跳到正文
研踪PaperTrail

YOUR RESEARCH, IN SIGHT

让下一篇重要论文，不再错过。

覆盖计算机全学科，以 AI 为快捷入口。关注方向、作者和关键词，把发现变成持续的研究积累。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+2471新增论文
+2025其中 AI 相关
+163已有论文修订
+100新增 AI 导读

arXiv 最近完成扫描：2026-11-10 20:36 · 全站最近入库：2026-11-10 15:22各信源进度 →

统计口径与数据新鲜度

统计 2026-11-09 21:44 — 2026-11-10 21:44（北京时间） · 点击数字查看对应简报；AI 相关包含在新增论文内。

最近完成检查：2026-11-10 20:36最近发现新增版本：2026-11-10 15:22库内论文最新更新时间：2026-11-10 02:59

每分钟刷新已保存的数据。空检查不代表采集异常；新收录可能包含历史论文。AI 导读按新增版本计数。

20,643 已收录论文17,531 AI 相关5,643 有修订112 HF 社区发现

浏览论文库

研究日报 / 周报 →按信源浏览 →

时间范围跟随排序：版本更新、本站收录或首次发表。最新收录可能包含历史论文。部分信源异常，查看状态 →

找到 15,443 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-10 20:33（北京时间）
版本更新 · v3arXiv:2605.16757

NeuroMAS: Multi-Agent Systems as Neural Networks with Joint Reinforcement Learning

Haoran Lu, Luyang Fang, Wenxuan Zhong, Ping Ma

原文摘要

Multi-agent language systems are often built as hand-designed workflows, where agents are assigned semantic roles and communication protocols are specified in advance. We propose NeuroMAS, a framework that treats a multi-agent language system as a trainable and scalable neural-network-like architecture with hidden textual intermediate signals. In NeuroMAS, language-agent nodes are role-free but structure-aware: the topology determines how information can flow, while reward-based training determines how nodes communicate, specialize, and coordinate. This formulation shifts multi-agent design from workflow engineering toward architecture design, where depth, width, connectivity, and growth protocol become scalable sources of capability. We provide a theoretical perspective showing why such modular text-valued computation can be parameter-efficient when tasks admit stable decompositions. Experiments show that NeuroMAS improves over both inference-time collaboration methods and trained baselines, and that its gains are not explained by trainable parameter count alone. We further find that organizational scaling is path-dependent: larger systems can fail when trained from scratch, but become stronger when grown progressively from smaller trained systems. These results suggest that learned multi-agent organization is a promising scaling axis for language-model reasoning and generation.

cs.AI · cs.MA · stat.ME · stat.ML

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08772

StarLoc: A Practical Approach to Encoding Geographic Locators in IPv6 for Starlink

Vanya Singla, Mostafa Abdollahi, Jinwei Zhao, Jianping Pan

原文摘要

LEO satellite networks such as Starlink provide Internet connectivity across continents, oceans, and airspace. IP geolocation providers map Starlink's IPv6 prefixes to locations mainly through its self-published GeoFeed. However, users may be far away from their assigned Points of Presence (PoPs), one label covers an entire prefix, and labels and PoP mappings can change while addresses stay the same. We characterize Starlink's IPv6 addressing using 925 historical GeoFeed snapshots (2024--2026), one year of PoP mappings, and over 7.9 million observed IPv6 addresses. Our findings reveal that customer prefixes lack clear geographic structure, prefix labels lie a median of 540km from their PoPs, and relabeling a single /40 can affect the reported locations of up to 65,536 possible /56 delegations. To address these challenges, we propose StarLoc, a geographic Locator--Identifier separation scheme that embeds terminal location into IPv6 addresses using a resolution-adaptive H3 Locator while preserving existing Starlink prefixes, /56 delegations, and stable Endpoint Identifiers. To balance accuracy and privacy, StarLoc reveals only cells with at least a certain number of terminals, and operators can cap resolution in sensitive regions or withhold location entirely. Unlike topology-based schemes, StarLoc encodes physical location and lets Locators follow moving users.

cs.NI

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.10050

Selected-Risk Calibration for Feasibility at Returned Decisions

Raghav Thind, Youran Sun, Ling Liang, Haizhao Yang

原文摘要

When an optimization problem uses a surrogate constraint, a decision that is feasible for the surrogate need not be feasible for the true constraint. We propose Selected-Risk Calibration (SRC), a framework that calibrates a learned constraint against the optimizer's returned decisions for objectives drawn from a given distribution. SRC guarantees that, with probability at least $1-δ$, every margin it certifies on held-out objectives has selected risk, the probability that the returned decision is infeasible, of at most $γ$. We also show that the random-input coverage guarantee of split conformal prediction does not bound the selected risk, and that likelihood-ratio reweighting for feedback covariate shift does not apply when the learned constraint is active at the returned decision. Our experiments show that a conformal margin covering about $90\%$ of random inputs covered as few as $17\%$ of the returned decisions, and every certified margin had a violation rate below its target on unseen objectives. Moreover, at equal cost, SRC's correction had fewer or equal infeasible solutions to the baselines in 57 of 62 comparisons. The SRC framework applies to any setting in which a surrogate model chooses a decision whose feasibility can be verified for a sample of instances.

