. Marth", "Erik P. Garrison", "Wen Fung Leong", "Alistair N. Ward", "Broad Institute of MIT and Harvard", "Guillermo del Angel", "Mark A. DePristo", "Stacey B. Gabriel", "Namrata Gupta", "Chris Hartl", "Ryan E. Poplin", "Cornell University", "Andrew G. Clark", "Juan L. Rodriguez-Flores", "European Molecular Biology Laboratory, European Bioinformatics Institute", "Paul Flicek", "Laura Clarke", "Richard E. Smith", "Xiangqun Zheng-Bradley", "Massachusetts General Hospital", "Daniel G. MacArthur", "McDonnell Genome Institute at Washington University", "Elaine R. Mardis", "Robert Fulton", "Daniel C. Koboldt", "McGill University", "Simon Gravel", "Stanford University", "Carlos D. Bustamante", "Translational Genomics Research Institute", "David W. Craig", "Alexis Christoforides", "Nils Homer", "Tyler Izatt", "US National Institutes of Health", "Stephen T. Sherry", "Chunlin Xiao", "University of Geneva", "Emmanouil T. Dermitzakis", "University of Michigan", "Gonçalo R. Abecasis", "Hyun Min Kang", "University of Oxford", "Gil A. McVean", "Yale University", "Mark B. Gerstein", "Suganthi Balasubramanian", "Lukas Habegger", "Functional interpretation group", "Cornell University", "Haiyuan Yu", "European Molecular Biology Laboratory, European Bioinformatics Institute", "Paul Flicek", "Laura Clarke", "Fiona Cunningham", "Ian Dunham", "Daniel Zerbino", "Xiangqun Zheng-Bradley", "Harvard University", "Kasper Lage", "Jakob Berg Jespersen", "Heiko Horn", "Stanford University", "Stephen B. Montgomery", "Marianne K. DeGorter", "Weill Cornell Medical College", "Ekta Khurana", "Wellcome Trust Sanger Institute", "Chris Tyler-Smith", "Yuan Chen", "Vincenza Colonna", "Yali Xue", "Yale University", "Mark B. Gerstein", "Suganthi Balasubramanian", "Yao Fu", "Donghoon Kim", "Chromosome Y group", "Albert Einstein College of Medicine", "Adam Auton", "Anthony Marcketta", "American Museum of Natural History", "Rob Desalle", "Apurva Narechania", "Arizona State University", "Melissa A. Wilson Sayres", "Boston College", "Erik P. Garrison", "Broad Institute of MIT and Harvard", "Robert E. Handsaker", "Seva Kashin", "Steven A. McCarroll", "Cornell University", "Juan L. Rodriguez-Flores", "European Molecular Biology Laboratory, European Bioinformatics Institute", "Paul Flicek", "Laura Clarke", "Xiangqun Zheng-Bradley", "New York Genome Center", "Yaniv Erlich", "Melissa Gymrek", "Thomas Frederick Willems", "Stanford University", "Carlos D. Bustamante", "Fernando L. Mendez", "G. David Poznik", "Peter A. Underhill", "The Jackson Laboratory for Genomic Medicine", "Charles Lee", "Eliza Cerveira", "Ankit Malhotra", "Mallory Romanovitch", "Chengsheng Zhang", "University of Michigan", "Gonçalo R. Abecasis", "University of Queensland", "Lachlan Coin", "Haojing Shao", "Virginia Bioinformatics Institute", "David Mittelman", "Wellcome Trust Sanger Institute", "Chris Tyler-Smith", "Qasim Ayub", "Ruby Banerjee", "Maria Cerezo", "Yuan Chen", "Thomas W. Fitzgerald", "Sandra Louzada", "Andrea Massaia", "Shane McCarthy", "Graham R. Ritchie", "Yali Xue", "Fengtang Yang", "Data coordination center group", "Baylor College of Medicine", "Richard A. Gibbs", "Christie Kovar", "Divya Kalra", "Walker Hale", "Donna Muzny", "Jeffrey G. Reid", "BGI-Shenzhen", "Jun Wang", "Xu Dan", "Xiaosen Guo", "Guoqing Li", "Yingrui Li", "Chen Ye", "Xiaole Zheng", "Broad Institute of MIT and Harvard", "David M. Altshuler", "European Molecular Biology Laboratory, European Bioinformatics Institute", "Paul Flicek", "Laura Clarke", "Xiangqun Zheng-Bradley", "Illumina", "David R. Bentley", "Anthony Cox", "Sean Humphray", "Scott Kahn", "Max Planck Institute for Molecular Genetics", "Ralf Sudbrak", "Marcus W. Albrecht", "Matthias Lienhard", "McDonnell Genome Institute at Washington University", "David Larson", "Translational Genomics Research Institute", "David W. Craig", "Tyler Izatt", "Ahmet A. Kurdoglu", "US National Institutes of Health", "Stephen T. Sherry", "Chunlin Xiao", "University of California, Santa Cruz", "David Haussler", "University of Michigan", "Gonçalo R. Abecasis", "University of Oxford", "Gil A. McVean", "Wellcome Trust Sanger Institute", "Richard M. Durbin", "Senduran Balasubramaniam", "Thomas M. Keane", "Shane McCarthy", "James Stalker", "Samples and ELSI group", "Aravinda Chakravarti", "Bartha M. Knoppers", "Gonçalo R. Abecasis", "Kathleen C. Barnes", "Christine Beiswanger", "Esteban G. Burchard", "Carlos D. Bustamante", "Hongyu Cai", "Hongzhi Cao", "Richard M. Durbin", "Norman P. Gerry", "Neda Gharani", "Richard A. Gibbs", "Christopher R. Gignoux", "Simon Gravel", "Brenna Henn", "Danielle Jones", "Lynn Jorde", "Jane S. Kaye", "Alon Keinan", "Alastair Kent", "Angeliki Kerasidou", "Yingrui Li", "Rasika Mathias", "Gil A. McVean", "Andres Moreno-Estrada", "Pilar N. Ossorio", "Michael Parker", "Alissa M. Resch", "Charles N. Rotimi", "Charmaine D. Royal", "Karla Sandoval", "Yeyang Su", "Ralf Sudbrak", "Zhongming Tian", "Sarah Tishkoff", "Lorraine H. Toji", "Chris Tyler-Smith", "Marc Via", "Yuhong Wang", "Huanming Yang", "Ling Yang", "Jiayong Zhu", "Sample collection", "British from England and Scotland (GBR)", "Walter Bodmer", "Colombians in Medellín, Colombia (CLM)", "Gabriel Bedoya", "Andres Ruiz-Linares", "Han Chinese South (CHS)", "Zhiming Cai", "Yang Gao", "Jiayou Chu", "Finnish in Finland (FIN)", "Leena Peltonen", "Iberian Populations in Spain (IBS)", "Andres Garcia-Montero", "Alberto Orfao", "Puerto Ricans in Puerto Rico (PUR)", "Julie Dutil", "Juan C. Martinez-Cruzado", "Taras K. Oleksyk", "African Caribbean in Barbados (ACB)", "Kathleen C. Barnes", "Rasika A. Mathias", "Anselm Hennis", "Harold Watson", "Colin McKenzie", "Bengali in Bangladesh (BEB)", "Firdausi Qadri", "Regina LaRocque", "Pardis C. Sabeti", "Chinese Dai in Xishuangbanna, China (CDX)", "Jiayong Zhu", "Xiaoyan Deng", "Esan in Nigeria (ESN)", "Pardis C. Sabeti", "Danny Asogun", "Onikepe Folarin", "Christian Happi", "Omonwunmi Omoniwa", "Matt Stremlau", "Ridhi Tariyal", "Gambian in Western Division – Mandinka (GWD)", "Muminatou Jallow", "Fatoumatta Sisay Joof", "Tumani Corrah", "Kirk Rockett", "Dominic Kwiatkowski", "Indian Telugu in the UK (ITU) and Sri Lankan Tamil in the UK (STU)", "Jaspal Kooner", "Kinh in Ho Chi Minh City, Vietnam (KHV)", "Trâ`n Tịnh Hiê`n", "Sarah J. Dunstan", "Nguyen Thuy Hang", "Mende in Sierra Leone (MSL)", "Richard Fonnie", "Robert Garry", "Lansana Kanneh", "Lina Moses", "Pardis C. Sabeti", "John Schieffelin", "Donald S. Grant", "Peruvian in Lima, Peru (PEL)", "Carla Gallo", "Giovanni Poletti", "Punjabi in Lahore, Pakistan (PJL)", "Danish Saleheen", "Asif Rasheed", "Scientific management", "Lisa D. Brooks", "Adam L. Felsenfeld", "Jean E. McEwen", "Yekaterina Vaydylevich", "Eric D. Green", "Audrey Duncanson", "Michael Dunn", "Jeffery A. Schloss", "Jun Wang", "Huanming Yang", "Writing group", "Adam Auton", "Lisa D. Brooks", "Richard M. Durbin", "Erik P. Garrison", "Hyun Min Kang", "Jan O. Korbel", "Jonathan L. Marchini", "Shane McCarthy", "Gil A. McVean", "Gonçalo R. Abecasis"], "year": 2015, "venue": "Nature", "url": "https://doi.org/10.1038/nature15393", "doi": "10.1038/nature15393", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-03", "2023-12", "2025-06"], "last_mentioned": "2025-06-07T05:35:52Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature21056", "slug": "439f95d483abedb7", "title": "Dermatologist-level classification of skin cancer with deep neural networks", "authors": ["Andre Esteva", "Brett Kuprel", "Roberto A. Novoa", "Justin Ko", "Susan M. Swetter", "Helen M. Blau", "Sebastian Thrun"], "year": 2017, "venue": "Nature", "url": "https://doi.org/10.1038/nature21056", "doi": "10.1038/nature21056", "topics": ["Machine learning", "Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-06", "2019-03", "2025-09"], "last_mentioned": "2025-09-07T15:51:44Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature22065", "slug": "9deeff4de2b99515", "title": "A 130,000-year-old archaeological site in southern California, USA", "authors": ["Steven R. Holen", "Thomas A. Deméré", "Daniel C. Fisher", "Richard Fullagar", "James B. Paces", "George T. Jefferson", "Jared M. Beeton", "Richard A. Cerutti", "Adam N. Rountrey", "Lawrence Vescera", "Kathleen A. Holen"], "year": 2017, "venue": "Nature", "url": "https://doi.org/10.1038/nature22065", "doi": "10.1038/nature22065", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-04", "2020-07", "2023-01"], "last_mentioned": "2023-01-26T01:10:59Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nature25975", "slug": "b9c1aca405415a51", "title": "Human hippocampal neurogenesis drops sharply in children to undetectable levels in adults", "authors": ["Shawn F. Sorrells", "Mercedes F. Paredes", "Arantxa Cebrian-Silla", "Kadellyn Sandoval", "Dashi Qi", "Kevin W. Kelley", "David James", "Simone Mayer", "Julia Chang", "Kurtis I. Auguste", "Edward F. Chang", "Antonio J. Gutierrez", "Arnold R. Kriegstein", "Gary W. Mathern", "Michael C. Oldham", "Eric J. Huang", "Jose Manuel Garcia-Verdugo", "Zhengang Yang", "Arturo Alvarez-Buylla"], "year": 2018, "venue": "Nature", "url": "https://doi.org/10.1038/nature25975", "doi": "10.1038/nature25975", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-04", "2018-11", "2019-03"], "last_mentioned": "2019-03-25T19:24:22Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nclimate2120", "slug": "5b4f6162b618a5ff", "title": "Taming hurricanes with arrays of offshore wind turbines", "authors": ["Mark Z. Jacobson", "Cristina L. Archer", "Willett Kempton"], "year": 2014, "venue": "Nature Climate Change", "url": "https://doi.org/10.1038/nclimate2120", "doi": "10.1038/nclimate2120", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-09", "2024-03", "2025-06"], "last_mentioned": "2025-06-04T02:56:14Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nclimate2921", "slug": "29eeb542088af9ea", "title": "Future cost-competitive electricity systems and their impact on US CO2 emissions", "authors": ["Alexander E. MacDonald", "Christopher T. M. Clack", "Anneliese Alexander", "Adam Dunbar", "James Wilczak", "Yuanfu Xie"], "year": 2016, "venue": "Nature Climate Change", "url": "https://doi.org/10.1038/nclimate2921", "doi": "10.1038/nclimate2921", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-06", "2023-03", "2025-04"], "last_mentioned": "2025-04-21T19:42:18Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/ncomms13276", "slug": "04b96342878d8fe3", "title": "Random synaptic feedback weights support error backpropagation for deep learning", "authors": ["Timothy P. Lillicrap", "Daniel Cownden", "Douglas B. Tweed", "Colin J. Akerman"], "year": 2016, "venue": "Nature Communications", "url": "https://doi.org/10.1038/ncomms13276", "doi": "10.1038/ncomms13276", "topics": ["Neuroscience", "Machine learning", "Algorithms"], "description": "Abstract The brain processes information through multiple layers of neurons.", "description_method": "abstract-excerpt", "abstract": "Abstract The brain processes information through multiple layers of neurons. This deep architecture is representationally powerful, but complicates learning because it is difficult to identify the responsible neurons when a mistake is made. In machine learning, the backpropagation algorithm assigns blame by multiplying error signals with all the synaptic weights on each neuron’s axon and further downstream. However, this involves a precise, symmetric backward connectivity pattern, which is thought to be impossible in the brain. Here we demonstrate that this strong architectural constraint is not required for effective error propagation. We present a surprisingly simple mechanism that assigns blame by multiplying errors by even random synaptic weights. This mechanism can transmit teaching signals across multiple layers of neurons and performs as effectively as backpropagation on a variety of tasks. Our results help reopen questions about how the brain could use error signals and dispel long-held assumptions about algorithmic constraints on learning.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-08", "2024-11", "2026-04"], "last_mentioned": "2026-04-13T01:47:38Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/news.2010.628", "slug": "2c260e227507562c", "title": "Lab animals and pets face obesity epidemic", "authors": ["Alla Katsnelson"], "year": 2010, "venue": "Nature", "url": "https://doi.org/10.1038/news.2010.628", "doi": "10.1038/news.2010.628", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-06", "2021-07", "2023-01"], "last_mentioned": "2023-01-11T12:22:58Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nn.3594", "slug": "9157c26a5ea19f60", "title": "Parental olfactory experience influences behavior and neural structure in subsequent generations", "authors": ["Brian G Dias", "Kerry J Ressler"], "year": 2013, "venue": "Nature Neuroscience", "url": "https://doi.org/10.1038/nn.3594", "doi": "10.1038/nn.3594", "topics": ["Psychology"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-12", "2025-11", "2025-12"], "last_mentioned": "2025-12-28T03:55:40Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/nphoton.2013.45", "slug": "950c98474a25087c", "title": "Towards high-capacity fibre-optic communications at the speed of light in vacuum", "authors": ["F. Poletti", "N. V. Wheeler", "M. N. Petrovich", "N. Baddela", "E. Numkam Fokoua", "J. R. Hayes", "D. R. Gray", "Z. Li", "R. Slavík", "D. J. Richardson"], "year": 2013, "venue": "Nature Photonics", "url": "https://doi.org/10.1038/nphoton.2013.45", "doi": "10.1038/nphoton.2013.45", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2015-12", "2017-10", "2020-02"], "last_mentioned": "2020-02-11T12:26:44Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41366-021-00805-6", "slug": "118f9928ccd47088", "title": "Adenovirus 36 prevalence and association with human obesity: a systematic review", "authors": ["Jaime da Silva Fernandes", "Fabiana Schuelter-Trevisol", "Ana Carolina Lobor Cancelier", "Helena Caetano Gonçalves e Silva", "Daiana Gomes de Sousa", "Richard L. Atkinson", "Daisson José Trevisol"], "year": 2021, "venue": "International Journal of Obesity", "url": "https://doi.org/10.1038/s41366-021-00805-6", "doi": "10.1038/s41366-021-00805-6", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-02", "2024-03", "2025-03"], "last_mentioned": "2025-03-23T15:38:21Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41413-020-0084-5", "slug": "f3817a174d349acb", "title": "Long-term bone and lung consequences associated with hospital-acquired severe acute respiratory syndrome: a 15-year follow-up from a prospective cohort study", "authors": ["Peixun Zhang", "Jia Li", "Huixin Liu", "Na Han", "Jiabao Ju", "Yuhui Kou", "Lei Chen", "Mengxi Jiang", "Feng Pan", "Yali Zheng", "Zhancheng Gao", "Baoguo Jiang"], "year": 2020, "venue": "Bone Research", "url": "https://doi.org/10.1038/s41413-020-0084-5", "doi": "10.1038/s41413-020-0084-5", "topics": ["Biology & medicine"], "description": "AbstractThe most severe sequelae after rehabilitation from SARS are femoral head necrosis and pulmonary fibrosis.", "description_method": "abstract-excerpt", "abstract": "AbstractThe most severe sequelae after rehabilitation from SARS are femoral head necrosis and pulmonary fibrosis. We performed a 15-year follow-up on the lung and bone conditions of SARS patients. We evaluated the recovery from lung damage and femoral head necrosis in an observational cohort study of SARS patients using pulmonary CT scans, hip joint MRI examinations, pulmonary function tests and hip joint function questionnaires. Eighty medical staff contracted SARS in 2003. Two patients died of SARS, and 78 were enrolled in this study from August 2003 to March 2018. Seventy-one patients completed the 15-year follow-up. The percentage of pulmonary lesions on CT scans diminished from 2003 (9.40 ± 7.83)% to 2004 (3.20 ± 4.78)% (P < 0.001) and remained stable thereafter until 2018 (4.60 ± 6.37)%. Between 2006 and 2018, the proportion of patients with interstitial changes who had improved pulmonary function was lower than that of patients without lesions, as demonstrated by the one-second ratio (FEV1/FVC%, t = 2.21, P = 0.04) and mid-flow of maximum expiration (FEF25%–75%, t = 2.76, P = 0.01). The volume of femoral head necrosis decreased significantly from 2003 (38.83 ± 21.01)% to 2005 (30.38 ± 20.23)% (P = 0.000 2), then declined slowly from 2005 to 2013 (28.99 ± 20.59)% and plateaued until 2018 (25.52 ± 15.51)%. Pulmonary interstitial damage and functional decline caused by SARS mostly recovered, with a greater extent of recovery within 2 years after rehabilitation. Femoral head necrosis induced by large doses of steroid pulse therapy in SARS patients was not progressive and was partially reversible.