跳到正文
研踪PaperTrail

论文库

检索计算机领域论文,按研究方向、作者或关键词筛选。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+1112新增论文
+875其中 AI 相关
+133已有论文修订
+94新增 AI 导读

arXiv 最近完成扫描:2026-11-20 00:30 · 全站最近入库:2026-11-19 19:09各信源进度 →

统计口径与数据新鲜度

统计 2026-11-19 01:12 — 2026-11-20 01:12(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:2026-11-20 00:30最近发现新增版本:2026-11-19 19:09库内论文最新更新时间:2026-11-19 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。

27,982 已收录论文23,249 AI 相关8,559 有修订112 HF 社区发现

浏览论文库

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 24 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-20 00:28(北京时间)
新论文 · v1arXiv:2611.24100

Lost Between Mind, Model Collaboration, and Motion: Understanding and Measuring Cross-Model Action Misalignment in Commercial Robots

Yuhang Huang, Junchao Li, Boyang Ma, Minghui Xu, Kaidi Xu, Pengfei Hu, Xiuzhen Cheng, Yue Zhang

原文摘要

Embodied robots increasingly rely on large language models (LLMs) to interpret user intent and coordinate with robot-specific action models for physical interaction. This multi-model pipeline creates a critical semantic boundary across mind, model collaboration, and motion: semantic consistency may break between what the large model understands, how the small action model translates that understanding into executable commands, and what physical actions the robot ultimately performs. As a security consequence, commercial robots may execute unintended physical actions even when user inputs contain no explicit action commands. To measure this problem, we develop TARS, a black-box framework that explores the command-reasoning-action pipeline and automatically discovers and validates unintended physical behaviors. We evaluate TARS on four commercial robots across 44 experimental settings, conducting 133,400 executions. TARS uncovers widespread cross-model action misalignment and achieves up to 94% verified action coverage.

cs.RO

新论文 · v1arXiv:2611.22825

A tiny object detection dataset for high-altitude open-water search and rescue

Hui Xu, Chenfei Hou, Henglai Wei, Hao Wang, Yidong Jin, Xiulin Niu, Haoyang Zhang, Hui Zhang 等 9 位作者

原文摘要

Large aerial platforms often operate above 1.5 km to improve the efficiency of open water search and rescue (OW-SAR). However, long imaging distances, wide fields of view, and limited sensor resolution reduce objects to a few pixels, increasing their susceptibility to environmental interference and making detection harder than for conventional small objects. To our knowledge, existing datasets focus on low-altitude scenarios and do not support automatic detection at 1.5-2.0 km. We therefore present TinyWOD, a tiny object detection dataset for high-altitude OW-SAR using lifebuoys as detection targets. It comprises 11,359 manually annotated images and 19,505 annotated instances spanning diverse environments, lighting conditions, water colors, and surface ripples. Compared with existing datasets, TinyWOD contains much smaller objects and a substantially higher proportion of tiny instances; its average equivalent side length (ESL) is 12.7 pixels, with 76.96% of in stances having ESLs below 15 pixels, reflecting actual high-altitude water-rescue imaging conditions. We define APtiny, a tiny object detection metric, and benchmark mainstream and state-of-the-art networks using this and conventional metrics. All evaluated networks effectively detect tiny waterborne objects, demonstrating TinyWOD's reusability and utility. We expect this dataset to support research on tiny object detection in high-altitude OW-SAR.

cs.CV · cs.AI

新论文 · v1arXiv:2611.22503

From Relevance to Utility: Rubric Alignment in Yuanbao Search

Yu Lu, Lixin Zou, Yangqi Chen, Hui Xu, Qiaolin Xia, Hao Wang, Qianwen Ma, Reed Li 等 11 位作者

原文摘要

Agentic search demands that a search engine move beyond per document relevance ranking to assemble document sets optimized for downstream LLM reasoning. Prior work addresses this either by improving individual pipeline stages in isolation or by applying RL based selection on top of relevance trained upstream modules, leaving the full stack unaligned to utility for reasoning. We propose rubric aligned search, a framework that unifies the traditional search pipeline under a single supervision surface. For each query, we derive a rubric from coarse to fine, specifying weighted positive criteria that the document set must satisfy and negative criteria that it must avoid. From this rubric we derive two training signals. First, a lightweight labeler distills rubric grounded judgments into per document utility labels, which supervise a contrastive dual encoder for retrieval and a multi task ranker trained with four rubric aligned scoring heads. Second, a rubric conditioned evaluator produces a set level reward that drives a dual branch RL selector, combining GRPO with best rollout distillation to resolve credit assignment in subset optimization. Deployed in Yuanbao Search, the pipeline improves end to end answer quality by 8.52% over the online production baseline.