cs.LG · stat.ML

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08770

ReKV: Overlapping Quantized KV Transfer and Suffix Recomputation

Guangjian Dong

原文摘要

Prefill--decode disaggregation moves a prompt's key--value (KV) cache between workers before the first token can be generated. Reducing this handoff to either cache compression or recomputation misses a useful choice: which layers to transmit, at what precision, and which layers to execute again on the decode worker. ReKV transfers a quantized prefix KV cache and one BF16 boundary state, then reconstructs a contiguous suffix while the prefix is in flight. In a model-shaped synthetic benchmark on cross-node H100s, the Q8 configuration meeting separately evaluated quality criteria reduces 8K handoff latency by $2.36\times$, $2.84\times$, and $3.65\times$ over BF16 transfer for Qwen2.5-7B, Qwen2.5-3B, and Qwen3-8B. It is also faster than the best measured transfer-only or recompute-only action at those points. A separate real-weight BF16 recomputation experiment reduces prefill-to-token-return time by $1.16$--$1.63\times$. The preferred action changes with prompt length and link bandwidth.

cs.DC

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2611.04085

Follow the Features: SAE Feature-Guided Data-Efficient Instruction-Tuning of Large Language Models

Ryan Swift, Konstantinos Psounis

原文摘要

Selecting small, effective instruction-tuning datasets requires identifying which aspects of the original data to retain. We introduce Follow the Features (FtF), a data-selection method that uses sparse autoencoder (SAE) features to characterize training data and guide subset construction. FtF ranks features using dataset-level activation statistics, scores samples by their expression of highly ranked features, and incorporates feature resampling to encourage diversity. The LLM-SAE pair used for selection is independent of the model being fine-tuned, enabling reuse across target models. We evaluate FtF on two model-dataset settings at a training budget of 100 samples with a fixed sequence length of 256 tokens. FtF achieves performance competitive with GRAPE and S2L, with relative rankings varying across settings. On Qwen 3 8B trained on Alpaca, FtF exceeds full-dataset training by 1.59 percentage points in the harmonic mean of an AlpacaEval-based win rate and Open LLM Leaderboard performance, although full-dataset training retains a higher Leaderboard score. These results support SAE feature statistics as a useful signal for data-efficient instruction tuning.

cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2611.04081

RAPID: Increasing Efficiency of Internal Distillation via Randomized Methods

Ryan Swift, Konstantinos Psounis

原文摘要

Knowledge distillation is a technique used to transfer the expertise of large teacher models into small, efficient student models. Recent advances have integrated the distillation of internal activations, more directly training students to learn similar representations as the teacher. One of the core challenges of internal distillation is the dimensional incompatibility between the student and teacher internal representations. Existing methods add learnable matrices to project the larger teacher activations into the smaller representation space of the student. This is a notable point of inefficiency, as learning additional parameters for the projectors increases training costs, only for them to be discarded immediately after training. We introduce RAPID (RAndomized Projection for Internal Distillation) a novel distillation methodology which eliminates the wasteful projection learning by leveraging randomized methods. RAPID utilizes an efficient transform construction based on the Johnson-Lindenstrauss lemma to compress the teacher representations into the student embedding space with strong theoretical guarantees of structural preservation. RAPID also employs a combination of exponential moving average (EMA) and homoscedastic uncertainty weighting in a novel multi-objective function designed for internal distillation. Our experiments in post-training for reasoning demonstrate that RAPID increases training efficiency by as much as 21.1% while maintaining performance comparable to existing methods.

cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2610.23047

DIFFLET: A Fast Inference Engine for Diffusion Transformer Models on AWS Trainium

Bin Ma, Min Yu, Vijay Kodamalla, Johnny Liu, Da Li, Dong Li

原文摘要

On AWS Trainium2, every component of a diffusion model runs as a program compiled ahead of time for fixed shapes, while pipelines differ across models, prompt lengths across requests, and caching decisions across denoising steps. We present Difflet, a serving engine in which five model pipelines share one compilation backend, joint attention is split across cores by attention head, and a host controller decides at each denoising step whether to run the denoiser or reuse its earlier output. A small probe program computes the caching signal on the device and returns one number per step, and a resident model process keeps every program loaded across requests. Across five image and video model families, fused attention accelerates denoiser steps by 1.2-4.3x over the same framework using PyTorch/XLA attention, and at 32.8k visual tokens it is the only attention that compiles. On four of the five models, resident serving is 1.5-3.4x faster than standalone runs that reload weights from a warm file cache for each generation.

cs.DC · cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2610.10854

KDFP: A first-principles approach to knowledge distillation in large language models

