We introduce the first model-stealing attack that extracts precise, nontrivial information from black-box production language models like OpenAI's ChatGPT or Google's PaLM-2. Specifically, our attack recovers the embedding projection layer (up to symmetries) of a transformer model, given typical API access. For under \$20 USD, our attack extracts the entire projection matrix of OpenAI's Ada and Babbage language models. We thereby confirm, for the first time, that these black-box models have a...| arXiv.org
The number of disclosed vulnerabilities has been steadily increasing over the years. At the same time, organizations face significant challenges patching their systems, leading to a need to prioritize vulnerability remediation in order to reduce the risk of attacks. Unfortunately, existing vulnerability scoring systems are either vendor-specific, proprietary, or are only commercially available. Moreover, these and other prioritization strategies based on vulnerability severity are poor predic...| arXiv.org
The recent advances in large language models (LLMs) attracted significant public and policymaker interest in its adoption patterns. In this paper, we systematically analyze LLM-assisted writing across four domains-consumer complaints, corporate communications, job postings, and international organization press releases-from January 2022 to September 2024. Our dataset includes 687,241 consumer complaints, 537,413 corporate press releases, 304.3 million job postings, and 15,919 United Nations (...| arXiv.org
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 ...| arXiv.org
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 (TR...| arXiv.org
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 intui...| arXiv.org
GPT-3 can perform numerous tasks when provided a natural language prompt that contains a few training examples. We show that this type of few-shot learning can be unstable: the choice of prompt format, training examples, and even the order of the training examples can cause accuracy to vary from near chance to near state-of-the-art. We demonstrate that this instability arises from the bias of language models towards predicting certain answers, e.g., those that are placed near the end of the p...| arXiv.org
DeepSeek-V3 and DeepSeek-R1 are leading open-source Large Language Models (LLMs) for general-purpose tasks and reasoning, achieving performance comparable to state-of-the-art closed-source models from companies like OpenAI and Anthropic -- while requiring only a fraction of their training costs. Understanding the key innovative techniques behind DeepSeek's success is crucial for advancing LLM research. In this paper, we review the core techniques driving the remarkable effectiveness and effic...| arXiv.org
We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrates remarkable reasoning capabilities. Through RL, DeepSeek-R1-Zero naturally emerges with numerous powerful and intriguing reasoning behaviors. However, it encounters challenges such as poor readability, and language mixing. To address these issues and further en...| arXiv.org
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...| arXiv.org
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 DeepSeekMo...| arXiv.org
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 per...| arXiv.org
In the era of large language models, Mixture-of-Experts (MoE) is a promising architecture for managing computational costs when scaling up model parameters. However, conventional MoE architectures like GShard, which activate the top-$K$ out of $N$ experts, face challenges in ensuring expert specialization, i.e. each expert acquires non-overlapping and focused knowledge. In response, we propose the DeepSeekMoE architecture towards ultimate expert specialization. It involves two principal strat...| arXiv.org
Recent advances in Reinforcement Learning with Verifiable Rewards (RLVR) have empowered large language models (LLMs) to tackle challenging reasoning tasks such as mathematics and programming. RLVR leverages verifiable outcome rewards to guide policy optimization, enabling LLMs to progressively improve output quality in a grounded and reliable manner. Despite its promise, the RLVR paradigm poses significant challenges, as existing methods often suffer from sparse reward signals and unstable po...| arXiv.org
Near-Earth asteroid 2024 YR4 was discovered on 2024-12-27 and its probability of Earth impact in December 2032 peaked at about 3% on 2025-02-18. Additional observations ruled out Earth impact by 2025-02-23. However, the probability of lunar impact in December 2032 then rose, reaching about 4% by the end of the apparition in May 2025. James Webb Space Telescope (JWST) observations on 2025-03-26 estimated the asteroid's diameter at 60 +/- 7 m. Studies of 2024 YR4's potential lunar impact effect...| arXiv.org