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-03", "2020-08", "2020-09"], "last_mentioned": "2020-09-09T22:35:03Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-019-08824-8", "slug": "e558aef204b45470", "title": "Room temperature CO2 reduction to solid carbon species on liquid metals featuring atomically thin ceria interfaces", "authors": ["Dorna Esrafilzadeh", "Ali Zavabeti", "Rouhollah Jalili", "Paul Atkin", "Jaecheol Choi", "Benjamin J. Carey", "Robert Brkljača", "Anthony P. O’Mullane", "Michael D. Dickey", "David L. Officer", "Douglas R. MacFarlane", "Torben Daeneke", "Kourosh Kalantar-Zadeh"], "year": 2019, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-019-08824-8", "doi": "10.1038/s41467-019-08824-8", "topics": ["Physics"], "description": "AbstractNegative carbon emission technologies are critical for ensuring a future stable climate.", "description_method": "abstract-excerpt", "abstract": "AbstractNegative carbon emission technologies are critical for ensuring a future stable climate. However, the gaseous state of CO2 does render the indefinite storage of this greenhouse gas challenging. Herein, we created a liquid metal electrocatalyst that contains metallic elemental cerium nanoparticles, which facilitates the electrochemical reduction of CO2 to layered solid carbonaceous species, at a low onset potential of −310 mV vs CO2/C. We exploited the formation of a cerium oxide catalyst at the liquid metal/electrolyte interface, which together with cerium nanoparticles, promoted the room temperature reduction of CO2. Due to the inhibition of van der Waals adhesion at the liquid interface, the electrode was remarkably resistant to deactivation via coking caused by solid carbonaceous species. The as-produced solid carbonaceous materials could be utilised for the fabrication of high-performance capacitor electrodes. Overall, this liquid metal enabled electrocatalytic process at room temperature may result in a viable negative emission technology.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-06", "2021-04", "2021-08"], "last_mentioned": "2021-08-10T00:53:22Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-020-14555-y", "slug": "320d192dc6d4b6f4", "title": "Global resource potential of seasonal pumped hydropower storage for energy and water storage", "authors": ["Julian D. Hunt", "Edward Byers", "Yoshihide Wada", "Simon Parkinson", "David E. H. J. Gernaat", "Simon Langan", "Detlef P. van Vuuren", "Keywan Riahi"], "year": 2020, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-020-14555-y", "doi": "10.1038/s41467-020-14555-y", "topics": ["Distributed systems"], "description": "AbstractSeasonal mismatches between electricity supply and demand is increasing due to expanded use of wind, solar and hydropower resources, which in turn raises the interest on low-cost seasonal energy storage options.", "description_method": "abstract-excerpt", "abstract": "AbstractSeasonal mismatches between electricity supply and demand is increasing due to expanded use of wind, solar and hydropower resources, which in turn raises the interest on low-cost seasonal energy storage options. Seasonal pumped hydropower storage (SPHS) can provide long-term energy storage at a relatively low-cost and co-benefits in the form of freshwater storage capacity. We present the first estimate of the global assessment of SPHS potential, using a novel plant-siting methodology based on high-resolution topographical and hydrological data. Here we show that SPHS costs vary from 0.007 to 0.2 US$ m−1 of water stored, 1.8 to 50 US$ MWh−1 of energy stored and 370 to 600 US$ kW−1 of installed power generation. This potential is unevenly distributed with mountainous regions demonstrating significantly more potential. The estimated world energy storage capacity below a cost of 50 US$ MWh−1 is 17.3 PWh, approximately 79% of the world electricity consumption in 2017.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-08", "2023-12", "2024-02"], "last_mentioned": "2024-02-16T14:10:56Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-021-24790-6", "slug": "3260cb45f24edf7c", "title": "Gene-drive suppression of mosquito populations in large cages as a bridge between lab and field", "authors": ["Andrew Hammond", "Paola Pollegioni", "Tania Persampieri", "Ace North", "Roxana Minuz", "Alessandro Trusso", "Alessandro Bucci", "Kyros Kyrou", "Ioanna Morianou", "Alekos Simoni", "Tony Nolan", "Ruth Müller", "Andrea Crisanti"], "year": 2021, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-021-24790-6", "doi": "10.1038/s41467-021-24790-6", "topics": ["Economics", "Psychology", "Statistics & probability"], "description": "Abstract CRISPR-based gene-drives targeting the gene doublesex in the malaria vector Anopheles gambiae effectively suppressed the reproductive capability of mosquito populations reared in small laboratory cages.", "description_method": "abstract-excerpt", "abstract": "Abstract CRISPR-based gene-drives targeting the gene doublesex in the malaria vector Anopheles gambiae effectively suppressed the reproductive capability of mosquito populations reared in small laboratory cages. To bridge the gap between laboratory and the field, this gene-drive technology must be challenged with vector ecology. Here we report the suppressive activity of the gene-drive in age-structured An. gambiae populations in large indoor cages that permit complex feeding and reproductive behaviours. The gene-drive element spreads rapidly through the populations, fully supresses the population within one year and without selecting for resistance to the gene drive. Approximate Bayesian computation allowed retrospective inference of life-history parameters from the large cages and a more accurate prediction of gene-drive behaviour under more ecologically-relevant settings. Generating data to bridge laboratory and field studies for invasive technologies is challenging. Our study represents a paradigm for the stepwise and sound development of vector control tools based on gene-drive.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-10", "2023-04", "2023-08"], "last_mentioned": "2023-08-04T22:17:06Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-023-39810-w", "slug": "dd600daf50ce62a2", "title": "Warning of a forthcoming collapse of the Atlantic meridional overturning circulation", "authors": ["Peter Ditlevsen", "Susanne Ditlevsen"], "year": 2023, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-023-39810-w", "doi": "10.1038/s41467-023-39810-w", "topics": ["Statistics & probability"], "description": "Abstract The Atlantic meridional overturning circulation (AMOC) is a major tipping element in the climate system and a future collapse would have severe impacts on the climate in the North Atlantic region.", "description_method": "abstract-excerpt", "abstract": "Abstract The Atlantic meridional overturning circulation (AMOC) is a major tipping element in the climate system and a future collapse would have severe impacts on the climate in the North Atlantic region. In recent years weakening in circulation has been reported, but assessments by the Intergovernmental Panel on Climate Change (IPCC), based on the Climate Model Intercomparison Project (CMIP) model simulations suggest that a full collapse is unlikely within the 21st century. Tipping to an undesired state in the climate is, however, a growing concern with increasing greenhouse gas concentrations. Predictions based on observations rely on detecting early-warning signals, primarily an increase in variance (loss of resilience) and increased autocorrelation (critical slowing down), which have recently been reported for the AMOC. Here we provide statistical significance and data-driven estimators for the time of tipping. We estimate a collapse of the AMOC to occur around mid-century under the current scenario of future emissions.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-02", "2024-04", "2024-11"], "last_mentioned": "2024-11-17T21:09:22Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41467-026-69285-4", "slug": "aead030ba7d41c2b", "title": "National analysis of cancer mortality and proximity to nuclear power plants in the United States", "authors": ["Yazan Alwadi", "Barrak Alahmad", "Carolina L. Zilli Vieira", "Philip J. Landrigan", "David C. Christiani", "Eric Garshick", "Marco Kaltofen", "Brent Coull", "Joel Schwartz", "John S. Evans", "Petros Koutrakis"], "year": 2026, "venue": "Nature Communications", "url": "https://doi.org/10.1038/s41467-026-69285-4", "doi": "10.1038/s41467-026-69285-4", "topics": ["Biology & medicine", "Psychology", "Economics"], "description": "Abstract Understanding the potential health implications of living near nuclear power plants is important given the renewed interest in nuclear energy as a low-carbon power source.", "description_method": "abstract-excerpt", "abstract": "Abstract Understanding the potential health implications of living near nuclear power plants is important given the renewed interest in nuclear energy as a low-carbon power source. Here we show that U.S. counties located closer to operational nuclear power plants have higher cancer mortality rates than those farther away. Using nationwide mortality data from 2000-2018, we assess long-term spatial patterns of cancer mortality in relation to proximity to nuclear facilities while accounting for socioeconomic, demographic, behavioral, environmental, and healthcare factors. Cancer mortality is higher across multiple age groups in both males and females, with the strongest associations among older adults, males aged 65–74 and females aged 55–64. While our findings cannot establish causality, they highlight the need for further research into potential exposure pathways, latency effects, and cancer-specific risks, emphasizing the importance of addressing these potentially substantial but overlooked risks to public health.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2026-02", "2026-03", "2026-06"], "last_mentioned": "2026-06-29T18:44:49Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41560-017-0032-9", "slug": "1f2f1699fb9ba5af", "title": "Understanding future emissions from low-carbon power systems by integration of life-cycle assessment and integrated energy modelling", "authors": ["Michaja Pehl", "Anders Arvesen", "Florian Humpenöder", "Alexander Popp", "Edgar G. Hertwich", "Gunnar Luderer"], "year": 2017, "venue": "Nature Energy", "url": "https://doi.org/10.1038/s41560-017-0032-9", "doi": "10.1038/s41560-017-0032-9", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-07", "2021-04", "2022-01"], "last_mentioned": "2022-01-02T10:45:37Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41562-017-0189-z", "slug": "b9844101606dff62", "title": "Redefine statistical significance", "authors": ["Daniel J. Benjamin", "James O. Berger", "Magnus Johannesson", "Brian A. Nosek", "E.