Ryan Swift, Konstantinos Psounis

原文摘要

Knowledge distillation is an established technique for improving the capabilities of small, efficient student models by training them with the representations of larger, more capable teacher models. Much of the recent work in the distillation of large language models (LLMs) has focused on distilling abilities learned during post-training, such as instruction following, chain-of-thought reasoning, and tool usage. This has left a large research gap in general knowledge distillation for LLMs, which is essential for developing efficient and private systems suitable for deployment on edge devices. We take a first-principles approach, evaluating previous lessons from prior works and conducting new explorations to develop a distillation methodology suitable for modern LLMs. We present KDFP, a novel methodology for white-box general knowledge distillation in LLMs. We demonstrate that KDFP outperforms existing methods by 1.6% $-$ 4.9% across 9 benchmarks while increasing training efficiency by up to 99.1% through ephemeral parameter reduction.

cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2509.00031

End-to-End On-Device Quantization-Aware Training for LLMs at Inference Cost

Qitao Tan, Xiaoying Song, Ningxi Cheng, Guoming Li, Jun Liu, Lingzi Hong, Caiwen Ding, Jundong Li 等 14 位作者

原文摘要

Quantization reduces the deployment cost of large language models (LLMs). However, post-training quantization (PTQ) cannot fine-tune model parameters and often degrades accuracy in low-bit settings, while quantization-aware training (QAT) requires backpropagation and incurs high memory overhead. We propose ZeroQAT, a zeroth-order QAT framework supporting both weight and activation quantization. By using forward-only gradient estimation, ZeroQAT removes backpropagation and substantially reduces memory and computation while retaining end-to-end optimization benefits. We further introduce a lightweight variant of ZeroQAT for quantized fine-tuning, which freezes and pre-quantizes most parameters to further cut memory usage. Experiments show that ZeroQAT consistently outperforms representative PTQ and QAT baselines while requiring significantly less memory. For example, ZeroQAT enables fine-tuning of a 13B model at extremely low bit-widths (e.g., 2-4 bits) on a single 8GB GPU, and even allows fine-tuning a 6.7B model on a OnePlus 12 smartphone, demonstrating its practicality for end-to-end QAT on resource-limited edge devices.

cs.LG · cs.AI

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.10048

Veterans' Mental Health Crisis Severity Annotation: Do LLMs Agree More Than Human Experts?

K M Sajjadul Islam, Pranav Yaramalla, Nadiyah Johnson, Mark Flower, Praveen Madiraju, Zeno Franco

原文摘要

Large language models (LLMs) offer a scalable approach for annotating mental health and crisis-related social media content, but their reliability in safety-critical tasks remains unclear. This study evaluates LLM-based crisis-severity annotation of veteran-related social media posts against annotations from eight Veteran Subject Matter Experts (SMEs). Using 149 tweets and five severity categories, we evaluate 15 open-source and 10 proprietary LLMs across eight prompt variants, examining annotation performance, inter-model agreement, prompt sensitivity, and high-severity crisis detection. Human reviewers showed higher ordinal than nominal agreement (α = 0.494), indicating that disagreements often occurred between neighboring severity levels. In contrast, selected LLMs demonstrated substantially higher inter-model consistency, with ordinal agreement (α) reaching 0.8336 under the baseline prompt. Proprietary models generally showed better agreement with expert annotations and greater stability than open-source models. GPT-5.6-Terra achieved the best overall performance, while the open-source Qwen2.5-7B showed the strongest performance in identifying Acute Crisis with Suicidal Ideation cases. However, higher model consistency did not necessarily correspond to agreement with human judgments or reliable detection of rare high-severity cases. Label-flipped demonstration experiments also revealed model-specific sensitivity to misleading in-context examples, while explanatory demonstrations did not consistently improve performance. Overall, LLMs show promise as scalable complements to expert annotation, but require careful evaluation before use in high-stakes mental health applications.

cs.AI

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08763

CrowdDirector: Runtime Orchestration of Crowd Simulation through Natural-Language Directives

Xiaohan Sun, Vuk Radovanovic, Merce Alvarez, Carol O'Sullivan, Yiorgos Chrysanthou, Panayiotis Charalambous

原文摘要

Natural language offers an intuitive way to direct a virtual crowd while a simulation is running. Handwritten rules execute efficiently but cannot respond flexibly to varied instructions, whereas asking a large language model (LLM) to control the crowd at every update introduces repeated latency, cost, and the risk of unsupported actions. We present CrowdDirector, a framework that uses an LLM only when the user adds or changes an instruction. The LLM translates the request into a structured instruction, and a scene validator checks whether the entities and actions it references exist in the current Unity scene. Accepted instructions are stored and remain continuously active as the simulation is running. At each simulation tick, a graph neural network uses the current scene state and active instruction to choose an available action and target for each agent, while Unity handles movement, interactions, and animation. The network is trained offline on simulated scenarios, with LLM guidance on how agents should respond to instructions. Our evaluations show the scene validator's feasibility decisions matched human judgements more often than the LLM's own assessment, and on a behavioural benchmark CrowdDirector follows user instructions more closely than handwritten rules while avoiding repeated LLM calls. It therefore provides a practical middle ground between fixed rules and continuous LLM control.