Interoperability is increasingly recognised as a foundational principle for fostering innovation, competition, and user autonomy in the evolving digital ecosystem. Existing research on interoperability predominantly focuses either on technological interoperability itself or on the legal regulations concerning interoperability, with insufficient exploration of their interdisciplinary intersection. This paper compares the technological interoperability in Web 3.0 with the theoretical framework ...| arXiv.org
Large language models (LLMs) have attracted significant attention due to their impressive general capabilities across diverse downstream tasks. However, without domain-specific optimization, they often underperform on specialized knowledge benchmarks and even produce hallucination. Recent studies show that strategically infusing domain knowledge during pretraining can substantially improve downstream performance. A critical challenge lies in balancing this infusion trade-off: injecting too li...| arXiv.org
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. Val...| arXiv.org
We introduce Florence-2, a novel vision foundation model with a unified, prompt-based representation for a variety of computer vision and vision-language tasks. While existing large vision models excel in transfer learning, they struggle to perform a diversity of tasks with simple instructions, a capability that implies handling the complexity of various spatial hierarchy and semantic granularity. Florence-2 was designed to take text-prompt as task instructions and generate desirable results ...| arXiv.org
We introduce Delayed Streams Modeling (DSM), a flexible formulation for streaming, multimodal sequence-to-sequence learning. Sequence-to-sequence generation is often cast in an offline manner, where the model consumes the complete input sequence before generating the first output timestep. Alternatively, streaming sequence-to-sequence rely on learning a policy for choosing when to advance on the input stream, or write to the output stream. DSM instead models already time-aligned streams with ...| arXiv.org
We introduce Hibiki, a decoder-only model for simultaneous speech translation. Hibiki leverages a multistream language model to synchronously process source and target speech, and jointly produces text and audio tokens to perform speech-to-text and speech-to-speech translation. We furthermore address the fundamental challenge of simultaneous interpretation, which unlike its consecutive counterpart, where one waits for the end of the source utterance to start translating, adapts its flow to ac...| arXiv.org
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 inform...| arXiv.org
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 sim...| arXiv.org
We propose a factorization-free method for orthogonal projection onto the positive semidefinite (PSD) cone, leveraging composite polynomial filtering. Inspired by recent advances in homomorphic encryption, our approach approximates the PSD cone projection operator using a carefully optimized composite polynomial evaluated exclusively via matrix-matrix multiplications. This approach enables efficient GPU implementations with low-precision arithmetic, significantly outperforming the classical P...| arXiv.org
Despite the widespread adoption of Large language models (LLMs), their remarkable capabilities remain limited to a few high-resource languages. Additionally, many low-resource languages (\eg African languages) are often evaluated only on basic text classification tasks due to the lack of appropriate or comprehensive benchmarks outside of high-resource languages. In this paper, we introduce IrokoBench -- a human-translated benchmark dataset for 17 typologically-diverse low-resource African lan...| arXiv.org
Frontier large language models (LLMs) are developed by researchers and practitioners with skewed cultural backgrounds and on datasets with skewed sources. However, LLMs' (lack of) multicultural knowledge cannot be effectively assessed with current methods for developing benchmarks. Existing multicultural evaluations primarily rely on expensive and restricted human annotations or potentially outdated internet resources. Thus, they struggle to capture the intricacy, dynamics, and diversity of c...| arXiv.org
Sycophancy is an undesirable behavior where models tailor their responses to follow a human user's view even when that view is not objectively correct (e.g., adapting liberal views once a user reveals that they are liberal). In this paper, we study the prevalence of sycophancy in language models and propose a simple synthetic-data intervention to reduce this behavior. First, on a set of three sycophancy tasks (Perez et al., 2022) where models are asked for an opinion on statements with no cor...| arXiv.org