cs.IR

版本更新 · v2arXiv:2607.01191

Perceive-to-Reason:解耦感知与推理的细粒度视觉推理

Perceive-to-Reason: Decoupling Perception and Reasoning for Fine-Grained Visual Reasoning

Hongxing Li, Xiufeng Huang, Dingming Li, Wenjing Jiang, Zixuan Wang, Haolei Xu, Hanrong Zhang, Haiwen Hong 等 14 位作者

AI 速读 基于摘要

P2R 将细粒度视觉推理解耦为“先定位证据、再基于证据作答”的两阶段框架,并用角色感知的强化学习策略对齐训练。

cs.CV

查看原文摘要

Fine-grained visual reasoning remains challenging for vision-language models, especially when small but critical visual cues are buried in high-resolution images. Existing approaches rely on repeated cropping or test-time visual search to introduce local evidence, but they typically do not explicitly distinguish perception from reasoning. In this paper, we propose Perceive-to-Reason (P2R), a unified framework that formulates fine-grained visual reasoning as a two-stage process: the model first localizes question-relevant evidence as a Perceiver, and then answers the question as a Reasoner based on the annotated image and cropped regions. To better align training with this decoupled formulation, we further introduce Perception-Reasoning Alternating GRPO (PRA-GRPO), a role-aware reinforcement learning strategy that alternates between perception-focused and reasoning-focused updates using only final-answer supervision. Built on top of Qwen3-VL-Instruct-2B/4B/8B, P2R consistently improves performance across model scales. In particular, P2R-4B achieves 93.2% on V-Star, 81.9% on HR-Bench-4K, and 80.5% on HR-Bench-8K, substantially outperforming its corresponding backbone. Further experiments show that the benefits of P2R extend beyond high-resolution benchmarks to broader multimodal reasoning tasks. These results suggest that explicitly decoupling perception from reasoning provides an effective framework for fine-grained visual reasoning.

版本更新 · v2arXiv:2611.16493

ReA-Bench: Benchmarking Responsible Agent Auditing

Zipeng Ye, Shenyu Dai, Yong Xie, Kun Wang, Jiyang Guan, Zhikai Chen, Jiaqi Weng, Jialing Tao 等 9 位作者

原文摘要

As agents perform tasks autonomously in local and cloud environments, auditing their execution trajectories can help service providers identify risks and improve agent harnesses and foundation models. Our analysis of production and red-team trajectories shows that some risks are repeatedly missed across auditing configurations. Overall recall obscures which risks are missed and whether higher recall comes with more false positives on legitimate behavior. Existing benchmarks offer limited insight into these failures. Generated trajectories can contain inconsistent tool calls and responses, conspicuous risky actions, or repetitive attack patterns. Grouping results by risk type also makes it difficult to tell whether auditors overlooked evidence, accepted an unsupported conclusion, or lacked information needed to make a judgment. We introduce ReA-Bench, a diagnostic benchmark for auditing responsible agent behavior. It organizes auditing challenges into six tiers: internal safety knowledge, contradictions within trajectories, evidence retrieval in long contexts, verification of sources and reasoning, indirect steering, and missing information. It includes challenging benign cases in which actions resemble risky ones but are consistent with the task and its authorization. It also provides paired trajectory versions to test whether auditors detect the same risks more reliably when the trajectory is shortened or missing information is supplied. Most samples are drawn from real production trajectories, with scenarios manually selected and adapted to preserve realistic tasks and risks that could plausibly arise during execution.

cs.AI

版本更新 · v4arXiv:2602.02343

Why Steering Works: Toward a Unified View of Language Model Parameter Dynamics

Ziwen Xu, Chenyan Wu, Hengyu Sun, Haiwen Hong, Mengru Wang, Yunzhi Yao, Longtao Huang, Hui Xue 等 12 位作者