cs.GR · cs.AI · cs.MA

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2607.09993

Beyond Bayesian Nash: Learning Minimax-Regret Equilibria for Adversarial Team Games under Asymmetric Information

Naman Aggarwal, Jonathan P. How

原文摘要

Adversarial team games (ATGs) with asymmetric information, such as adversarial path-finding, goal search, and reachability games on graphs, require strategies that are robust to hidden opponent types, such as a hidden goal flag, and to deception. Under asymmetric information, deception is seen as strategic shifts in the type distribution such that the omniscient opponent can collude with Nature and condition its play on the observed type. Existing risk-neutral solution concepts, such as Bayesian Nash equilibrium (BNE), are sensitive to distribution shifts, while distributionally robust approaches provide guarantees only within a prescribed ambiguity set. To address these limitations, we introduce Probabilistically Robust Minimax-Regret Equilibrium (PR-MRE), a novel equilibrium concept that combines the distribution-free robustness of minimax-regret reasoning with probabilistic information from a nominal type distribution. PR-MRE minimizes worst-case regret over a high-confidence subset of the type space, providing protection against strategic redistribution of probability mass while avoiding the conservatism of fully distribution-free approaches. We show that, for finite normal-form Bayesian games, approximate PR-MRE computation can be formulated as a robust bilinear program and derive a tractable semidefinite relaxation. We then adapt this relaxation into a novel meta-solver within a robust double-oracle framework, PRMRE-PSRO, enabling population-based learning of approximate PR-MRE strategies via deep reinforcement learning best responses. Experiments on graph-structured adversarial team games demonstrate that PR-MRE discovers strategies with substantially improved worst-case performance across hidden types compared to risk-neutral Bayesian Nash equilibrium solutions, resulting in more robust behavior under strategic distribution shifts.

cs.GT · cs.AI · cs.LG · cs.MA

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2605.20079

Probability-Conserving Flow Guidance

Parsa Esmati, Junha Hyung, Amirhossein Dadashzadeh, Jaegul Choo, Majid Mirmehdi

原文摘要

Diffusion and flow-based generative models dominate visual synthesis, with guidance aligning samples to user input and improving perceptual quality. However, Classifier-Free Guidance (CFG) and extrapolation-based methods are heuristic linear combinations of velocities/scores that ignore the generative manifold geometry, breaking probability conservation and driving samples off the learned manifold under strong guidance. We analyse guidance through the continuity equation and show its effect decomposes into a divergence term and a score-parallel term defined invariantly across parameterizations. We prove the divergence term blows up structurally as sampling approaches the data manifold, motivating a time-dependent schedule alongside score-parallel attenuation. The resulting plug-and-play rule, Adaptive Manifold Guidance (AdaMaG), targets both terms at no additional inference cost. Finally, we show that several existing saturation and manifold preserving heuristics can be read as acting on one of these two terms. On image generation benchmarks, AdaMaG improves realism, reduces visible artefacts, and induces controlled desaturation under strong guidance.

cs.CV · cs.AI · cs.LG · eess.IV

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08756

Debate to Deceive: Poisoning LLM's Fact-Checking via Game-Theoretic Multi-Agent Debate

Liting Huang, Huizhi Liang, Yi Lei, Xingjun Ma, Varun Ojha

原文摘要

Knowledge attacks on Large Language Models, such as manipulating fact-checking judgments, pose significant risks to AI reliability. Existing approaches -- adversarial suffix/template injection and single-agent self-reflection -- lack robustness across diverse defenses and are prone to confirmation bias and costly trial-and-error, respectively. We propose Debate-to-Deceive (D2D), a multi-agent debate framework grounded in human debate structure and rhetorical style, enabling agents to generate adversarial passages that are semantically coherent, rhetorically effective, and resistant to early confirmation bias. D2D formalizes the adversarial process as an evolutionary game in which agents adaptively evolve their rhetorical strategies (Logos, Pathos, Ethos) through competition. To manage the intractably large LLM output space, we decouple agent behavior into a strategic layer, which selects a rhetorical action from a compact strategy space via evolutionary-inspired learning dynamics, and a generation layer, where the LLM produces a concrete passage conditioned on that strategy and the debate context. On four fact-checking benchmarks (FEVER, HealthVer, CLIMATE-FEVER, and SciFact), D2D substantially outperforms existing attacks and evades most defenses.

cs.AI

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08757

Goals as a Representational Layer for Human-AI Collaboration

Gaia Molinaro, Lev Tankelevitch, Sean Rintel, Scott Counts

原文摘要

Goals are central to how people, teams, and organizations structure their work. Yet, while generative artificial intelligence (AI) systems increasingly operate across a worker's context, information about why a person is pursuing a task, beyond local prompts and interactions, is typically missing. We argue that goals should become an explicit representational layer mediating interactions among people, AI systems, and organizations. Because human goals are hierarchical, partially latent, and dynamically evolving, such a layer must be multidimensional, both grounded in human articulation and inferred from available evidence, and continuously updated. A goal layer could connect immediate requests to broader outcomes, improve the organization and retrieval of relevant context, support reflection and coordination, and provide a foundation for training, aligning, and evaluating AI systems in terms of human-centered outcomes. We highlight these opportunities while identifying open challenges involving goal inference, socio-cognitive frictions, alignment, privacy, and governance.