As language models (LMs) scale, they develop many novel behaviors, good and bad, exacerbating the need to evaluate how they behave. Prior work creates evaluations with crowdwork (which is time-consuming and expensive) or existing data sources (which are not always available). Here, we automatically generate evaluations with LMs. We explore approaches with varying amounts of human effort, from instructing LMs to write yes/no questions to making complex Winogender schemas with multiple stages o...| arXiv.org
Given the broad capabilities of large language models, it should be possible to work towards a general-purpose, text-based assistant that is aligned with human values, meaning that it is helpful, honest, and harmless. As an initial foray in this direction we study simple baseline techniques and evaluations, such as prompting. We find that the benefits from modest interventions increase with model size, generalize to a variety of alignment evaluations, and do not compromise the performance of ...| arXiv.org
This is an extended version of my 2018 Heinemann prize lecture describing the work for which I got the prize. The citation is very broad so this describes virtually all my work prior to 1995 and some afterwards. It discusses work in non-relativistic quantum mechanics, constructive quantum field theory and statistical mechanics.| arXiv.org
1 Introduction| arxiv.org
We present LatentCSI, a novel method for generating images of the physical environment from WiFi CSI measurements that leverages a pretrained latent diffusion model (LDM). Unlike prior approaches that rely on complex and computationally intensive techniques such as GANs, our method employs a lightweight neural network to map CSI amplitudes directly into the latent space of an LDM. We then apply the LDM's denoising diffusion model to the latent representation with text-based guidance before de...| arXiv.org
This report examines what I see as the core argument for concern about existential risk from misaligned artificial intelligence. I proceed in two stages. First, I lay out a backdrop picture that informs such concern. On this picture, intelligent agency is an extremely powerful force, and creating agents much more intelligent than us is playing with fire -- especially given that if their objectives are problematic, such agents would plausibly have instrumental incentives to seek power over hum...| arXiv.org
Reward hacking -- where RL agents exploit gaps in misspecified reward functions -- has been widely observed, but not yet systematically studied. To understand how reward hacking arises, we construct four RL environments with misspecified rewards. We investigate reward hacking as a function of agent capabilities: model capacity, action space resolution, observation space noise, and training time. More capable agents often exploit reward misspecifications, achieving higher proxy reward and lowe...| arXiv.org
We study goal misgeneralization, a type of out-of-distribution generalization failure in reinforcement learning (RL). Goal misgeneralization failures occur when an RL agent retains its capabilities out-of-distribution yet pursues the wrong goal. For instance, an agent might continue to competently avoid obstacles, but navigate to the wrong place. In contrast, previous works have typically focused on capability generalization failures, where an agent fails to do anything sensible at test time....| arXiv.org
Yes, it can. Catalogs produced by networks of Gravitational-wave interferometers are subject to complicated selection effects, and the gold-standard remains direct measurements of the detection probability through large injection campaigns. I leverage public data products from the LIGO-Virgo-KAGRA Collaborations' 3rd and 4th observing runs to show that there are non-trivial temporal variations within the detection probability that are well-described by a weekly cycle. There are clear differen...| arXiv.org
Recent text-to-image diffusion models such as MidJourney and Stable Diffusion threaten to displace many in the professional artist community. In particular, models can learn to mimic the artistic style of specific artists after "fine-tuning" on samples of their art. In this paper, we describe the design, implementation and evaluation of Glaze, a tool that enables artists to apply "style cloaks" to their art before sharing online. These cloaks apply barely perceptible perturbations to images, ...| arXiv.org
Scientists are increasingly overwhelmed by the volume of articles being published. Total articles indexed in Scopus and Web of Science have grown exponentially in recent years; in 2022 the article total was approximately ~47% higher than in 2016, which has outpaced the limited growth - if any - in the number of practising scientists. Thus, publication workload per scientist (writing, reviewing, editing) has increased dramatically. We define this problem as the strain on scientific publishing....| arXiv.org