-J. Wagenmakers", "Richard Berk", "Kenneth A. Bollen", "Björn Brembs", "Lawrence Brown", "Colin Camerer", "David Cesarini", "Christopher D. Chambers", "Merlise Clyde", "Thomas D. Cook", "Paul De Boeck", "Zoltan Dienes", "Anna Dreber", "Kenny Easwaran", "Charles Efferson", "Ernst Fehr", "Fiona Fidler", "Andy P. Field", "Malcolm Forster", "Edward I. George", "Richard Gonzalez", "Steven Goodman", "Edwin Green", "Donald P. Green", "Anthony G. Greenwald", "Jarrod D. Hadfield", "Larry V. Hedges", "Leonhard Held", "Teck Hua Ho", "Herbert Hoijtink", "Daniel J. Hruschka", "Kosuke Imai", "Guido Imbens", "John P. A. Ioannidis", "Minjeong Jeon", "James Holland Jones", "Michael Kirchler", "David Laibson", "John List", "Roderick Little", "Arthur Lupia", "Edouard Machery", "Scott E. Maxwell", "Michael McCarthy", "Don A. Moore", "Stephen L. Morgan", "Marcus Munafó", "Shinichi Nakagawa", "Brendan Nyhan", "Timothy H. Parker", "Luis Pericchi", "Marco Perugini", "Jeff Rouder", "Judith Rousseau", "Victoria Savalei", "Felix D. Schönbrodt", "Thomas Sellke", "Betsy Sinclair", "Dustin Tingley", "Trisha Van Zandt", "Simine Vazire", "Duncan J. Watts", "Christopher Winship", "Robert L. Wolpert", "Yu Xie", "Cristobal Young", "Jonathan Zinman", "Valen E. Johnson"], "year": 2017, "venue": "Nature Human Behaviour", "url": "https://doi.org/10.1038/s41562-017-0189-z", "doi": "10.1038/s41562-017-0189-z", "topics": ["Statistics & probability"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2017-09", "2019-03", "2021-09"], "last_mentioned": "2021-09-15T04:01:35Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41562-018-0520-3", "slug": "12484270d12ac22f", "title": "Extreme opponents of genetically modified foods know the least but think they know the most", "authors": ["Philip M. Fernbach", "Nicholas Light", "Sydney E. Scott", "Yoel Inbar", "Paul Rozin"], "year": 2019, "venue": "Nature Human Behaviour", "url": "https://doi.org/10.1038/s41562-018-0520-3", "doi": "10.1038/s41562-018-0520-3", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-01", "2020-11", "2023-09"], "last_mentioned": "2023-09-14T19:33:01Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41562-022-01516-2", "slug": "fdd8973fb12bf1b8", "title": "Evidence of a predictive coding hierarchy in the human brain listening to speech", "authors": ["Charlotte Caucheteux", "Alexandre Gramfort", "Jean-Rémi King"], "year": 2023, "venue": "Nature Human Behaviour", "url": "https://doi.org/10.1038/s41562-022-01516-2", "doi": "10.1038/s41562-022-01516-2", "topics": ["Neuroscience", "Language models", "Psychology"], "description": "Abstract Considerable progress has recently been made in natural language processing: deep learning algorithms are increasingly able to generate, summarize, translate and classify texts.", "description_method": "abstract-excerpt", "abstract": "Abstract Considerable progress has recently been made in natural language processing: deep learning algorithms are increasingly able to generate, summarize, translate and classify texts. Yet, these language models still fail to match the language abilities of humans. Predictive coding theory offers a tentative explanation to this discrepancy: while language models are optimized to predict nearby words, the human brain would continuously predict a hierarchy of representations that spans multiple timescales. To test this hypothesis, we analysed the functional magnetic resonance imaging brain signals of 304 participants listening to short stories. First, we confirmed that the activations of modern language models linearly map onto the brain responses to speech. Second, we showed that enhancing these algorithms with predictions that span multiple timescales improves this brain mapping. Finally, we showed that these predictions are organized hierarchically: frontoparietal cortices predict higher-level, longer-range and more contextual representations than temporal cortices. Overall, these results strengthen the role of hierarchical predictive coding in language processing and illustrate how the synergy between neuroscience and artificial intelligence can unravel the computational bases of human cognition.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-03", "2023-08", "2023-11"], "last_mentioned": "2023-11-10T06:04:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41562-024-01882-z", "slug": "5e58e62c6d5bf7f8", "title": "Testing theory of mind in large language models and humans", "authors": ["James W. A. Strachan", "Dalila Albergo", "Giulia Borghini", "Oriana Pansardi", "Eugenio Scaliti", "Saurabh Gupta", "Krati Saxena", "Alessandro Rufo", "Stefano Panzeri", "Guido Manzi", "Michael S. A. Graziano", "Cristina Becchio"], "year": 2024, "venue": "Nature Human Behaviour", "url": "https://doi.org/10.1038/s41562-024-01882-z", "doi": "10.1038/s41562-024-01882-z", "topics": ["Language models", "Psychology", "Robotics"], "description": "Abstract At the core of what defines us as humans is the concept of theory of mind: the ability to track other people’s mental states.", "description_method": "abstract-excerpt", "abstract": "Abstract At the core of what defines us as humans is the concept of theory of mind: the ability to track other people’s mental states. The recent development of large language models (LLMs) such as ChatGPT has led to intense debate about the possibility that these models exhibit behaviour that is indistinguishable from human behaviour in theory of mind tasks. Here we compare human and LLM performance on a comprehensive battery of measurements that aim to measure different theory of mind abilities, from understanding false beliefs to interpreting indirect requests and recognizing irony and faux pas. We tested two families of LLMs (GPT and LLaMA2) repeatedly against these measures and compared their performance with those from a sample of 1,907 human participants. Across the battery of theory of mind tests, we found that GPT-4 models performed at, or even sometimes above, human levels at identifying indirect requests, false beliefs and misdirection, but struggled with detecting faux pas. Faux pas, however, was the only test where LLaMA2 outperformed humans. Follow-up manipulations of the belief likelihood revealed that the superiority of LLaMA2 was illusory, possibly reflecting a bias towards attributing ignorance. By contrast, the poor performance of GPT originated from a hyperconservative approach towards committing to conclusions rather than from a genuine failure of inference. These findings not only demonstrate that LLMs exhibit behaviour that is consistent with the outputs of mentalistic inference in humans but also highlight the importance of systematic testing to ensure a non-superficial comparison between human and artificial intelligences.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-06", "2025-08", "2025-10"], "last_mentioned": "2025-10-27T17:31:21Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41565-018-0337-2", "slug": "591b854f1d323109", "title": "A note on the Lena image", "authors": [], "year": 2018, "venue": "Nature Nanotechnology", "url": "https://doi.org/10.1038/s41565-018-0337-2", "doi": "10.1038/s41565-018-0337-2", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-03", "2023-06", "2023-12"], "last_mentioned": "2023-12-07T17:46:26Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41567-023-02230-0", "slug": "b75b483c02cfd956", "title": "Unfounded concerns about photovoltaic module toxicity and waste are slowing decarbonization", "authors": ["Heather Mirletz", "Henry Hieslmair", "Silvana Ovaitt", "Taylor L. Curtis", "Teresa M. Barnes"], "year": 2023, "venue": "Nature Physics", "url": "https://doi.org/10.1038/s41567-023-02230-0", "doi": "10.1038/s41567-023-02230-0", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-10", "2024-01", "2025-07"], "last_mentioned": "2025-07-11T01:31:08Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41577-020-00436-4", "slug": "24ed8eca037827d6", "title": "Not just antibodies: B cells and T cells mediate immunity to COVID-19", "authors": ["Rebecca J. Cox", "Karl A. Brokstad"], "year": 2020, "venue": "Nature Reviews Immunology", "url": "https://doi.org/10.1038/s41577-020-00436-4", "doi": "10.1038/s41577-020-00436-4", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-09", "2020-12", "2022-02"], "last_mentioned": "2022-02-03T18:23:24Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-018-0102-6", "slug": "03b7fb31ccb71730", "title": "Vector-based navigation using grid-like representations in artificial agents", "authors": ["Andrea Banino", "Caswell Barry", "Benigno Uria", "Charles Blundell", "Timothy Lillicrap", "Piotr Mirowski", "Alexander Pritzel", "Martin J. Chadwick", "Thomas Degris", "Joseph Modayil", "Greg Wayne", "Hubert Soyer", "Fabio Viola", "Brian Zhang", "Ross Goroshin", "Neil Rabinowitz", "Razvan Pascanu", "Charlie Beattie", "Stig Petersen", "Amir Sadik", "Stephen Gaffney", "Helen King", "Koray Kavukcuoglu", "Demis Hassabis", "Raia Hadsell", "Dharshan Kumaran"], "year": 2018, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-018-0102-6", "doi": "10.1038/s41586-018-0102-6", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-05", "2024-06", "2025-11"], "last_mentioned": "2025-11-11T07:15:52Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-018-0106-2", "slug": "43c7bbbdad533758", "title": "An unexpected and persistent increase in global emissions of ozone-depleting CFC-11", "authors": ["Stephen A. Montzka", "Geoff S. Dutton", "Pengfei Yu", "Eric Ray", "Robert W. Portmann", "John S. Daniel", "Lambert Kuijpers", "Brad D. Hall", "Debra Mondeel", "Carolina Siso", "J. David Nance", "Matt Rigby", "Alistair J. Manning", "Lei Hu", "Fred Moore", "Ben R. Miller", "James W. Elkins"], "year": 2018, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-018-0106-2", "doi": "10.1038/s41586-018-0106-2", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-05", "2018-07", "2026-07"], "last_mentioned": "2026-07-29T04:52:26Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-019-1043-4", "slug": "27297d4d79cf5456", "title": "RETRACTED ARTICLE: Complex societies precede moralizing gods throughout world history", "authors": ["Harvey Whitehouse", "Pieter François", "Patrick E. Savage", "Thomas E. Currie", "Kevin C. Feeney", "Enrico Cioni", "Rosalind Purcell", "Robert M. Ross", "Jennifer Larson", "John Baines", "Barend ter Haar", "Alan Covey", "Peter Turchin"], "year": 2019, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-019-1043-4", "doi": "10.1038/s41586-019-1043-4", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-04", "2021-03", "2021-07"], "last_mentioned": "2021-07-06T15:09:47Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2010-9", "slug": "2d837cf11ff1cc53", "title": "Power generation from ambient humidity using protein nanowires", "authors": ["Xiaomeng Liu", "Hongyan Gao", "Joy E. Ward", "Xiaorong Liu", "Bing Yin", "Tianda Fu", "Jianhan Chen", "Derek R. Lovley", "Jun Yao"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2010-9", "doi": "10.1038/s41586-020-2010-9", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-04", "2023-06", "2023-07"], "last_mentioned": "2023-07-04T23:41:32Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2342-5", "slug": "80581873dba12018", "title": "Pathogenesis and transmission of SARS-CoV-2 in golden hamsters", "authors": ["Sin Fun Sia", "Li-Meng Yan", "Alex W. H. Chin", "Kevin Fung", "Ka-Tim Choy", "Alvina Y. L. Wong", "Prathanporn Kaewpreedee", "Ranawaka A. P. M. Perera", "Leo L. M. Poon", "John M. Nicholls", "Malik Peiris", "Hui-Ling Yen"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2342-5", "doi": "10.1038/s41586-020-2342-5", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-09", "2020-10", "2022-01"], "last_mentioned": "2022-01-18T20:26:25Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2639-4", "slug": "af7deb23571b37a2", "title": "Phase I/II study of COVID-19 RNA vaccine BNT162b1 in adults", "authors": ["Mark J. Mulligan", "Kirsten E. Lyke", "Nicholas Kitchin", "Judith Absalon", "Alejandra Gurtman", "Stephen Lockhart", "Kathleen Neuzil", "Vanessa Raabe", "Ruth Bailey", "Kena A. Swanson", "Ping Li", "Kenneth Koury", "Warren Kalina", "David Cooper", "Camila Fontes-Garfias", "Pei-Yong Shi", "Özlem Türeci", "Kristin R. Tompkins", "Edward E. Walsh", "Robert Frenck", "Ann R. Falsey", "Philip R. Dormitzer", "William C. Gruber", "Uğur Şahin", "Kathrin U. Jansen"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2639-4", "doi": "10.1038/s41586-020-2639-4", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-09", "2021-11", "2022-04"], "last_mentioned": "2022-04-28T19:02:21Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-020-2939-8", "slug": "c466f07e8821cc14", "title": "Autonomous navigation of stratospheric balloons using reinforcement learning", "authors": ["Marc G. Bellemare", "Salvatore Candido", "Pablo Samuel Castro", "Jun Gong", "Marlos C. Machado", "Subhodeep Moitra", "Sameera S. Ponda", "Ziyu Wang"], "year": 2020, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-020-2939-8", "doi": "10.1038/s41586-020-2939-8", "topics": ["Robotics", "Machine learning"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-12", "2021-01", "2023-02"], "last_mentioned": "2023-02-03T01:13:00Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-021-03380-y", "slug": "a08133a94adbb962", "title": "People systematically overlook subtractive changes", "authors": ["Gabrielle S. Adams", "Benjamin A. Converse", "Andrew H. Hales", "Leidy E. Klotz"], "year": 2021, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-021-03380-y", "doi": "10.1038/s41586-021-03380-y", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-06", "2023-04", "2026-02"], "last_mentioned": "2026-02-25T20:50:37Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-021-04223-6", "slug": "faf6e6b3508ed3e8", "title": "Deep physical neural networks trained with backpropagation", "authors": ["Logan G. Wright", "Tatsuhiro Onodera", "Martin M. Stein", "Tianyu Wang", "Darren T. Schachter", "Zoey Hu", "Peter L. McMahon"], "year": 2022, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-021-04223-6", "doi": "10.1038/s41586-021-04223-6", "topics": ["Machine learning", "Algorithms", "Physics"], "description": "Abstract Deep-learning models have become pervasive tools in science and engineering.", "description_method": "abstract-excerpt", "abstract": "Abstract Deep-learning models have become pervasive tools in science and engineering. However, their energy requirements now increasingly limit their scalability 1 . Deep-learning accelerators 2–9 aim to perform deep learning energy-efficiently, usually targeting the inference phase and often by exploiting physical substrates beyond conventional electronics. Approaches so far 10–22 have been unable to apply the backpropagation algorithm to train unconventional novel hardware in situ. The advantages of backpropagation have made it the de facto training method for large-scale neural networks, so this deficiency constitutes a major impediment. Here we introduce a hybrid in situ–in silico algorithm, called physics-aware training, that applies backpropagation to train controllable physical systems. Just as deep learning realizes computations with deep neural networks made from layers of mathematical functions, our approach allows us to train deep physical neural networks made from layers of controllable physical systems, even when the physical layers lack any mathematical isomorphism to conventional artificial neural network layers. To demonstrate the universality of our approach, we train diverse physical neural networks based on optics, mechanics and electronics to experimentally perform audio and image classification tasks. Physics-aware training combines the scalability of backpropagation with the automatic mitigation of imperfections and noise achievable with in situ algorithms. Physical neural networks have the potential to perform machine learning faster and more energy-efficiently than conventional electronic processors and, more broadly, can endow physical systems with automatically designed physical functionalities, for example, for robotics 23–26 , materials 27–29 and smart sensors 30–32 .", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-02", "2022-04", "2025-01"], "last_mentioned": "2025-01-15T23:47:08Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-021-04301-9", "slug": "02273b708a3db1ba", "title": "Magnetic control of tokamak plasmas through deep reinforcement learning", "authors": ["Jonas Degrave", "Federico