cs.HC · cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2608.13410

Who Speaks Matters: Authority-Aware Multi-View RAG over Italian Parliamentary Proceedings

Mirko Tritella, Riccardo Pozzi, Matteo Palmonari

原文摘要

Parliamentary proceedings are a primary record of democratic deliberation, yet their volume and fragmentation make multi-perspective access difficult for citizens, journalists, and researchers. Applying Retrieval-Augmented Generation (RAG) to parliamentary transcripts introduces three specific risks: dominance of the most frequent speakers, inability to weight speakers according to topical expertise, and citation misattribution in politically sensitive text. We present ParliamentRAG, a RAG system for the Italian Chamber of Deputies that addresses these risks jointly. Its core contribution is a topic-dependent authority model that estimates each speaker's authority as a function of the current query, combining interpretable components such as profession, education, and previous interventions. Given a user query, the system retrieves relevant speech chunks, identifies topic-relevant experts across parliamentary groups, and generates a summary synthesizing their perspectives, accompanied by supporting quotations. ParliamentRAG is evaluated against Google NotebookLM on 15 policy topics via a two-level protocol combining automated metrics and blind A/B human evaluation by six domain experts. The system achieves higher coverage across political groups (0.97 vs. 0.95), perfect quotation faithfulness (1.00 vs. 0.95), and stronger expert preferences on source-related dimensions, while NotebookLM remains stronger on prose-oriented dimensions.

cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2608.13660

What to Preserve, Where to Adapt: A Depth-Wise Analysis of Forgetting in Continual Gynecological Image Segmentation

Amal Saqib, Tausifa Jan Saleem, Numan Saeed, Mohammad Yaqub

原文摘要

The clinical management of gynecological diseases often relies on medical imaging for diagnosis, treatment planning, and follow-up. Segmentation in this setting is challenging because successive tasks may differ in imaging modality, target anatomy, pathology, and annotation structure. Continual learning allows models to adapt to new tasks without simultaneous access to previous datasets. However, when successive tasks differ substantially, learning a new task can degrade performance on earlier ones, a problem known as catastrophic forgetting. Understanding where adaptation disrupts previous knowledge can help guide the design of more targeted continual-learning strategies. We investigate how forgetting changes as different parts of an encoder--decoder network are allowed to adapt. We progressively expand the trainable region of a 3D nnU-Net backbone from the bottleneck toward input- and output-proximal blocks. Under a shared learning rate, adaptation near the bottleneck largely preserves previous-task performance but provides limited current-task learning, whereas broader adaptation improves current-task performance but sharply increases forgetting. This trade-off persists even when the average change in trainable backbone parameters is approximately comparable. Assigning different learning rates to different blocks substantially reduces forgetting when part of the backbone is trainable, although this changes both the size and location of the updates. Forgetting still increases as more blocks are trained and remains severe when the full backbone is updated. These results show that forgetting depends not only on how much the model changes, but also on which parts of the model are allowed to change.

cs.CV · cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v4arXiv:2604.11748

LangFlow: Continuous Diffusion Rivals Discrete in Language Modeling

Yuxin Chen, Chumeng Liang, Hangke Sui, Ruihan Guo, Chaoran Cheng, Jiaxuan You, Ge Liu

原文摘要

Continuous diffusion has served as a foundation for high-fidelity, controllable, and few-step generation across continuous data modalities such as images, videos, and molecular structures. However, in language modeling, prior continuous diffusion language models (DLMs) lag behind discrete counterparts. Existing categorical and simplex-based approaches operate over extremely large and sparse language spaces, while prior embedding-space approaches avoid this sparsity but lack a well-selected design space. In this work, we close this gap with LangFlow by connecting embedding-space DLMs to Flow Matching, alongside three key innovations: (1) we derive a novel ODE-based NLL bound for principled evaluation of continuous flow-based language models; (2) we propose an information-uniform principle for setting the noise schedule, which motivates a learnable noise scheduler based on a Gumbel distribution; and (3) we revise prior training protocols by incorporating self-conditioning, which improves both likelihood and sample quality for embedding-space DLMs and behaves differently from its use in discrete diffusion. Putting everything together, LangFlow is competitive with top discrete DLMs on generative quality, reaching a generative perplexity of 92.3 on LM1B and 36.5 on OpenWebText. It also outperforms discrete counterparts on reasoning tasks like Sudoku. As the first continuous DLM shown to rival discrete diffusion in generative quality, LangFlow provides clear evidence that continuous diffusion is a promising paradigm for language modeling. Code: https://github.com/nealchen2003/LangFlow.

cs.CL · cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.10046

Topological Deep Learning Challenge 2026: Bridging the Gap

Guillermo Bernárdez, Lev Telyatnikov, Mathilde Papillon, Marco Montagna, Louisa Cornelis, Louis Van Langendonck, Antonio Purificato, Bruno Neri 等 71 位作者