Conformal cyclic cosmology (CCC) posits the existence of an aeon preceding our Big Bang 'B', whose conformal infinity 'I' is identified, conformally, with 'B', now regarded as a spacelike 3-surface. Black-hole encounters, within bound galactic clusters in that previous aeon, would have the observable effect, in our CMB sky, of families of concentric circles over which the temperature variance is anomalously low, the centre of each such family representing the point of 'I' at which the cluster...| arXiv.org
Training large language models (LLMs) for complex reasoning via Reinforcement Learning with Verifiable Rewards (RLVR) is effective but limited by reliance on costly, domain-specific supervision. We explore Reinforcement Learning from Internal Feedback (RLIF), a framework that enables LLMs to learn from intrinsic signals without external rewards or labeled data. We propose Intuitor, an RLIF method that uses a model's own confidence, termed self-certainty, as its sole reward signal. Intuitor re...| arXiv.org
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,...| arXiv.org
Widely used alignment techniques, such as reinforcement learning from human feedback (RLHF), rely on the ability of humans to supervise model behavior - for example, to evaluate whether a model faithfully followed instructions or generated safe outputs. However, future superhuman models will behave in complex ways too difficult for humans to reliably evaluate; humans will only be able to weakly supervise superhuman models. We study an analogy to this problem: can weak model supervision elicit...| arXiv.org
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 ini...| arXiv.org
arXiv is a free distribution service and an open-access archive for nearly 2.4 million| arxiv.org
Current approaches to building general-purpose AI systems tend to produce systems with both beneficial and harmful capabilities. Further progress in AI development could lead to capabilities that pose extreme risks, such as offensive cyber capabilities or strong manipulation skills. We explain why model evaluation is critical for addressing extreme risks. Developers must be able to identify dangerous capabilities (through "dangerous capability evaluations") and the propensity of models to app...| arXiv.org
1 Introduction| arxiv.org
Numerous powerful large language models (LLMs) are now available for use as writing support tools, idea generators, and beyond. Although these LLMs are marketed as helpful creative assistants, several works have shown that using an LLM as a creative partner results in a narrower set of creative outputs. However, these studies only consider the effects of interacting with a single LLM, begging the question of whether such narrowed creativity stems from using a particular LLM -- which arguably ...| arXiv.org
How diverse are the outputs of large language models when diversity is desired? We examine the diversity of responses of various models to questions with multiple possible answers, comparing them with human responses. Our findings suggest that models' outputs are highly concentrated, reflecting a narrow, mainstream 'worldview', in comparison to humans, whose responses exhibit a much longer-tail. We examine three ways to increase models' output diversity: 1) increasing generation randomness vi...| arXiv.org
1 Introduction| arxiv.org
1 Introduction| arxiv.org
The increasing size of large neural network models, specifically language models and foundational image models, poses deployment challenges, prompting efforts to reduce memory requirements and enhance computational efficiency. These efforts are critical to ensure practical deployment and effective utilization of these models across various applications. In this work, a novel type of neural network layers and models is developed that uses only single-bit parameters. In this novel type of model...| arXiv.org
Neural embedding models have become a fundamental component of modern information retrieval (IR) pipelines. These models produce a single embedding $x \in \mathbb{R}^d$ per data-point, allowing for fast retrieval via highly optimized maximum inner product search (MIPS) algorithms. Recently, beginning with the landmark ColBERT paper, multi-vector models, which produce a set of embedding per data point, have achieved markedly superior performance for IR tasks. Unfortunately, using these models ...| arXiv.org
Recent progresses of electronics, essentially due to its miniaturization, are opening new fields that were just dreamed of, notably in astronomy. At start in paragraph 3, we introduce the time variation of images expressing the dual nature of the optical signal (ZO) and we expose several useful applications where the optical signal variations are not faster than CCD. However we prefered to initiate the article with a deeper question posed inadvertently in paragraph 2: what causes the rapid, w...| arXiv.org
People often strive for deep engagement in activities which is usually associated with feelings of flow: a state of full task absorption accompanied by a sense of control and fulfillment. The intrinsic factors driving such engagement and facilitating subjective feelings of flow remain unclear. Building on computational theories of intrinsic motivation, this study examines how learning progress predicts engagement and directs cognitive control. Results showed that task engagement, indicated by...| arXiv.org