Felici", "Jonas Buchli", "Michael Neunert", "Brendan Tracey", "Francesco Carpanese", "Timo Ewalds", "Roland Hafner", "Abbas Abdolmaleki", "Diego de las Casas", "Craig Donner", "Leslie Fritz", "Cristian Galperti", "Andrea Huber", "James Keeling", "Maria Tsimpoukelli", "Jackie Kay", "Antoine Merle", "Jean-Marc Moret", "Seb Noury", "Federico Pesamosca", "David Pfau", "Olivier Sauter", "Cristian Sommariva", "Stefano Coda", "Basil Duval", "Ambrogio Fasoli", "Pushmeet Kohli", "Koray Kavukcuoglu", "Demis Hassabis", "Martin Riedmiller"], "year": 2022, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-021-04301-9", "doi": "10.1038/s41586-021-04301-9", "topics": ["Machine learning", "Physics", "Robotics"], "description": "Abstract Nuclear fusion using magnetic confinement, in particular in the tokamak configuration, is a promising path towards sustainable energy.", "description_method": "abstract-excerpt", "abstract": "Abstract Nuclear fusion using magnetic confinement, in particular in the tokamak configuration, is a promising path towards sustainable energy. A core challenge is to shape and maintain a high-temperature plasma within the tokamak vessel. This requires high-dimensional, high-frequency, closed-loop control using magnetic actuator coils, further complicated by the diverse requirements across a wide range of plasma configurations. In this work, we introduce a previously undescribed architecture for tokamak magnetic controller design that autonomously learns to command the full set of control coils. This architecture meets control objectives specified at a high level, at the same time satisfying physical and operational constraints. This approach has unprecedented flexibility and generality in problem specification and yields a notable reduction in design effort to produce new plasma configurations. We successfully produce and control a diverse set of plasma configurations on the Tokamak à Configuration Variable 1,2 , including elongated, conventional shapes, as well as advanced configurations, such as negative triangularity and ‘snowflake’ configurations. Our approach achieves accurate tracking of the location, current and shape for these configurations. We also demonstrate sustained ‘droplets’ on TCV, in which two separate plasmas are maintained simultaneously within the vessel. This represents a notable advance for tokamak feedback control, showing the potential of reinforcement learning to accelerate research in the fusion domain, and is one of the most challenging real-world systems to which reinforcement learning has been applied.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-08", "2023-12", "2024-05"], "last_mentioned": "2024-05-09T14:37:29Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41586-023-06083-8", "slug": "26e1fc223afa12bd", "title": "Safe and just Earth system boundaries", "authors": ["Johan Rockström", "Joyeeta Gupta", "Dahe Qin", "Steven J. Lade", "Jesse F. Abrams", "Lauren S. Andersen", "David I. Armstrong McKay", "Xuemei Bai", "Govindasamy Bala", "Stuart E. Bunn", "Daniel Ciobanu", "Fabrice DeClerck", "Kristie Ebi", "Lauren Gifford", "Christopher Gordon", "Syezlin Hasan", "Norichika Kanie", "Timothy M. Lenton", "Sina Loriani", "Diana M. Liverman", "Awaz Mohamed", "Nebojsa Nakicenovic", "David Obura", "Daniel Ospina", "Klaudia Prodani", "Crelis Rammelt", "Boris Sakschewski", "Joeri Scholtens", "Ben Stewart-Koster", "Thejna Tharammal", "Detlef van Vuuren", "Peter H. Verburg", "Ricarda Winkelmann", "Caroline Zimm", "Elena M. Bennett", "Stefan Bringezu", "Wendy Broadgate", "Pamela A. Green", "Lei Huang", "Lisa Jacobson", "Christopher Ndehedehe", "Simona Pedde", "Juan Rocha", "Marten Scheffer", "Lena Schulte-Uebbing", "Wim de Vries", "Cunde Xiao", "Chi Xu", "Xinwu Xu", "Noelia Zafra-Calvo", "Xin Zhang"], "year": 2023, "venue": "Nature", "url": "https://doi.org/10.1038/s41586-023-06083-8", "doi": "10.1038/s41586-023-06083-8", "topics": ["Other research"], "description": "Abstract The stability and resilience of the Earth system and human well-being are inseparably linked 1–3 , yet their interdependencies are generally under-recognized; consequently, they are often treated independently 4,5 .", "description_method": "abstract-excerpt", "abstract": "Abstract The stability and resilience of the Earth system and human well-being are inseparably linked 1–3 , yet their interdependencies are generally under-recognized; consequently, they are often treated independently 4,5 . Here, we use modelling and literature assessment to quantify safe and just Earth system boundaries (ESBs) for climate, the biosphere, water and nutrient cycles, and aerosols at global and subglobal scales. We propose ESBs for maintaining the resilience and stability of the Earth system (safe ESBs) and minimizing exposure to significant harm to humans from Earth system change (a necessary but not sufficient condition for justice) 4 . The stricter of the safe or just boundaries sets the integrated safe and just ESB. Our findings show that justice considerations constrain the integrated ESBs more than safety considerations for climate and atmospheric aerosol loading. Seven of eight globally quantified safe and just ESBs and at least two regional safe and just ESBs in over half of global land area are already exceeded. We propose that our assessment provides a quantitative foundation for safeguarding the global commons for all people now and into the future.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-06", "2023-07", "2023-12"], "last_mentioned": "2023-12-24T07:47:42Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-021-01571-8", "slug": "bfd4d66e67aab2b1", "title": "No causal effect of school closures in Japan on the spread of COVID-19 in spring 2020", "authors": ["Kentaro Fukumoto", "Charles T. McClean", "Kuninori Nakagawa"], "year": 2021, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-021-01571-8", "doi": "10.1038/s41591-021-01571-8", "topics": ["Statistics & probability", "Distributed systems", "Biology & medicine"], "description": "Abstract Among tool kits to combat the coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2, school closures are one of the most frequent non-pharmaceutical interventions.", "description_method": "abstract-excerpt", "abstract": "Abstract Among tool kits to combat the coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2, school closures are one of the most frequent non-pharmaceutical interventions. However, school closures bring about substantial costs, such as learning loss. To date, studies have not reached a consensus about the effectiveness of these policies at mitigating community transmission, partly because they lack rigorous causal inference. Here we assess the causal effect of school closures in Japan on reducing the spread of COVID-19 in spring 2020. By matching each municipality with open schools to a municipality with closed schools that is the most similar in terms of potential confounders, we can estimate how many cases the municipality with open schools would have had if it had closed its schools. We do not find any evidence that school closures in Japan reduced the spread of COVID-19. Our null results suggest that policies on school closures should be reexamined given the potential negative consequences for children and parents.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-11", "2022-01", "2022-03"], "last_mentioned": "2022-03-29T06:47:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-022-01810-6", "slug": "b1cf6d5d4c056536", "title": "Unexplained post-acute infection syndromes", "authors": ["Jan Choutka", "Viraj Jansari", "Mady Hornig", "Akiko Iwasaki"], "year": 2022, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-022-01810-6", "doi": "10.1038/s41591-022-01810-6", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-08", "2024-01", "2024-11"], "last_mentioned": "2024-11-30T10:05:16Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41591-022-01816-0", "slug": "64f194724e629cf4", "title": "Infectious viral load in unvaccinated and vaccinated individuals infected with ancestral, Delta or Omicron SARS-CoV-2", "authors": ["Olha Puhach", "Kenneth Adea", "Nicolas Hulo", "Pascale Sattonnet", "Camille Genecand", "Anne Iten", "Frédérique Jacquérioz", "Laurent Kaiser", "Pauline Vetter", "Isabella Eckerle", "Benjamin Meyer"], "year": 2022, "venue": "Nature Medicine", "url": "https://doi.org/10.1038/s41591-022-01816-0", "doi": "10.1038/s41591-022-01816-0", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-07", "2022-09", "2023-10"], "last_mentioned": "2023-10-01T17:46:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-017-17497-6", "slug": "2f6064be0ce7b343", "title": "Long term outcomes in survivors of epidemic Influenza A (H7N9) virus infection", "authors": ["Jiajia Chen", "Jie Wu", "Shaorui Hao", "Meifang Yang", "Xiaoqing Lu", "Xiaoxiao Chen", "Lanjuan Li"], "year": 2017, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-017-17497-6", "doi": "10.1038/s41598-017-17497-6", "topics": ["Psychology", "Physics"], "description": "Abstract Patients who survive influenza A (H7N9) virus infection are at risk of physical and psychological complications of lung injury and multi-organ dysfunction.", "description_method": "abstract-excerpt", "abstract": "Abstract Patients who survive influenza A (H7N9) virus infection are at risk of physical and psychological complications of lung injury and multi-organ dysfunction. However, there were no prospectively individualized assessments of physiological, functional and quality-of-life measures after hospital discharge. The current study aims to assess the main determinants of functional disability of these patients during the follow-up. Fifty-six influenza A (H7N9) survivors were investigated during the 2-year after discharge from the hospital. Results show interstitial change and fibrosis on pulmonary imaging remained 6 months after hospital discharge. Both ventilation and diffusion dysfunction improved, but restrictive and obstructive patterns on ventilation function test persisted throughout the follow-up period. For patients with acute respiratory distress syndrome lung functions improved faster during the first six months. Role-physical and Role-emotional domains in the 36-Item Short-Form Health Survey were worse than those of a sex- and age-matched general population group. The quality of life of survivors with ARDS was lower than those with no ARDS. Our findings suggest that pulmonary function and imaging findings improved during the first 6 months especially for those with ARDS, however long-term lung disability and psychological impairment in H7N9 survivors persisted at 2 years after discharge from the hospital.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-08", "2021-08", "2022-09"], "last_mentioned": "2022-09-06T05:43:26Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-018-23948-5", "slug": "74b1baf526bea6dc", "title": "The tremendous potential of deep-sea mud as a source of rare-earth elements", "authors": ["Yutaro Takaya", "Kazutaka Yasukawa", "Takehiro Kawasaki", "Koichiro Fujinaga", "Junichiro Ohta", "Yoichi Usui", "Kentaro Nakamura", "Jun-Ichi Kimura", "Qing Chang", "Morihisa Hamada", "Gjergj Dodbiba", "Tatsuo Nozaki", "Koichi Iijima", "Tomohiro Morisawa", "Takuma Kuwahara", "Yasuyuki Ishida", "Takao Ichimura", "Masaki Kitazume", "Toyohisa Fujita", "Yasuhiro Kato"], "year": 2018, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-018-23948-5", "doi": "10.1038/s41598-018-23948-5", "topics": ["Economics"], "description": "Abstract Potential risks of supply shortages for critical metals including rare-earth elements and yttrium (REY) have spurred great interest in commercial mining of deep-sea mineral resources.", "description_method": "abstract-excerpt", "abstract": "Abstract Potential risks of supply shortages for critical metals including rare-earth elements and yttrium (REY) have spurred great interest in commercial mining of deep-sea mineral resources. Deep-sea mud containing over 5,000 ppm total REY content was discovered in the western North Pacific Ocean near Minamitorishima Island, Japan, in 2013. This REY-rich mud has great potential as a rare-earth metal resource because of the enormous amount available and its advantageous mineralogical features. Here, we estimated the resource amount in REY-rich mud with Geographical Information System software and established a mineral processing procedure to greatly enhance its economic value. The resource amount was estimated to be 1.2 Mt of rare-earth oxide for the most promising area (105 km 2 × 0–10 mbsf), which accounts for 62, 47, 32, and 56 years of annual global demand for Y, Eu, Tb, and Dy, respectively. Moreover, using a hydrocyclone separator enabled us to recover selectively biogenic calcium phosphate grains, which have high REY content (up to 22,000 ppm) and constitute the coarser domain in the grain-size distribution. The enormous resource amount and the effectiveness of the mineral processing are strong indicators that this new REY resource could be exploited in the near future.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-04", "2018-07", "2019-05"], "last_mentioned": "2019-05-31T16:01:54Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-018-38089-y", "slug": "9c485d211bfd0f73", "title": "Sedimentary record from Patagonia, southern Chile supports cosmic-impact triggering of biomass burning, climate change, and megafaunal extinctions at 12.8 ka", "authors": ["Mario Pino", "Ana M. Abarzúa", "Giselle Astorga", "Alejandra Martel-Cea", "Nathalie Cossio-Montecinos", "R. Ximena Navarro", "Maria Paz Lira", "Rafael Labarca", "Malcolm A. LeCompte", "Victor Adedeji", "Christopher R. Moore", "Ted E. Bunch", "Charles Mooney", "Wendy S. Wolbach", "Allen West", "James P. Kennett"], "year": 2019, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-018-38089-y", "doi": "10.1038/s41598-018-38089-y", "topics": ["Statistics & probability", "Physics"], "description": "Abstract The Younger Dryas (YD) impact hypothesis posits that fragments of a large, disintegrating asteroid/comet struck North America, South America, Europe, and western Asia ~12,800 years ago.", "description_method": "abstract-excerpt", "abstract": "Abstract The Younger Dryas (YD) impact hypothesis posits that fragments of a large, disintegrating asteroid/comet struck North America, South America, Europe, and western Asia ~12,800 years ago. Multiple airbursts/impacts produced the YD boundary layer (YDB), depositing peak concentrations of platinum, high-temperature spherules, meltglass, and nanodiamonds, forming an isochronous datum at >50 sites across ~50 million km² of Earth’s surface. This proposed event triggered extensive biomass burning, brief impact winter, YD climate change, and contributed to extinctions of late Pleistocene megafauna. In the most extensive investigation south of the equator, we report on a ~12,800-year-old sequence at Pilauco, Chile (~40°S), that exhibits peak YD boundary concentrations of platinum, gold, high-temperature iron- and chromium-rich spherules, and native iron particles rarely found in nature. A major peak in charcoal abundance marks an intense biomass-burning episode, synchronous with dramatic changes in vegetation, including a high-disturbance regime, seasonality in precipitation, and warmer conditions. This is anti-phased with northern-hemispheric cooling at the YD onset, whose rapidity suggests atmospheric linkage. The sudden disappearance of megafaunal remains and dung fungi in the YDB layer at Pilauco correlates with megafaunal extinctions across the Americas. The Pilauco record appears consistent with YDB impact evidence found at sites on four continents.