原文摘要

This paper describes the 2026 edition of the Topological Deep Learning Challenge: Bridging the Gap, hosted at the second Topology, Algebra, and Geometry in Data Science (TAG-DS) Conference. Where previous editions grew the modelling and data ecosystem of Topological Deep Learning (TDL), this edition set out to place Graph Neural Networks (GNNs) and Topological Neural Networks (TNNs) side by side under a single evaluation pipeline. Participants implemented recent state-of-the-art architectures in TopoBench, with Track 1 covering GNNs and Track 2 covering TNNs. They evaluated these implementations on shared synthetic benchmarks generated with GraphUniverse, which independently controls homophily, average degree and degree-distribution heaviness. The challenge yielded 68 qualifying submissions from 34 teams, spanning the graph, simplicial, cellular, hypergraph and combinatorial domains. For the first time, submissions were also tracked on a public leaderboard, which we summarize and analyze here. Our analysis finds that the differences between the submitted tracks are strongly task-dependent: the mean accuracy gap is modest on community detection, while on triangle counting the Track 1 mean error is 1.85 times the Track 2 mean. Several structurally informed GNNs also outperform the Track 2 mean on triangle counting.

cs.LG · cs.AI

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08755

AtariBench: Evaluating Spatiotemporal Reasoning of Vision-Language Models in Atari Gameplay

Mingjia Huo, Yao Fu, Bo Chang, Yaqing Wang, Dawei Zhu, Dylan Zhang, Xinyang Yi, Lichan Hong 等 11 位作者

原文摘要

Vision-Language Models (VLMs) have achieved strong performance on static visual understanding benchmarks, yet their ability to act in dynamic environments remains less well understood. We introduce AtariBench, a controlled benchmark that evaluates VLMs across 20 Atari 2600 games, requiring models to make sequential decisions from high-temporal-resolution gameplay streams. By decoupling fast in-game dynamics from wall-clock inference speed, AtariBench focuses on whether models can perceive rapid visual state changes, reason over spatiotemporal dynamics, and ground actions in the current observation. Our experiments reveal substantial limitations in state-of-the-art VLMs: models often fail to infer object motion and direction, exhibit brittle sequential decision-making, and over-rely on textual interaction history despite newly appended visual feedback. We further introduce a separability analysis to identify which in-game scores provide reliable signals of model competence, and use the benchmark to study how memory design and prompting strategies affect performance. Overall, AtariBench exposes a persistent gap between passive visual perception and active decision-making, providing a lightweight and rigorous testbed for developing VLMs that can reason and act in dynamic environments.

cs.CV · cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2511.20798

Physics Steering: Causal Control of Cross-Domain Concepts in a Physics Foundation Model

Rio Alexa Fear, Payel Mukhopadhyay, Michael McCabe, Alberto Bietti, Miles Cranmer

原文摘要

Foundation models trained on text and images have been shown to develop abstract internal features that align with human concepts and that can be directly manipulated via activation steering to alter model behaviour. Whether scientific foundation models learn similarly abstract and domain-general representations has remained an open question. Inspired by recent work identifying single directions in activation space that control complex behaviours in large language models, we show that Walrus, a large physics foundation model, learns linearly steerable representations of physical phenomena. By computing the difference between activations representing contrasting physical regimes, we identify single directions in activation space that correspond to vorticity, diffusion, and even temporal progression. We find that injecting these concept directions back into the model during inference enables fine-grained causal control: vortices can be induced or removed, diffusion enhanced or suppressed, and simulations sped up or slowed down. Moreover, directions derived from one physical system also steer the behaviour of other physical systems, suggesting that they are domain-general. These results provide early evidence that scientific foundation models can learn general representations of physical principles, and further support for the Linear Representation Hypothesis.

cs.LG · cs.AI · physics.comp-ph

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08754

Measurement-Aware Multimodal LLM for No-Reference Low-Dose CT Quality Assessment

Kerem Basim, Tolga Demiroglu, Mehmet Ozan Unal, Metin Ertas, Isa Yildirim

原文摘要

No-reference image quality assessment (IQA) of low-dose CT predicts the score that radiologists would assign to a slice without a full-dose reference, as needed to compare dose protocols and denoisers. Trained CT IQA networks agree well with radiologists but output only a score, whereas multimodal large language models (MLLMs) can explain a score but perform poorly on CT when only prompted. We propose a measurement-aware MLLM, built on Qwen3.8-27B, that rates a slice from its image together with six image-domain measurements of noise, sharpness, high-frequency content and streaks, given to the language model as percentiles in text. On the LDCT-IQA 2023 test set, it reaches an Overall score of 2.735, on par with the retrained challenge winner (2.735) and MedSigLIP-FiLM (2.713), and substantially above its base MLLM used as a zero-shot agent. Shuffling the measurements at test time reduces the score by 0.048, mostly through noise, but the same measurements also lead the model to assign higher scores to over-smoothed denoised images.