The use of NLP in the realm of financial technology is broad and complex, with applications ranging from sentiment analysis and named entity recognition to question answering. Large Language Models (LLMs) have been shown to be effective on a variety of tasks; however, no LLM specialized for the financial domain has been reported in literature. In this work, we present BloombergGPT, a 50 billion parameter language model that is trained on a wide range of financial data. We construct a 363 bill...| arXiv.org
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 earl...| arXiv.org
We introduce Paper2Agent, an automated framework that converts research papers into AI agents. Paper2Agent transforms research output from passive artifacts into active systems that can accelerate downstream use, adoption, and discovery. Conventional research papers require readers to invest substantial effort to understand and adapt a paper's code, data, and methods to their own work, creating barriers to dissemination and reuse. Paper2Agent addresses this challenge by automatically converti...| arXiv.org
JSON is a popular standard for data interchange on the Internet. Ingesting JSON documents can be a performance bottleneck. A popular parsing strategy consists in converting the input text into a tree-based data structure -- sometimes called a Document Object Model or DOM. We designed and implemented a novel JSON parsing interface -- called On-Demand -- that appears to the programmer like a conventional DOM-based approach. However, the underlying implementation is a pointer iterating through t...| arXiv.org
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 qua...| arXiv.org
We examine whether substantial AI automation could accelerate global economic growth by about an order of magnitude, akin to the economic growth effects of the Industrial Revolution. We identify three primary drivers for such growth: 1) the scalability of an AI "labor force" restoring a regime of increasing returns to scale, 2) the rapid expansion of an AI labor force, and 3) a massive increase in output from rapid automation occurring over a brief period of time. Against this backdrop, we ev...| arXiv.org
Thanks to rapid progress in artificial intelligence, we have entered an era when technology and philosophy intersect in interesting ways. Sitting squarely at the centre of this intersection are large language models (LLMs). The more adept LLMs become at mimicking human language, the more vulnerable we become to anthropomorphism, to seeing the systems in which they are embedded as more human-like than they really are. This trend is amplified by the natural tendency to use philosophically loade...| arXiv.org
With rising life expectancies around the world and an older scientific workforce than ever before, what does aging mean for individual scientists, and what do aging scientists mean for scientific progress as a whole? Here we examine how scientists and scholars age in terms of how their ideas and contributions relate to the evolving frontier of knowledge and how demographically aging fields relate to field-level advance. At the individual level, we examine how research experiences and choices ...| arXiv.org
This study presents estimates of the global expenditure on article processing charges (APCs) paid to six publishers for open access between 2019 and 2023. APCs are fees charged for publishing in some fully open access journals (gold) and in subscription journals to make individual articles open access (hybrid). There is currently no way to systematically track institutional, national or global expenses for open access publishing due to a lack of transparency in APC prices, what articles they ...| arXiv.org
Chain-of-thought (CoT) offers a potential boon for AI safety as it allows monitoring a model's CoT to try to understand its intentions and reasoning processes. However, the effectiveness of such monitoring hinges on CoTs faithfully representing models' actual reasoning processes. We evaluate CoT faithfulness of state-of-the-art reasoning models across 6 reasoning hints presented in the prompts and find: (1) for most settings and models tested, CoTs reveal their usage of hints in at least 1% o...| arXiv.org
Information on the web, such as scientific publications and Wikipedia, often surpasses users' reading level. To help address this, we used a self-refinement approach to develop a LLM capability for minimally lossy text simplification. To validate our approach, we conducted a randomized study involving 4563 participants and 31 texts spanning 6 broad subject areas: PubMed (biomedical scientific articles), biology, law, finance, literature/philosophy, and aerospace/computer science. Participants...| arXiv.org