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-04", "2019-05", "2025-07"], "last_mentioned": "2025-07-02T22:10:39Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-020-66170-y", "slug": "baa82563f3a34ef0", "title": "Fuelling walking and cycling: human powered locomotion is associated with non-negligible greenhouse gas emissions", "authors": ["Anja Mizdrak", "Linda J Cobiac", "Christine L Cleghorn", "Alistair Woodward", "Tony Blakely"], "year": 2020, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-020-66170-y", "doi": "10.1038/s41598-020-66170-y", "topics": ["Economics", "Robotics"], "description": "AbstractReducing motorized transport and increasing active transport (i.e.", "description_method": "abstract-excerpt", "abstract": "AbstractReducing motorized transport and increasing active transport (i.e. transport by walking, cycling and other active modes) may reduce greenhouse gas (GHG) emissions and improve health. But, active modes of transport are not zero emitters. We aimed to quantify GHG emissions from food production required to fuel extra physical activity for walking and cycling. We estimate the emissions (in kgCO2e) per kilometre travelled for walking and cycling from energy intake required to compensate for increased energy expenditure, and data on food-related GHG emissions. We assume that persons who shift from passive modes of transport (e.g. driving) have increased energy expenditure that may be compensated with increased food consumption. The GHG emissions associated with food intake required to fuel a kilometre of walking range between 0.05 kgCO2e/km in the least economically developed countries to 0.26 kgCO2e/km in the most economically developed countries. Emissions for cycling are approximately half those of walking. Emissions from food required for walking and cycling are not negligible in economically developed countries which have high dietary-related emissions. There is high uncertainty about the actual emissions associated with walking and cycling, and high variability based on country economic development. Our study highlights the need to consider emissions from other sectors when estimating net-emissions impacts from transport interventions.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-01", "2025-01", "2025-06"], "last_mentioned": "2025-06-21T17:54:18Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-021-89909-7", "slug": "fe28c7562dc8943a", "title": "Satellite mega-constellations create risks in Low Earth Orbit, the atmosphere and on Earth", "authors": ["Aaron C. Boley", "Michael Byers"], "year": 2021, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-021-89909-7", "doi": "10.1038/s41598-021-89909-7", "topics": ["Other research"], "description": "AbstractThe rapid development of mega-constellations risks multiple tragedies of the commons, including tragedies to ground-based astronomy, Earth orbit, and Earth’s upper atmosphere.", "description_method": "abstract-excerpt", "abstract": "AbstractThe rapid development of mega-constellations risks multiple tragedies of the commons, including tragedies to ground-based astronomy, Earth orbit, and Earth’s upper atmosphere. Moreover, the connections between the Earth and space environments are inadequately taken into account by the adoption of a consumer electronic model applied to space assets. For example, we point out that satellite re-entries from the Starlink mega-constellation alone could deposit more aluminum into Earth’s upper atmosphere than what is done through meteoroids; they could thus become the dominant source of high-altitude alumina. Using simple models, we also show that untracked debris will lead to potentially dangerous on-orbit collisions on a regular basis due to the large number of satellites within mega-constellation orbital shells. The total cross-section of satellites in these constellations also greatly increases the risk of impacts due to meteoroids. De facto orbit occupation by single actors, inadequate regulatory frameworks, and the possibility of free-riding exacerbate these risks. International cooperation is urgently needed, along with a regulatory system that takes into account the effects of tens of thousands of satellites.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-07", "2023-09", "2025-10"], "last_mentioned": "2025-10-09T06:38:10Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-021-91470-2", "slug": "83a84d31c12a9aaa", "title": "Animal sales from Wuhan wet markets immediately prior to the COVID-19 pandemic", "authors": ["Xiao Xiao", "Chris Newman", "Christina D. Buesching", "David W. Macdonald", "Zhao-Min Zhou"], "year": 2021, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-021-91470-2", "doi": "10.1038/s41598-021-91470-2", "topics": ["Economics", "Biology & medicine"], "description": "AbstractHere we document 47,381 individuals from 38 species, including 31 protected species sold between May 2017 and November 2019 in Wuhan’s markets.", "description_method": "abstract-excerpt", "abstract": "AbstractHere we document 47,381 individuals from 38 species, including 31 protected species sold between May 2017 and November 2019 in Wuhan’s markets. We note that no pangolins (or bats) were traded, supporting reformed opinion that pangolins were not likely the spillover host at the source of the current coronavirus (COVID-19) pandemic. While we caution against the misattribution of COVID-19’s origins, the wild animals on sale in Wuhan suffered poor welfare and hygiene conditions and we detail a range of other zoonotic infections they can potentially vector. Nevertheless, in a precautionary response to COVID-19, China’s Ministries temporarily banned all wildlife trade on 26th Jan 2020 until the COVID-19 pandemic concludes, and permanently banned eating and trading terrestrial wild (non-livestock) animals for food on 24th Feb 2020. These interventions, intended to protect human health, redress previous trading and enforcement inconsistencies, and will have collateral benefits for global biodiversity conservation and animal welfare.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-07", "2021-08", "2022-09"], "last_mentioned": "2022-09-06T17:49:57Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-021-95565-8", "slug": "19012bd7cf4eb878", "title": "More than 50 long-term effects of COVID-19: a systematic review and meta-analysis", "authors": ["Sandra Lopez-Leon", "Talia Wegman-Ostrosky", "Carol Perelman", "Rosalinda Sepulveda", "Paulina A. Rebolledo", "Angelica Cuapio", "Sonia Villapol"], "year": 2021, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-021-95565-8", "doi": "10.1038/s41598-021-95565-8", "topics": ["Biology & medicine", "Science & research"], "description": "Abstract COVID-19 can involve persistence, sequelae, and other medical complications that last weeks to months after initial recovery.", "description_method": "abstract-excerpt", "abstract": "Abstract COVID-19 can involve persistence, sequelae, and other medical complications that last weeks to months after initial recovery. This systematic review and meta-analysis aims to identify studies assessing the long-term effects of COVID-19. LitCOVID and Embase were searched to identify articles with original data published before the 1st of January 2021, with a minimum of 100 patients. For effects reported in two or more studies, meta-analyses using a random-effects model were performed using the MetaXL software to estimate the pooled prevalence with 95% CI. PRISMA guidelines were followed. A total of 18,251 publications were identified, of which 15 met the inclusion criteria. The prevalence of 55 long-term effects was estimated, 21 meta-analyses were performed, and 47,910 patients were included (age 17–87 years). The included studies defined long-COVID as ranging from 14 to 110 days post-viral infection. It was estimated that 80% of the infected patients with SARS-CoV-2 developed one or more long-term symptoms. The five most common symptoms were fatigue (58%), headache (44%), attention disorder (27%), hair loss (25%), and dyspnea (24%). Multi-disciplinary teams are crucial to developing preventive measures, rehabilitation techniques, and clinical management strategies with whole-patient perspectives designed to address long COVID-19 care.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-10", "2021-12", "2022-07"], "last_mentioned": "2022-07-08T05:14:15Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-022-06749-9", "slug": "9848976895c48df6", "title": "Chlamydia pneumoniae can infect the central nervous system via the olfactory and trigeminal nerves and contributes to Alzheimer’s disease risk", "authors": ["Anu Chacko", "Ali Delbaz", "Heidi Walkden", "Souptik Basu", "Charles W. Armitage", "Tanja Eindorf", "Logan K. Trim", "Edith Miller", "Nicholas P. West", "James A. St John", "Kenneth W. Beagley", "Jenny A. K. Ekberg"], "year": 2022, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-022-06749-9", "doi": "10.1038/s41598-022-06749-9", "topics": ["Biology & medicine", "Neuroscience"], "description": "Abstract Chlamydia pneumoniae is a respiratory tract pathogen but can also infect the central nervous system (CNS).", "description_method": "abstract-excerpt", "abstract": "Abstract Chlamydia pneumoniae is a respiratory tract pathogen but can also infect the central nervous system (CNS). Recently, the link between C. pneumoniae CNS infection and late-onset dementia has become increasingly evident. In mice, CNS infection has been shown to occur weeks to months after intranasal inoculation. By isolating live C. pneumoniae from tissues and using immunohistochemistry, we show that C. pneumoniae can infect the olfactory and trigeminal nerves, olfactory bulb and brain within 72 h in mice. C. pneumoniae infection also resulted in dysregulation of key pathways involved in Alzheimer’s disease pathogenesis at 7 and 28 days after inoculation. Interestingly, amyloid beta accumulations were also detected adjacent to the C. pneumoniae inclusions in the olfactory system. Furthermore, injury to the nasal epithelium resulted in increased peripheral nerve and olfactory bulb infection, but did not alter general CNS infection. In vitro, C. pneumoniae was able to infect peripheral nerve and CNS glia. In summary, the nerves extending between the nasal cavity and the brain constitute invasion paths by which C. pneumoniae can rapidly invade the CNS likely by surviving in glia and leading to Aβ deposition.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-10", "2024-09", "2025-08"], "last_mentioned": "2025-08-11T13:29:39Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-022-24053-4", "slug": "198dd0b25a402b04", "title": "Association between vitamin D supplementation and COVID-19 infection and mortality", "authors": ["Jason B. Gibbons", "Edward C. Norton", "Jeffrey S. McCullough", "David O. Meltzer", "Jill Lavigne", "Virginia C. Fiedler", "Robert D. Gibbons"], "year": 2022, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-022-24053-4", "doi": "10.1038/s41598-022-24053-4", "topics": ["Other research"], "description": "AbstractVitamin D deficiency has long been associated with reduced immune function that can lead to viral infection.", "description_method": "abstract-excerpt", "abstract": "AbstractVitamin D deficiency has long been associated with reduced immune function that can lead to viral infection. Several studies have shown that Vitamin D deficiency is associated with increases the risk of infection with COVID-19. However, it is unknown if treatment with Vitamin D can reduce the associated risk of COVID-19 infection, which is the focus of this study. In the population of US veterans, we show that Vitamin D2and D3fills were associated with reductions in COVID-19 infection of 28% and 20%, respectively [(D3Hazard Ratio (HR) = 0.80, [95% CI 0.77, 0.83]), D2HR = 0.72, [95% CI 0.65, 0.79]]. Mortality within 30-days of COVID-19 infection was similarly 33% lower with Vitamin D3and 25% lower with D2(D3HR = 0.67, [95% CI 0.59, 0.75]; D2HR = 0.75, [95% CI 0.55, 1.04]). We also find that after controlling for vitamin D blood levels, veterans receiving higher dosages of Vitamin D obtained greater benefits from supplementation than veterans receiving lower dosages. Veterans with Vitamin D blood levels between 0 and 19 ng/ml exhibited the largest decrease in COVID-19 infection following supplementation. Black veterans received greater associated COVID-19 risk reductions with supplementation than White veterans. As a safe, widely available, and affordable treatment, Vitamin D may help to reduce the severity of the COVID-19 pandemic.