eess.IV · cs.CV

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08753

Random Symbolic Boosting: A Geometric Approach to Symbolic Regression

José R. Ortegaa, José Manuel Muñoz Contreras, Leonardo Trujillo

原文摘要

Symbolic models offer interpretability, but their construction through generational evolution can be computationally expensive. This article presents Random Symbolic Boosting (RSB), a method that constructs explicit additive symbolic models via boosting over populations of randomly generated syntax trees, eliminating the need for generational evolution. The idea combines two observations: high-dimensional spaces can be spanned by enough linearly independent vectors, and the semantics of syntax trees can be directly manipulated as such vectors. It is shown that, for Ordinary Least Squares, the boosting process guarantees a monotonic reduction of the training error for both regression and classification. Furthermore, an adaptive geometric early stopping mechanism is introduced. It is based on the angle between the prediction and target vectors, is independent of the problems scale, and requires no adjustments across tasks. Evaluated on a standard benchmark for symbolic regression, the method is located at the Pareto-optimal point between predictive accuracy and training time, ranking first on the real-world datasets. Additionally, it was tested on a large tabular classification benchmark with 200 problems, achieving competitive results.

cs.LG · cs.AI

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08752

AttentionSwitch: Jailbreak Defense for LLMs via Joint Learning of Sparse Attention-Head Selection and Scaling

Doniyorkhon Obidov, Honggang Yu, Xiaolong Guo, Kaichen Yang

原文摘要

Large language models have achieved remarkable performance across a wide range of tasks, yet they remain susceptible to jailbreak attacks that circumvent safety mechanisms and elicit harmful or unintended outputs. Although safety can be improved through model fine-tuning, such approaches can be computationally expensive and may degrade previously acquired capabilities through catastrophic forgetting. Recent jailbreak defenses have therefore explored targeted interventions on specific model components associated with unsafe behavior. However, existing approaches typically decouple two inherently related decisions: where to intervene and how to modify the selected components. Such sequential optimization may be suboptimal because the optimal intervention location can depend on the intervention applied there, and vice versa. In this work, we introduce AttentionSwitch, the first jailbreak defense to jointly optimize intervention sites and strengths within a frozen LLM. Specifically, AttentionSwitch learns a sparse set of attention heads together with query-conditioned multiplicative scaling factors, allowing the outputs of the selected heads to be either attenuated or amplified without altering the directions of their representations. Across popular language models and jailbreak attacks, AttentionSwitch outperforms existing defenses, with particularly strong results when training data is scarce.

cs.AI · cs.CL · cs.CR

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08749

How are viral trends different from educational and commercial uses of LLM image generation and editing? A privacy-preserving descriptive analysis

Will Tucker-Ray, Marika Swanberg, Alexander Knop, Marvin Chow, Eric Culbertson, Julian Jacobs, Pritish Kamath, Ravi Kumar 等 12 位作者

原文摘要

Gemini's Nano Banana models, first introduced in late 2025, led to a shift in the capabilities and use of image generation and editing. We utilize Urania, a Google-developed framework for generating insights about LLM chatbot interactions with rigorous differential privacy (DP) guarantees, to analyze use cases and intents, techniques and ways of prompting, without ever looking at a single image or individual prompt. We select a subsample of 2.1M conversations in early 2026 which utilized Nano Banana models, into roughly 800 clusters in five categories: (1) Social & Viral Sharing, (2) Brand & Marketing Collateral, (3) Educational & Technical Information Media, (4) Product Visualizations & E-commerce, and (5) Architectural & Spatial Designs. We then compare these clusters on 21 features of use, measuring aspects of prompts' (a) editing techniques, (b) artistic style requests, (c) composition, and (d) chatbot interactions. This allows us to describe a baseline of how people are using image generation and editing tools for different use cases---(including growing and emergent use cases which were not possible before Nano Banana)---in a privacy-preserving manner.

cs.CY · cs.AI

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.10045

Don't Blame The Preferences: Examining The Emergence of Sycophancy in LLM Training

Chao Peter Yang, Catherine Zehr, Pranav Bijith, Brandon Fain

原文摘要

AI sycophancy, the tendency of large language models (LLMs) to flatter users and affirm their beliefs, can undermine accuracy and contribute to misinformation, dependence, and overconfidence. Although prior work has benchmarked sycophancy in released models and proposed mitigations, its origins during training remain poorly understood. Preference alignment methods such as RLHF and DPO are often blamed, but their effects are rarely compared with earlier training stages. We investigate sycophancy across the training pipelines of three model families and three benchmarks. We find that sycophancy is largely established by the end of supervised fine-tuning (SFT), rather than introduced by preference alignment. Across all nine model benchmark combinations, SFT checkpoints attain at least 92% of preference aligned models' sycophancy scores and at least 85% of final models' scores; on average, they are about as sycophantic as final models. Preference alignment increases sycophancy in only four of nine cases, by an average of 4.2 percentage points when it does. To probe the underlying causes, we isolate model representations associated with sycophancy using mechanistic interpretability and measure their activation on training datasets from each stage. SFT datasets show roughly the same average activation as other training data but contribute substantially more examples, and the highest-scoring individual datasets all come from SFT. Our findings suggest that addressing sycophancy requires interventions throughout the training pipeline, particularly in data selection and augmentation for pre-training and SFT.

cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2505.15417

Robust Multimodal Learning via Entropy-Gated Contrastive Fusion

Leon Chlon, Maggie Chlon, MarcAntonio M. Awada

原文摘要

Real-world multimodal systems routinely face missing-input scenarios, and in reality, robots lose audio in a factory or a clinical record omits lab tests at inference time. Standard fusion layers either preserve robustness or calibration but never both. We introduce Adaptive Entropy-Gated Contrastive Fusion (AECF), a single light-weight layer that (i) adapts its entropy coefficient per instance, (ii) enforces monotone calibration across all modality subsets, and (iii) drives a curriculum mask directly from training-time entropy. On AV-MNIST and MS-COCO, AECF improves masked-input mAP by +18 pp at a 50% drop rate while reducing ECE by up to 200%, yet adds 1% run-time. All back-bones remain frozen, making AECF an easy drop-in layer for robust, calibrated multimodal inference.

stat.ML · cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2610.12407

Diffusion-Steering-Based MPC with a JEPA World Action Model

Shashank Hegde, Alexander Popov, Elie Aljalbout, Nikolai Smolyanskiy

原文摘要

World action models (WAMs) predict actions and future observations, typically from a reconstruction-based representation that carries noisy, redundant information which can complicate downstream predictions. We introduce LeWAM, a bidirectional transformer for forward, backward, inverse dynamics and policy prediction, on a decoder-free JEPA latent trained end-to-end through all four modes. We see the following benefits: 1) Alignment: linear probes read robot and object state from LeWAM's latent better than from a regular Le World Model (a forward-only JEPA world model), while the latent ignores visual distractors as well as LeWM does and far better than a reconstruction-based WAM. 2) Acting: Closed-loop evaluations of LeWAM match a regular flow-matching policy trained on the same encoder at matched size, while also providing a world model. 3) Planning: Sampling raw actions when planning with WAMs lets MPC exploit dynamics-model inaccuracies; planning in the noise space of the policy head instead improves the closed-loop performance of these WAMs.

cs.RO · cs.AI · cs.CV

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.08745

I Don't Care How You Hide It, I Care What You Do: Runtime Detection of Malicious ML Models

Hala Ali, Andrew Case

原文摘要

Pre-trained machine learning (ML) models shared through public platforms can be weaponized to execute malicious payloads during loading or inference. Existing scanners prove inadequate for detecting malicious models as they rely on static analysis and function denylists, support limited serialization formats, and miss attacks that replace denylisted functions with equivalent alternatives, exploit unsupported loading paths, or abuse legitimate framework APIs. To address this problem, we present ModelShield, the first runtime behavioral detection system for malicious ML models that attributes observed system-level behavior to specific model operations. It combines application-level instrumentation of ML framework APIs with eBPF-based syscall tracing, tracking model behavior across spawned threads and processes to separate model-triggered activity from environment noise. Attributed syscall sequences are classified using 195 detection patterns mapped to MITRE ATT&CK techniques. We implemented ModelShield to support 17 ML frameworks and 36 formats. On an independent dataset, it is the only evaluated tool to achieve zero false positives, with false negatives limited to payloads that produce no malicious runtime behavior. It detected all 29 published evasion variants spanning function substitution, loading-path obfuscation, and TensorFlow API abuse, where no existing scanner achieved complete coverage. Across 145,000 models scanned from public platforms, it captured 41 malicious models on Hugging Face, including a reverse-shell payload and a payload that performed host reconnaissance and outbound exfiltration. Among these, ModelScan, PickleScan, Fickling, and ModelAudit detected only 29%, 56%, 80%, and 93%, respectively. Despite runtime overhead, ModelShield has a median analysis time of 12.3 seconds, with 90% of scans completing within 35.5 seconds, making it practical for hub-scale screening.

cs.CR

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2509.10834

Landscape Analysis of Simultaneous Blind Deconvolution and Phase Retrieval via Structured Low-Rank Tensor Recovery

Xiao Liang, Zhen Qin, Zhihui Zhu, Shuang Li

原文摘要

This paper studies the landscape of the simultaneous blind deconvolution and phase retrieval (BDPR) problem via a structured low-rank tensor recovery framework. Since the complicated sensing structure makes direct landscape analysis technically challenging, we introduce a tractable tensor sensing surrogate that preserves the essential low-rank structure. We first analyze its population risk, characterize the global landscape on the unit sphere, and establish local linear convergence of Riemannian gradient descent (RGD) under mild conditions. We then extend the analysis to tensor sensing under a tensor restricted isometry property (TRIP), establishing local geometric and RGD convergence guarantees, and quantify robustness to measurement noise. Our theoretical results are further supported by extensive numerical experiments. These findings offer insights into the optimization landscape of the surrogate tensor sensing problem, thereby providing guidance for studying and solving the original BDPR problem.

eess.SP · cs.IT

查看论文
arXiv ↗PDF ↗