We investigate practical and scalable algorithms for training large language models (LLMs) with user-level differential privacy (DP) in order to provably safeguard all the examples contributed by each user. We study two variants of DP-SGD with: (1) example-level sampling (ELS) and per-example gradient clipping, and (2) user-level sampling (ULS) and per-user gradient clipping. We derive a novel user-level DP accountant that allows us to compute provably tight privacy guarantees for ELS. Using ...| arXiv.org
We introduce Dataset Grouper, a library to create large-scale group-structured (e.g., federated) datasets, enabling federated learning simulation at the scale of foundation models. This library facilitates the creation of group-structured versions of existing datasets based on user-specified partitions and directly leads to a variety of useful heterogeneous datasets that can be plugged into existing software frameworks. Dataset Grouper offers three key advantages. First, it scales to settings...| arXiv.org
We demonstrate that it is possible to train large recurrent language models with user-level differential privacy guarantees with only a negligible cost in predictive accuracy. Our work builds on recent advances in the training of deep networks on user-partitioned data and privacy accounting for stochastic gradient descent. In particular, we add user-level privacy protection to the federated averaging algorithm, which makes "large step" updates from user-level data. Our work demonstrates that ...| arXiv.org
Large Language Model (LLM) agents, acting on their users' behalf to manipulate and analyze data, are likely to become the dominant workload for data systems in the future. When working with data, agents employ a high-throughput process of exploration and solution formulation for the given task, one we call agentic speculation. The sheer volume and inefficiencies of agentic speculation can pose challenges for present-day data systems. We argue that data systems need to adapt to more natively s...| arXiv.org
Human cognitive performance is enhanced by the use of tools. For example, a human can produce a much greater, and more accurate, volume of mathematical calculation in a unit of time using a calculator or a spreadsheet application on a computer. Such tools have taken over the burden of lower level cognitive grunt work but the human still serves the role of the expert performing higher level thinking and reasoning. Recently, however, unsupervised, deep, machine learning has produced cognitive s...| arXiv.org
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-dr...| arXiv.org
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...| arXiv.org
Recent advancements in large language models (LLMs) have significantly advanced the automation of software development tasks, including code synthesis, program repair, and test generation. More recently, researchers and industry practitioners have developed various autonomous LLM agents to perform end-to-end software development tasks. These agents are equipped with the ability to use tools, run commands, observe feedback from the environment, and plan for future actions. However, the complex...| arXiv.org
Foundation models have revolutionized natural language processing through a ``train once, deploy anywhere'' paradigm, where a single pre-trained model adapts to countless downstream tasks without retraining. Access to a Physics Foundation Model (PFM) would be transformative -- democratizing access to high-fidelity simulations, accelerating scientific discovery, and eliminating the need for specialized solver development. Yet current physics-aware machine learning approaches remain fundamental...| arXiv.org
Logical qubits are considered an essential component for achieving utility-scale quantum computation. Multiple recent demonstrations of logical qubits on neutral atoms have relied on coherent qubit motion into entangling zones. However, this architecture requires motion prior to every entangling gate, incurring significant cost in wall clock runtime and motion-related error accumulation. We propose and experimentally realize an alternative architecture which unites qubit motion and in-place e...| arXiv.org
We prove that $S(5) = 47,176,870$ using the Coq proof assistant. The Busy Beaver value $S(n)$ is the maximum number of steps that an $n$-state 2-symbol Turing machine can perform from the all-zero tape before halting, and $S$ was historically introduced by Tibor Radó in 1962 as one of the simplest examples of an uncomputable function. The proof enumerates $181,385,789$ Turing machines with 5 states and, for each machine, decides whether it halts or not. Our result marks the first determinati...| arXiv.org