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-11", "2022-12", "2023-01"], "last_mentioned": "2023-01-27T15:09:23Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41598-025-97652-6", "slug": "39be789eec6afe80", "title": "AI tutoring outperforms in-class active learning: an RCT introducing a novel research-based design in an authentic educational setting", "authors": ["Greg Kestin", "Kelly Miller", "Anna Klales", "Timothy Milbourne", "Gregorio Ponti"], "year": 2025, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/s41598-025-97652-6", "doi": "10.1038/s41598-025-97652-6", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2025-07", "2025-09", "2026-07"], "last_mentioned": "2026-07-15T01:47:28Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s41893-019-0451-7", "slug": "8579f4ae4e8b4329", "title": "Barriers and enablers for prescribed burns for wildfire management in California", "authors": ["Rebecca K. Miller", "Christopher B. Field", "Katharine J. Mach"], "year": 2020, "venue": "Nature Sustainability", "url": "https://doi.org/10.1038/s41893-019-0451-7", "doi": "10.1038/s41893-019-0451-7", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-09", "2020-10", "2023-06"], "last_mentioned": "2023-06-07T23:53:56Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s42256-021-00338-7", "slug": "89d73b09b7c59140", "title": "AI for radiographic COVID-19 detection selects shortcuts over signal", "authors": ["Alex J. DeGrave", "Joseph D. Janizek", "Su-In Lee"], "year": 2021, "venue": "Nature Machine Intelligence", "url": "https://doi.org/10.1038/s42256-021-00338-7", "doi": "10.1038/s42256-021-00338-7", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-11", "2023-06", "2025-09"], "last_mentioned": "2025-09-25T17:06:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s43016-022-00530-x", "slug": "01b19838d72aeed8", "title": "A hybrid inorganic–biological artificial photosynthesis system for energy-efficient food production", "authors": ["Elizabeth C. Hann", "Sean Overa", "Marcus Harland-Dunaway", "Andrés F. Narvaez", "Dang N. Le", "Martha L. Orozco-Cárdenas", "Feng Jiao", "Robert E. Jinkerson"], "year": 2022, "venue": "Nature Food", "url": "https://doi.org/10.1038/s43016-022-00530-x", "doi": "10.1038/s43016-022-00530-x", "topics": ["Biology & medicine"], "description": "Abstract Artificial photosynthesis systems are proposed as an efficient alternative route to capture CO 2 to produce additional food for growing global demand.", "description_method": "abstract-excerpt", "abstract": "Abstract Artificial photosynthesis systems are proposed as an efficient alternative route to capture CO 2 to produce additional food for growing global demand. Here a two-step CO 2 electrolyser system was developed to produce a highly concentrated acetate stream with a 57% carbon selectivity (CO 2 to acetate), allowing its direct use for the heterotrophic cultivation of yeast, mushroom-producing fungus and a photosynthetic green alga, in the dark without inputs from biological photosynthesis. An evaluation of nine crop plants found that carbon from exogenously supplied acetate incorporates into biomass through major metabolic pathways. Coupling this approach to existing photovoltaic systems could increase solar-to-food energy conversion efficiency by about fourfold over biological photosynthesis, reducing the solar footprint required. This technology allows for a reimagination of how food can be produced in controlled environments.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-06", "2023-11", "2024-10"], "last_mentioned": "2024-10-25T23:26:21Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s43247-022-00498-3", "slug": "cf42a83a57bad4e5", "title": "The Arctic has warmed nearly four times faster than the globe since 1979", "authors": ["Mika Rantanen", "Alexey Yu. Karpechko", "Antti Lipponen", "Kalle Nordling", "Otto Hyvärinen", "Kimmo Ruosteenoja", "Timo Vihma", "Ari Laaksonen"], "year": 2022, "venue": "Communications Earth &amp; Environment", "url": "https://doi.org/10.1038/s43247-022-00498-3", "doi": "10.1038/s43247-022-00498-3", "topics": ["Other research"], "description": "Abstract In recent decades, the warming in the Arctic has been much faster than in the rest of the world, a phenomenon known as Arctic amplification.", "description_method": "abstract-excerpt", "abstract": "Abstract In recent decades, the warming in the Arctic has been much faster than in the rest of the world, a phenomenon known as Arctic amplification. Numerous studies report that the Arctic is warming either twice, more than twice, or even three times as fast as the globe on average. Here we show, by using several observational datasets which cover the Arctic region, that during the last 43 years the Arctic has been warming nearly four times faster than the globe, which is a higher ratio than generally reported in literature. We compared the observed Arctic amplification ratio with the ratio simulated by state-of-the-art climate models, and found that the observed four-fold warming ratio over 1979–2021 is an extremely rare occasion in the climate model simulations. The observed and simulated amplification ratios are more consistent with each other if calculated over a longer period; however the comparison is obscured by observational uncertainties before 1979. Our results indicate that the recent four-fold Arctic warming ratio is either an extremely unlikely event, or the climate models systematically tend to underestimate the amplification.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-08", "2023-04", "2024-06"], "last_mentioned": "2024-06-09T18:11:07Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/s43587-021-00138-z", "slug": "9368aae4690b3ad1", "title": "Endophenotype-based in silico network medicine discovery combined with insurance record data mining identifies sildenafil as a candidate drug for Alzheimer’s disease", "authors": ["Jiansong Fang", "Pengyue Zhang", "Yadi Zhou", "Chien-Wei Chiang", "Juan Tan", "Yuan Hou", "Shaun Stauffer", "Lang Li", "Andrew A. Pieper", "Jeffrey Cummings", "Feixiong Cheng"], "year": 2021, "venue": "Nature Aging", "url": "https://doi.org/10.1038/s43587-021-00138-z", "doi": "10.1038/s43587-021-00138-z", "topics": ["Biology & medicine"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-12", "2023-07", "2023-09"], "last_mentioned": "2023-09-10T19:34:38Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/srep13945", "slug": "9242281c44db6d68", "title": "AGM2015: Antineutrino Global Map 2015", "authors": ["S.M. Usman", "G.R. Jocher", "S.T. Dye", "W.F. McDonough", "J.G. Learned"], "year": 2015, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/srep13945", "doi": "10.1038/srep13945", "topics": ["Physics"], "description": "AbstractEvery second greater than 1025 antineutrinos radiate to space from Earth, shining like a faint antineutrino star.", "description_method": "abstract-excerpt", "abstract": "AbstractEvery second greater than 1025 antineutrinos radiate to space from Earth, shining like a faint antineutrino star. Underground antineutrino detectors have revealed the rapidly decaying fission products inside nuclear reactors, verified the long-lived radioactivity inside our planet and informed sensitive experiments for probing fundamental physics. Mapping the anisotropic antineutrino flux and energy spectrum advance geoscience by defining the amount and distribution of radioactive power within Earth while critically evaluating competing compositional models of the planet. We present the Antineutrino Global Map 2015 (AGM2015), an experimentally informed model of Earth’s surface antineutrino flux over the 0 to 11 MeV energy spectrum, along with an assessment of systematic errors. The open source AGM2015 provides fundamental predictions for experiments, assists in strategic detector placement to determine neutrino mass hierarchy and aids in identifying undeclared nuclear reactors. We use cosmochemically and seismologically informed models of the radiogenic lithosphere/mantle combined with the estimated antineutrino flux, as measured by KamLAND and Borexino, to determine the Earth’s total antineutrino luminosity at \"Equation missing\". We find a dominant flux of geo-neutrinos, predict sub-equal crust and mantle contributions, with ~1% of the total flux from man-made nuclear reactors.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-06", "2022-01", "2026-03"], "last_mentioned": "2026-03-04T05:11:26Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/srep40700", "slug": "33002a9c6bc829f5", "title": "Neurotypical Peers are Less Willing to Interact with Those with Autism based on Thin Slice Judgments", "authors": ["Noah J. Sasson", "Daniel J. Faso", "Jack Nugent", "Sarah Lovell", "Daniel P. Kennedy", "Ruth B. Grossman"], "year": 2017, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/srep40700", "doi": "10.1038/srep40700", "topics": ["Psychology", "Neuroscience"], "description": "Abstract Individuals with autism spectrum disorder (ASD), including those who otherwise require less support, face severe difficulties in everyday social interactions.", "description_method": "abstract-excerpt", "abstract": "Abstract Individuals with autism spectrum disorder (ASD), including those who otherwise require less support, face severe difficulties in everyday social interactions. Research in this area has primarily focused on identifying the cognitive and neurological differences that contribute to these social impairments, but social interaction by definition involves more than one person and social difficulties may arise not just from people with ASD themselves, but also from the perceptions, judgments, and social decisions made by those around them. Here, across three studies, we find that first impressions of individuals with ASD made from thin slices of real-world social behavior by typically-developing observers are not only far less favorable across a range of trait judgments compared to controls, but also are associated with reduced intentions to pursue social interaction. These patterns are remarkably robust, occur within seconds, do not change with increased exposure, and persist across both child and adult age groups. However, these biases disappear when impressions are based on conversational content lacking audio-visual cues, suggesting that style, not substance, drives negative impressions of ASD. Collectively, these findings advocate for a broader perspective of social difficulties in ASD that considers both the individual’s impairments and the biases of potential social partners.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-02", "2024-05", "2025-10"], "last_mentioned": "2025-10-03T16:34:50Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1038/srep46173", "slug": "f8c9036580af8ee3", "title": "The presence of microplastics in commercial salts from different countries", "authors": ["Ali Karami", "Abolfazl Golieskardi", "Cheng Keong Choo", "Vincent Larat", "Tamara S. Galloway", "Babak Salamatinia"], "year": 2017, "venue": "Scientific Reports", "url": "https://doi.org/10.1038/srep46173", "doi": "10.1038/srep46173", "topics": ["Physics"], "description": "AbstractThe occurrence of microplastics (MPs) in saltwater bodies is relatively well studied, but nothing is known about their presence in most of the commercial salts that are widely consumed by humans across the globe.", "description_method": "abstract-excerpt", "abstract": "AbstractThe occurrence of microplastics (MPs) in saltwater bodies is relatively well studied, but nothing is known about their presence in most of the commercial salts that are widely consumed by humans across the globe. Here, we extracted MP-like particles larger than 149 μm from 17 salt brands originating from 8 different countries followed by the identification of their polymer composition using micro-Raman spectroscopy. Microplastics were absent in one brand while others contained between 1 to 10 MPs/Kg of salt. Out of the 72 extracted particles, 41.6% were plastic polymers, 23.6% were pigments, 5.50% were amorphous carbon, and 29.1% remained unidentified. The particle size (mean ± SD) was 515 ± 171 μm. The most common plastic polymers were polypropylene (40.0%) and polyethylene (33.3%). Fragments were the primary form of MPs (63.8%) followed by filaments (25.6%) and films (10.6%). According to our results, the low level of anthropogenic particles intake from the salts (maximum 37 particles per individual per annum) warrants negligible health impacts. However, to better understand the health risks associated with salt consumption, further development in extraction protocols are needed to isolate anthropogenic particles smaller than 149 μm.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2018-08", "2020-10", "2023-07"], "last_mentioned": "2023-07-06T00:40:12Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.1400005", "slug": "92d2feb67bbbe790", "title": "Systematic inequality and hierarchy in faculty hiring networks", "authors": ["Aaron Clauset", "Samuel Arbesman", "Daniel B. Larremore"], "year": 2015, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.1400005", "doi": "10.1126/sciadv.1400005", "topics": ["Economics", "Psychology"], "description": "An analysis of networks of graduate-to-faculty hires reveals systematic hiring biases and patterns.", "description_method": "abstract-excerpt", "abstract": "An analysis of networks of graduate-to-faculty hires reveals systematic hiring biases and patterns.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-02", "2022-06", "2025-09"], "last_mentioned": "2025-09-21T17:54:38Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.aav5916", "slug": "38f876ff0f043e6a", "title": "Kids these days: Why the youth of today seem lacking", "authors": ["John Protzko", "Jonathan W. Schooler"], "year": 2019, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.aav5916", "doi": "10.1126/sciadv.aav5916", "topics": ["Psychology"], "description": "We view kids these days unfavorably, especially on traits at which we excel, partly because we have a biased view of the past.", "description_method": "abstract-excerpt", "abstract": "We view kids these days unfavorably, especially on traits at which we excel, partly because we have a biased view of the past.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-07", "2022-09", "2024-02"], "last_mentioned": "2024-02-23T15:57:17Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.aay2631", "slug": "a44159e8128b6b34", "title": "AI Feynman: A physics-inspired method for symbolic regression", "authors": ["Silviu-Marian Udrescu", "Max Tegmark"], "year": 2020, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.aay2631", "doi": "10.1126/sciadv.aay2631", "topics": ["Physics", "Statistics & probability", "Algorithms"], "description": "Our physics-inspired algorithm for symbolic regression is able to discover complex physics equations from mere tables of numbers.", "description_method": "abstract-excerpt", "abstract": "Our physics-inspired algorithm for symbolic regression is able to discover complex physics equations from mere tables of numbers.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-03", "2024-05", "2024-09"], "last_mentioned": "2024-09-12T14:54:46Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.abd1705", "slug": "8bfbdb90a8f42646", "title": "Nonreplicable publications are cited more than replicable ones", "authors": ["Marta Serra-Garcia", "Uri Gneezy"], "year": 2021, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.abd1705", "doi": "10.1126/sciadv.abd1705", "topics": ["Science & research"], "description": "Published papers that fail to replicate are cited more than those that replicate, even after the failure is published.", "description_method": "abstract-excerpt", "abstract": "Published papers that fail to replicate are cited more than those that replicate, even after the failure is published.