Plaintiffs and defendants in copyright lawsuits over generative AI often make sweeping, opposing claims about the extent to which large language models (LLMs) have memorized plaintiffs' protected expression. Drawing on adversarial ML and copyright law, we show that these polarized positions dramatically oversimplify the relationship between memorization and copyright. To do so, we leverage a recent probabilistic extraction technique to extract pieces of the Books3 dataset from 17 open-weight ...| arXiv.org
The New York Times's copyright lawsuit against OpenAI and Microsoft alleges OpenAI's GPT models have "memorized" NYT articles. Other lawsuits make similar claims. But parties, courts, and scholars disagree on what memorization is, whether it is taking place, and what its copyright implications are. These debates are clouded by ambiguities over the nature of "memorization." We attempt to bring clarity to the conversation. We draw on the technical literature to provide a firm foundation for leg...| arXiv.org
Directed greybox fuzzing (DGF) focuses on efficiently reaching specific program locations or triggering particular behaviors, making it essential for tasks like vulnerability detection and crash reproduction. However, existing methods often suffer from path explosion and randomness in input mutation, leading to inefficiencies in exploring and exploiting target paths. In this paper, we propose HGFuzzer, an automatic framework that leverages the large language model (LLM) to address these chall...| arXiv.org
Multimodal systems, which process multiple input types such as text, audio, and images, are becoming increasingly prevalent in software systems, enabled by the huge advancements in Machine Learning. This triggers the need to easily define the requirements linked to these new types of user interactions, potentially involving more than one modality at the same time. This remains an open challenge due to the lack of languages and methods adapted to the diverse nature of multimodal interactions, ...| arXiv.org
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 th...| arXiv.org
In recent years, with the rapid development of the depth and breadth of large language models' capabilities, various corresponding evaluation benchmarks have been emerging in increasing numbers. As a quantitative assessment tool for model performance, benchmarks are not only a core means to measure model capabilities but also a key element in guiding the direction of model development and promoting technological innovation. We systematically review the current status and development of large ...| arXiv.org
Serving Large Language Models (LLMs) is a GPU-intensive task where traditional autoscalers fall short, particularly for modern Prefill-Decode (P/D) disaggregated architectures. This architectural shift, while powerful, introduces significant operational challenges, including inefficient use of heterogeneous hardware, network bottlenecks, and critical imbalances between prefill and decode stages. We introduce HeteroScale, a coordinated autoscaling framework that addresses the core challenges o...| arXiv.org
On October 22, 1934, in a famous experiment, Enrico Fermi and his colleagues discovered that a significant increase in induced radioactivity can be obtained when neutrons are slowed down by means of hydrogen atoms. This discovery and its explanation earned him the 1938 Nobel Prize in Physics. One year later, on October 1935, Fermi held a public speech in Palermo, Italy, presenting his findings at the 24th congress of the Italian Society for the Progress of Sciences. The transcription of his s...| arXiv.org
The Wigner's Friend thought experiment stands as one of the most intellectually provocative and challenging conceptual puzzles in quantum mechanics. It compels us to confront profound questions concerning the fundamental nature of reality, the very act of observation, and the possible role that consciousness might play within the quantum measurement process. This article gives a general presentation, beginning with Eugene Wigner's seminal proposal of the original thought experiment. In this p...| arXiv.org
I dispute the conventional claim that the second law of thermodynamics is saved from a "Maxwell's Demon" by the entropy cost of information erasure, and show that instead it is measurement that incurs the entropy cost. Thus Brillouin, who identified measurement as savior of the second law, was essentially correct, and putative refutations of his view, such as Bennett's claim to measure without entropy cost, are seen to fail when the applicable physics is taken into account. I argue that the t...| arXiv.org
Pig-butchering scams have emerged as a complex form of fraud that combines elements of romance, investment fraud, and advanced social engineering tactics to systematically exploit victims. In this paper, we present the first qualitative analysis of pig-butchering scams, informed by in-depth semi-structured interviews with $N=26$ victims. We capture nuanced, first-hand accounts from victims, providing insight into the lifecycle of pig-butchering scams and the complex emotional and financial ma...| arXiv.org
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 ...| arXiv.org
Administrators and developers use SSH client keys and signatures for authentication, for example, to access internet backbone servers or to commit new code on platforms like GitHub. However, unlike servers, SSH clients cannot be measured through internet scans. We close this gap in two steps. First, we collect SSH client public keys. Such keys are regularly published by their owners on open development platforms like GitHub and GitLab. We systematize previous non-academic work by subjecting t...| arXiv.org
Caveats: While we have taken considerable effort to extract reliable data for all charts there are many factors which affect results.| arxiv.org
Cosine-similarity is the cosine of the angle between two vectors, or equivalently the dot product between their normalizations. A popular application is to quantify semantic similarity between high-dimensional objects by applying cosine-similarity to a learned low-dimensional feature embedding. This can work better but sometimes also worse than the unnormalized dot-product between embedded vectors in practice. To gain insight into this empirical observation, we study embeddings derived from r...| arXiv.org
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 A...| arXiv.org
As LLMs become more widely deployed, there is increasing interest in directly optimizing for feedback from end users (e.g. thumbs up) in addition to feedback from paid annotators. However, training to maximize human feedback creates a perverse incentive structure for the AI to resort to manipulative or deceptive tactics to obtain positive feedback from users who are vulnerable to such strategies. We study this phenomenon by training LLMs with Reinforcement Learning with simulated user feedbac...| arXiv.org
In the 1980s, three related impossibility results emerged in the field of distributed computing. First, Fischer, Lynch, and Paterson demonstrated that deterministic consensus is unattainable in an asynchronous message-passing system when a single process may crash-stop. Subsequently, Loui and Abu-Amara showed the infeasibility of achieving consensus in asynchronous shared-memory systems, given the possibility of one crash-stop failure. Lastly, Santoro and Widmayer established the impossibilit...| arXiv.org
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 A...| arXiv.org
This paper introduces a reinforcement learning (RL) platform that enhances end-to-end user journeys in healthcare digital tools through personalization. We explore a case study with SwipeRx, the most popular all-in-one app for pharmacists in Southeast Asia, demonstrating how the platform can be used to personalize and adapt user experiences. Our RL framework is tested through a series of experiments with product recommendations tailored to each pharmacy based on real-time information on their...| arXiv.org
By providing evidence-based clinical decision support, digital tools and electronic health records can revolutionize patient management, especially in resource-poor settings where fewer health workers are available and often need more training. When these tools are integrated with AI, they can offer personalized support and adaptive interventions, effectively connecting community health workers (CHWs) and healthcare facilities. The CHARM (Community Health Access & Resource Management) app is ...| arXiv.org
Malaria can be prevented, diagnosed, and treated; however, every year, there are more than 200 million cases and 200.000 preventable deaths. Malaria remains a pressing public health concern in low- and middle-income countries, especially in sub-Saharan Africa. We describe how by means of mobile health applications, machine-learning-based adaptive interventions can strengthen malaria surveillance and treatment adherence, increase testing, measure provider skills and quality of care, improve pu...| arXiv.org
How can one build a feature-rich, general-purpose, Rust-based operating system (OS) with a minimal and sound Trusted Computing Base (TCB) for memory safety? Existing Rust-based OSes fall short due to their improper use of unsafe Rust in kernel development. To address this challenge, we propose a novel OS architecture called framekernel that realizes Rust's full potential to achieve intra-kernel privilege separation, ensuring TCB minimality and soundness. We present OSTD, a streamlined framewo...| arXiv.org
Combining the exquisite angular resolution of Gaia with optical light curves and WISE photometry, the Gaia Gravitational Lenses group (GraL) uses machine learning techniques to identify candidate strongly lensed quasars, and has confirmed over two dozen new strongly lensed quasars from the Gaia Data Release 2. This paper reports on the 12 quadruply-imaged quasars identified by this effort to date, which is approximately a 20% increase in the total number of confirmed quadruply-imaged quasars....| arXiv.org