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-12", "2025-09", "2025-12"], "last_mentioned": "2025-12-01T02:53:34Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/sciadv.abl8213", "slug": "4be07e314fc9b32c", "title": "ChAdOx1 interacts with CAR and PF4 with implications for thrombosis with thrombocytopenia syndrome", "authors": ["Alexander T. Baker", "Ryan J. Boyd", "Daipayan Sarkar", "Alicia Teijeira-Crespo", "Chun Kit Chan", "Emily Bates", "Kasim Waraich", "John Vant", "Eric Wilson", "Chloe D. Truong", "Magdalena Lipka-Lloyd", "Petra Fromme", "Josh Vermaas", "Dewight Williams", "LeeAnn Machiesky", "Meike Heurich", "Bolni M. Nagalo", "Lynda Coughlan", "Scott Umlauf", "Po-Lin Chiu", "Pierre J. Rizkallah", "Taylor S. Cohen", "Alan L. Parker", "Abhishek Singharoy", "Mitesh J. Borad"], "year": 2021, "venue": "Science Advances", "url": "https://doi.org/10.1126/sciadv.abl8213", "doi": "10.1126/sciadv.abl8213", "topics": ["Biology & medicine"], "description": "We observe previously unknown interactions between clinically important adenovirus vector capsids, platelet factor 4, and CAR.", "description_method": "abstract-excerpt", "abstract": "We observe previously unknown interactions between clinically important adenovirus vector capsids, platelet factor 4, and CAR.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-12", "2022-01", "2022-04"], "last_mentioned": "2022-04-29T19:48:25Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.1121066", "slug": "1ac8cae8f9e0e069", "title": "Experimental Study of Inequality and Unpredictability in an Artificial Cultural Market", "authors": ["Matthew J. Salganik", "Peter Sheridan Dodds", "Duncan J. Watts"], "year": 2006, "venue": "Science", "url": "https://doi.org/10.1126/science.1121066", "doi": "10.1126/science.1121066", "topics": ["Economics"], "description": "Hit songs, books, and movies are many times more successful than average, suggesting that “the best” alternatives are qualitatively different from “the rest”; yet experts routinely fail to predict which products will succeed.", "description_method": "abstract-excerpt", "abstract": "Hit songs, books, and movies are many times more successful than average, suggesting that “the best” alternatives are qualitatively different from “the rest”; yet experts routinely fail to predict which products will succeed. We investigated this paradox experimentally, by creating an artificial “music market” in which 14,341 participants downloaded previously unknown songs either with or without knowledge of previous participants' choices. Increasing the strength of social influence increased both inequality and unpredictability of success. Success was also only partly determined by quality: The best songs rarely did poorly, and the worst rarely did well, but any other result was possible.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-09", "2022-10", "2023-04"], "last_mentioned": "2023-04-05T16:40:13Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.1245824", "slug": "092d976071e5fe9c", "title": "A Different Form of Color Vision in Mantis Shrimp", "authors": ["Hanne H. Thoen", "Martin J. How", "Tsyr-Huei Chiou", "Justin Marshall"], "year": 2014, "venue": "Science", "url": "https://doi.org/10.1126/science.1245824", "doi": "10.1126/science.1245824", "topics": ["Psychology", "Algorithms"], "description": "One of the most complex eyes in the animal kingdom can be found in species of stomatopod crustaceans (mantis shrimp), some of which have 12 different photoreceptor types, each sampling a narrow set of wavelengths ranging from deep ultraviolet to far red (300 to 720 nanometers) ( 1 – 3 ).", "description_method": "abstract-excerpt", "abstract": "One of the most complex eyes in the animal kingdom can be found in species of stomatopod crustaceans (mantis shrimp), some of which have 12 different photoreceptor types, each sampling a narrow set of wavelengths ranging from deep ultraviolet to far red (300 to 720 nanometers) ( 1 – 3 ). Functionally, this chromatic complexity has presented a mystery ( 3 – 5 ). Why use 12 color channels when three or four are sufficient for fine color discrimination? Behavioral wavelength discrimination tests (Δλ functions) in stomatopods revealed a surprisingly poor performance, ruling out color vision that makes use of the conventional color-opponent coding system ( 6 – 8 ). Instead, our experiments suggest that stomatopods use a previously unknown color vision system based on temporal signaling combined with scanning eye movements, enabling a type of color recognition rather than discrimination.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-11", "2023-02", "2025-07"], "last_mentioned": "2025-07-01T19:28:10Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.aad6253", "slug": "b58d007b00118282", "title": "Design and synthesis of a minimal bacterial genome", "authors": ["Clyde A. Hutchison", "Ray-Yuan Chuang", "Vladimir N. Noskov", "Nacyra Assad-Garcia", "Thomas J. Deerinck", "Mark H. Ellisman", "John Gill", "Krishna Kannan", "Bogumil J. Karas", "Li Ma", "James F. Pelletier", "Zhi-Qing Qi", "R. Alexander Richter", "Elizabeth A. Strychalski", "Lijie Sun", "Yo Suzuki", "Billyana Tsvetanova", "Kim S. Wise", "Hamilton O. Smith", "John I. Glass", "Chuck Merryman", "Daniel G. Gibson", "J. Craig Venter"], "year": 2016, "venue": "Science", "url": "https://doi.org/10.1126/science.aad6253", "doi": "10.1126/science.aad6253", "topics": ["Biology & medicine"], "description": "Designing and building a minimal genome A goal in biology is to understand the molecular and biological function of every gene in a cell.", "description_method": "abstract-excerpt", "abstract": "Designing and building a minimal genome A goal in biology is to understand the molecular and biological function of every gene in a cell. One way to approach this is to build a minimal genome that includes only the genes essential for life. In 2010, a 1079-kb genome based on the genome of Mycoplasma mycoides (JCV-syn1.0) was chemically synthesized and supported cell growth when transplanted into cytoplasm. Hutchison III et al. used a design, build, and test cycle to reduce this genome to 531 kb (473 genes). The resulting JCV-syn3.0 retains genes involved in key processes such as transcription and translation, but also contains 149 genes of unknown function. Science , this issue p. 10.1126/science.aad6253", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2019-08", "2022-01", "2023-08"], "last_mentioned": "2023-08-11T06:32:31Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.aam9744", "slug": "ac40e60f13134448", "title": "There’s plenty of room at the Top: What will drive computer performance after Moore’s law?", "authors": ["Charles E. Leiserson", "Neil C. Thompson", "Joel S. Emer", "Bradley C. Kuszmaul", "Butler W. Lampson", "Daniel Sanchez", "Tao B. Schardl"], "year": 2020, "venue": "Science", "url": "https://doi.org/10.1126/science.aam9744", "doi": "10.1126/science.aam9744", "topics": ["Algorithms"], "description": "From bottom to topThe doubling of the number of transistors on a chip every 2 years, a seemly inevitable trend that has been called Moore's law, has contributed immensely to improvements in computer performance.", "description_method": "abstract-excerpt", "abstract": "From bottom to topThe doubling of the number of transistors on a chip every 2 years, a seemly inevitable trend that has been called Moore's law, has contributed immensely to improvements in computer performance. However, silicon-based transistors cannot get much smaller than they are today, and other approaches should be explored to keep performance growing. Leisersonet al.review recent examples and argue that the most promising place to look is at the top of the computing stack, where improvements in software, algorithms, and hardware architecture can bring the much-needed boost.Science, this issue p.eaam9744", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-12", "2022-03", "2024-07"], "last_mentioned": "2024-07-04T02:40:53Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abn7950", "slug": "0f39b22de959f78d", "title": "Exceeding 1.5°C global warming could trigger multiple climate tipping points", "authors": ["David I. Armstrong McKay", "Arie Staal", "Jesse F. Abrams", "Ricarda Winkelmann", "Boris Sakschewski", "Sina Loriani", "Ingo Fetzer", "Sarah E. Cornell", "Johan Rockström", "Timothy M. Lenton"], "year": 2022, "venue": "Science", "url": "https://doi.org/10.1126/science.abn7950", "doi": "10.1126/science.abn7950", "topics": ["Other research"], "description": "Climate tipping points occur when change in a part of the climate system becomes self-perpetuating beyond a warming threshold, leading to substantial Earth system impacts.", "description_method": "abstract-excerpt", "abstract": "Climate tipping points occur when change in a part of the climate system becomes self-perpetuating beyond a warming threshold, leading to substantial Earth system impacts. Synthesizing paleoclimate, observational, and model-based studies, we provide a revised shortlist of global “core” tipping elements and regional “impact” tipping elements and their temperature thresholds. Current global warming of ~1.1°C above preindustrial temperatures already lies within the lower end of some tipping point uncertainty ranges. Several tipping points may be triggered in the Paris Agreement range of 1.5 to <2°C global warming, with many more likely at the 2 to 3°C of warming expected on current policy trajectories. This strengthens the evidence base for urgent action to mitigate climate change and to develop improved tipping point risk assessment, early warning capability, and adaptation strategies.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-09", "2022-10", "2024-08"], "last_mentioned": "2024-08-13T15:06:02Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1126/science.abp9364", "slug": "bb40d802b4a93ba4", "title": "How do social media feed algorithms affect attitudes and behavior in an election campaign?", "authors": ["Andrew M. Guess", "Neil Malhotra", "Jennifer Pan", "Pablo Barberá", "Hunt Allcott", "Taylor Brown", "Adriana Crespo-Tenorio", "Drew Dimmery", "Deen Freelon", "Matthew Gentzkow", "Sandra González-Bailón", "Edward Kennedy", "Young Mie Kim", "David Lazer", "Devra Moehler", "Brendan Nyhan", "Carlos Velasco Rivera", "Jaime Settle", "Daniel Robert Thomas", "Emily Thorson", "Rebekah Tromble", "Arjun Wilkins", "Magdalena Wojcieszak", "Beixian Xiong", "Chad Kiewiet de Jonge", "Annie Franco", "Winter Mason", "Natalie Jomini Stroud", "Joshua A. Tucker"], "year": 2023, "venue": "Science", "url": "https://doi.org/10.1126/science.abp9364", "doi": "10.1126/science.abp9364", "topics": ["Algorithms", "Psychology"], "description": "We investigated the effects of Facebook’s and Instagram’s feed algorithms during the 2020 US election.", "description_method": "abstract-excerpt", "abstract": "We investigated the effects of Facebook’s and Instagram’s feed algorithms during the 2020 US election. We assigned a sample of consenting users to reverse-chronologically-ordered feeds instead of the default algorithms. Moving users out of algorithmic feeds substantially decreased the time they spent on the platforms and their activity. The chronological feed also affected exposure to content: The amount of political and untrustworthy content they saw increased on both platforms, the amount of content classified as uncivil or containing slur words they saw decreased on Facebook, and the amount of content from moderate friends and sources with ideologically mixed audiences they saw increased on Facebook. Despite these substantial changes in users’ on-platform experience, the chronological feed did not significantly alter levels of issue polarization, affective polarization, political knowledge, or other key attitudes during the 3-month study period.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2024-09", "2024-12", "2025-01"], "last_mentioned": "2025-01-16T18:37:24Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1136/bmj.k5094", "slug": "504ffb7fb9b67f0e", "title": "Parachute use to prevent death and major trauma when jumping from aircraft: randomized controlled trial", "authors": ["Robert W Yeh", "Linda R Valsdottir", "Michael W Yeh", "Changyu Shen", "Daniel B Kramer", "Jordan B Strom", "Eric A Secemsky", "Joanne L Healy", "Robert M Domeier", "Dhruv S Kazi", "Brahmajee K Nallamothu"], "year": 2018, "venue": "BMJ", "url": "https://doi.org/10.1136/bmj.k5094", "doi": "10.1136/bmj.k5094", "topics": ["Biology & medicine"], "description": "Abstract Objective To determine if using a parachute prevents death or major traumatic injury when jumping from an aircraft.", "description_method": "abstract-excerpt", "abstract": "Abstract Objective To determine if using a parachute prevents death or major traumatic injury when jumping from an aircraft. Design Randomized controlled trial. Setting Private or commercial aircraft between September 2017 and August 2018. Participants 92 aircraft passengers aged 18 and over were screened for participation. 23 agreed to be enrolled and were randomized. Intervention Jumping from an aircraft (airplane or helicopter) with a parachute versus an empty backpack (unblinded). Main outcome measures Composite of death or major traumatic injury (defined by an Injury Severity Score over 15) upon impact with the ground measured immediately after landing. Results Parachute use did not significantly reduce death or major injury (0% for parachute v 0% for control; P>0.9). This finding was consistent across multiple subgroups. Compared with individuals screened but not enrolled, participants included in the study were on aircraft at significantly lower altitude (mean of 0.6 m for participants v mean of 9146 m for non-participants; P<0.001) and lower velocity (mean of 0 km/h v mean of 800 km/h; P<0.001). Conclusions Parachute use did not reduce death or major traumatic injury when jumping from aircraft in the first randomized evaluation of this intervention. However, the trial was only able to enroll participants on small stationary aircraft on the ground, suggesting cautious extrapolation to high altitude jumps. When beliefs regarding the effectiveness of an intervention exist in the community, randomized trials might selectively enroll individuals with a lower perceived likelihood of benefit, thus diminishing the applicability of the results to clinical practice.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-04", "2025-09", "2025-10"], "last_mentioned": "2025-10-10T21:04:50Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/1028976.1028982", "slug": "32df6b65196a25ed", "title": "A unified theory of garbage collection", "authors": ["David F. Bacon", "Perry Cheng", "V. T. Rajan"], "year": 2004, "venue": "Proceedings of the 19th annual ACM SIGPLAN conference on Object-oriented programming, systems, languages, and applications", "url": "https://doi.org/10.1145/1028976.1028982", "doi": "10.1145/1028976.1028982", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-01", "2023-02", "2023-12"], "last_mentioned": "2023-12-30T21:13:22Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/2408776.2408794", "slug": "35e809461ab5c481", "title": "The tail at scale", "authors": ["Jeffrey Dean", "Luiz André Barroso"], "year": 2013, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/2408776.2408794", "doi": "10.1145/2408776.2408794", "topics": ["Other research"], "description": "Software techniques that tolerate latency variability are vital to building responsive large-scale Web services.", "description_method": "abstract-excerpt", "abstract": "Software techniques that tolerate latency variability are vital to building responsive large-scale Web services.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-03", "2020-04", "2020-09"], "last_mentioned": "2020-09-02T10:01:08Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/2534973", "slug": "0d051a17be193232", "title": "An Empirical Investigation into Programming Language Syntax", "authors": ["Andreas Stefik", "Susanna Siebert"], "year": 2013, "venue": "ACM Transactions on Computing Education", "url": "https://doi.org/10.1145/2534973", "doi": "10.1145/2534973", "topics": ["Programming languages"], "description": "Recent studies in the literature have shown that syntax remains a significant barrier to novice computer science students in the field.", "description_method": "abstract-excerpt", "abstract": "Recent studies in the literature have shown that syntax remains a significant barrier to novice computer science students in the field. While this syntax barrier is known to exist, whether and how it varies across programming languages has not been carefully investigated. For this article, we conducted four empirical studies on programming language syntax as part of a larger analysis into the, so called, programming language wars. We first present two surveys conducted with students on the intuitiveness of syntax, which we used to garner formative clues on what words and symbols might be easy for novices to understand. We followed up with two studies on the accuracy rates of novices using a total of six programming languages: Ruby, Java, Perl, Python, Randomo, and Quorum. Randomo was designed by randomly choosing some keywords from the ASCII table (a metaphorical placebo). To our surprise, we found that languages using a more traditional C-style syntax (both Perl and Java) did not afford accuracy rates significantly higher than a language with randomly generated keywords, but that languages which deviate (Quorum, Python, and Ruby) did. These results, including the specifics of syntax that are particularly problematic for novices, may help teachers of introductory programming courses in choosing appropriate first languages and in helping students to overcome the challenges they face with syntax.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-08", "2022-11", "2025-10"], "last_mentioned": "2025-10-14T12:00:48Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/2789168.2790109", "slug": "9043323908a9f938", "title": "Keystroke Recognition Using WiFi Signals", "authors": ["Kamran Ali", "Alex X. Liu", "Wei Wang", "Muhammad Shahzad"], "year": 2015, "venue": "Proceedings of the 21st Annual International Conference on Mobile Computing and Networking", "url": "https://doi.org/10.1145/2789168.2790109", "doi": "10.1145/2789168.2790109", "topics": ["Psychology"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-01", "2025-06", "2025-09"], "last_mentioned": "2025-09-04T18:24:51Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3158140", "slug": "2738e4baaa0dfc08", "title": "Collapsing towers of interpreters", "authors": ["Nada Amin", "Tiark Rompf"], "year": 2017, "venue": "Proceedings of the ACM on Programming Languages", "url": "https://doi.org/10.1145/3158140", "doi": "10.1145/3158140", "topics": ["Programming languages", "Mathematics", "Language models"], "description": "Given a tower of interpreters, i.e., a sequence of multiple interpreters interpreting one another as input programs, we aim to collapse this tower into a compiler that removes all interpretive overhead and runs in a single pass.", "description_method": "abstract-excerpt", "abstract": "Given a tower of interpreters, i.e., a sequence of multiple interpreters interpreting one another as input programs, we aim to collapse this tower into a compiler that removes all interpretive overhead and runs in a single pass. In the real world, a use case might be Python code executed by an x86 runtime, on a CPU emulated in a JavaScript VM, running on an ARM CPU. Collapsing such a tower can not only exponentially improve runtime performance, but also enable the use of base-language tools for interpreted programs, e.g., for analysis and verification. In this paper, we lay the foundations in an idealized but realistic setting. We present a multi-level lambda calculus that features staging constructs and stage polymorphism: based on runtime parameters, an evaluator either executes source code (thereby acting as an interpreter) or generates code (thereby acting as a compiler). We identify stage polymorphism, a programming model from the domain of high-performance program generators, as the key mechanism to make such interpreters compose in a collapsible way. We present Pink, a meta-circular Lisp-like evaluator on top of this calculus, and demonstrate that we can collapse arbitrarily many levels of self-interpretation, including levels with semantic modifications. We discuss several examples: compiling regular expressions through an interpreter to base code, building program transformers from modi ed interpreters, and others. We develop these ideas further to include reflection and reification, culminating in Purple, a reflective language inspired by Brown, Blond, and Black, which realizes a conceptually infinite tower, where every aspect of the semantics can change dynamically. Addressing an open challenge, we show how user programs can be compiled and recompiled under user-modified semantics.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-08", "2025-08", "2025-12"], "last_mentioned": "2025-12-14T05:20:16Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/3158154", "slug": "9107604d452b487a", "title": "RustBelt: securing the foundations of the Rust programming language", "authors": ["Ralf Jung", "Jacques-Henri Jourdan", "Robbert Krebbers", "Derek Dreyer"], "year": 2017, "venue": "Proceedings of the ACM on Programming Languages", "url": "https://doi.org/10.1145/3158154", "doi": "10.1145/3158154", "topics": ["Programming languages"], "description": "Rust is a new systems programming language that promises to overcome the seemingly fundamental tradeoff between high-level safety guarantees and low-level control over resource management.", "description_method": "abstract-excerpt", "abstract": "Rust is a new systems programming language that promises to overcome the seemingly fundamental tradeoff between high-level safety guarantees and low-level control over resource management. Unfortunately, none of Rust's safety claims have been formally proven, and there is good reason to question whether they actually hold. Specifically, Rust employs a strong, ownership-based type system, but then extends the expressive power of this core type system through libraries that internally use unsafe features. In this paper, we give the first formal (and machine-checked) safety proof for a language representing a realistic subset of Rust. Our proof is extensible in the sense that, for each new Rust library that uses unsafe features, we can say what verification condition it must satisfy in order for it to be deemed a safe extension to the language. We have carried out this verification for some of the most important libraries that are used throughout the Rust ecosystem.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-09", "2023-09", "2024-12"], "last_mentioned": "2024-12-21T14:01:30Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/319838.319870", "slug": "1ff5f42ac0afe48d", "title": "Connection Machine Lisp", "authors": ["Guy L. Steele", "W. Daniel Hillis"], "year": 1986, "venue": "Proceedings of the 1986 ACM conference on LISP and functional programming  - LFP '86", "url": "https://doi.org/10.1145/319838.319870", "doi": "10.1145/319838.319870", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-03", "2022-05", "2024-06"], "last_mentioned": "2024-06-19T00:43:07Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/360248.360253", "slug": "bb633a14dddb7df5", "title": "Ethernet", "authors": ["Robert M. Metcalfe", "David R. Boggs"], "year": 1976, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/360248.360253", "doi": "10.1145/360248.360253", "topics": ["Networks", "Distributed systems", "Statistics & probability"], "description": "Ethernet is a branching broadcast communication system for carrying digital data packets among locally distributed computing stations.", "description_method": "abstract-excerpt", "abstract": "Ethernet is a branching broadcast communication system for carrying digital data packets among locally distributed computing stations. The packet transport mechanism provided by Ethernet has been used to build systems which can be viewed as either local computer networks or loosely coupled multiprocessors. An Ethernet's shared communication facility, its Ether, is a passive broadcast medium with no central control. Coordination of access to the Ether for packet broadcasts is distributed among the contending transmitting stations using controlled statistical arbitration. Switching of packets to their destinations on the Ether is distributed among the receiving stations using packet address recognition. Design principles and implementation are described based on experience with an operating Ethernet of 100 nodes along a kilometer of coaxial cable. A model for estimating performance under heavy loads and a packet protocol for error controlled communication are included for completeness.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-10", "2023-03", "2024-04"], "last_mentioned": "2024-04-26T00:10:10Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/363744.363749", "slug": "0a1e919983f3e79a", "title": "Correspondence between ALGOL 60 and Church's Lambda-notation", "authors": ["P. J. Landin"], "year": 1965, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/363744.363749", "doi": "10.1145/363744.363749", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-10", "2023-07", "2024-10"], "last_mentioned": "2024-10-16T23:52:57Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/367487.367501", "slug": "5755187096ecc1f6", "title": "A method for overlapping and erasure of lists", "authors": ["George E. Collins"], "year": 1960, "venue": "Communications of the ACM", "url": "https://doi.org/10.1145/367487.367501", "doi": "10.1145/367487.367501", "topics": ["Other research"], "description": "An important property of the Newell Shaw-Simon scheme for computer storage of lists is that data having multiple occurrences need not be stored at more than one place in the computer.", "description_method": "abstract-excerpt", "abstract": "An important property of the Newell Shaw-Simon scheme for computer storage of lists is that data having multiple occurrences need not be stored at more than one place in the computer. That is, lists may be “overlapped.” Unfortunately, overlapping poses a problem for subsequent erasure. Given a list that is no longer needed, it is desired to erase just those parts that do not overlap other lists. In LISP, McCarthy employs an elegant but inefficient solution to the problem. The present paper describes a general method which enables efficient erasure. The method employs interspersed reference counts to describe the extent of the overlapping.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2020-06", "2022-11", "2025-09"], "last_mentioned": "2025-09-13T15:53:35Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "doi:10.1145/96709.96748", "slug": "ae855bdaa238b1bd", "title": "Deciding ML typability is complete for deterministic exponential time", "authors": ["Harry G. Mairson"], "year": 1990, "venue": "Proceedings of the 17th ACM SIGPLAN-SIGACT symposium on Principles of programming languages  - POPL '90", "url": "https://doi.org/10.1145/96709.96748", "doi": "10.1145/96709.96748", "topics": ["Other research"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2021-07", "2024-06", "2025-07"], "last_mentioned": "2025-07-07T11:52:07Z", "metadata_source": "crossref", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "url:https://aclanthology.org/N19-1423/", "slug": "8e6ad1c3f0f2b7ce", "title": "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding", "authors": ["Jacob Devlin", "Ming-Wei Chang", "Kenton Lee", "Kristina Toutanova"], "year": 2019, "venue": "Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers)", "url": "https://aclanthology.org/N19-1423/", "doi": "10.18653/v1/N19-1423", "topics": ["Language models"], "description": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, Kristina Toutanova.", "description_method": "abstract-excerpt", "abstract": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, Kristina Toutanova. Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers). 2019.", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2023-02", "2023-11", "2026-01"], "last_mentioned": "2026-01-06T20:17:12Z", "metadata_source": "publisher-citation-tags", "review_status": "unreviewed", "topic_method": "rules-v1"}
{"id": "url:https://papers.nips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html", "slug": "ad5f105ba492c8d3", "title": "ImageNet Classification with Deep Convolutional Neural Networks", "authors": ["Krizhevsky, Alex", "Sutskever, Ilya", "Hinton, Geoffrey E."], "year": 2012, "venue": "Advances in Neural Information Processing Systems", "url": "https://papers.nips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html", "doi": null, "topics": ["Machine learning", "Computer vision"], "description": "", "description_method": null, "abstract": "", "score": 0.0, "mention_count": 3, "user_count": 3, "months": ["2022-04", "2022-08", "2026-07"], "last_mentioned": "2026-07-07T19:03:23Z", "metadata_source": "publisher-citation-tags", "review_status": "unreviewed", "topic_method": "rules-v1"}