原文摘要

Methods for controlling large language models (LLMs), including local weight fine-tuning, LoRA-based adaptation, and activation-based interventions, are often studied in isolation, obscuring their connections and making comparison difficult. In this work, we present a unified view that frames these interventions as dynamic weight updates induced by a control signal, placing them within a single conceptual framework. Building on this view, we propose a unified preference-utility analysis that separates control effects into preference, defined as the tendency toward a target concept, and utility, defined as coherent and task-valid generation, and measures both on a shared log-odds scale using polarity-paired contrastive examples. Across methods, we observe a consistent trade-off between preference and utility: stronger control increases preference while predictably reducing utility. We further explain this behavior through an activation manifold perspective, in which control shifts representations along target-concept directions to enhance preference, while utility declines primarily when interventions push representations off the model's valid-generation manifold. Finally, we introduce a new steering approach SPLIT guided by this analysis that improves preference while better preserving utility. Code is available at https://github.com/zjunlp/EasyEdit/blob/main/examples/SPLIT.md.

cs.CL · cs.AI · cs.CV · cs.IR · cs.LG

版本更新 · v3arXiv:2603.02578

How Controllable Are Large Language Models? A Unified Evaluation across Behavioral Granularities

Ziwen Xu, Kewei Xu, Haoming Xu, Haiwen Hong, Longtao Huang, Hui Xue, Ningyu Zhang, Yongliang Shen 等 11 位作者

原文摘要

Large Language Models (LLMs) are increasingly deployed in socially sensitive domains, yet their unpredictable behaviors, ranging from misaligned intent to inconsistent personality, pose significant risks. We introduce SteerEval, a hierarchical benchmark for evaluating LLM controllability across three domains: language features, sentiment, and personality. Each domain is structured into three specification levels: L1 (what to express), L2 (how to express), and L3 (how to instantiate), connecting high-level behavioral intent to concrete textual output. Using SteerEval, we systematically evaluate contemporary steering methods, revealing that control often degrades at finer-grained levels. Our benchmark offers a principled and interpretable framework for safe and controllable LLM behavior, serving as a foundation for future research.

cs.CL · cs.AI · cs.HC · cs.LG

版本更新 · v2arXiv:2604.08541

Seeing but Not Thinking: Routing Distraction in Multimodal Mixture-of-Experts

Haolei Xu, Haiwen Hong, Hongxing Li, Rui Zhou, Yang Zhang, Longtao Huang, Hui Xue, Yongliang Shen 等 10 位作者

原文摘要

Multimodal Mixture-of-Experts (MoE) models have achieved remarkable performance on vision-language tasks. However, we identify a puzzling phenomenon termed Seeing but Not Thinking: models accurately perceive image content yet fail in subsequent reasoning, while correctly solving identical problems presented as pure text. Through systematic analysis, we first verify that cross-modal semantic sharing exists in MoE architectures, ruling out semantic alignment failure as the sole explanation. We then reveal that visual experts and domain experts exhibit layer-wise separation, with image inputs inducing significant routing divergence from text inputs in middle layers where domain experts concentrate. Based on these findings, we propose the Routing Distraction hypothesis: when processing visual inputs, the routing mechanism fails to adequately activate task-relevant reasoning experts. To validate this hypothesis, we design a routing-guided intervention method that enhances domain expert activation. Experiments on three multimodal MoE models across six benchmarks demonstrate consistent improvements, with gains of up to 3.17% on complex visual reasoning tasks. Our analysis further reveals that domain expert identification locates cognitive functions rather than sample-specific solutions, enabling effective transfer across tasks with different information structures.

cs.CV · cs.AI · cs.CL

版本更新 · v2arXiv:2611.15315

MIGRO: A Decoupled and Portable Long-Term Memory System for Language Models

Mingxi Zou, Yinghui Xu, Jiaxiang Chen, Langzhang Liang, Wei Zhu, Peng Ye, Runkai Zhao, Ruoxi Jiang 等 13 位作者

原文摘要

Language models evolve, but the observations, revisions, and context accumulated through user interactions should remain usable. Each new model must be able to retrieve past evidence, track revisions, and add new observations. We present MIGRO, a decoupled long-term memory system that treats experience as a persistent state shared across successive language models. MIGRO follows a simple principle: preserve historical evidence exactly, organize it approximately, and decide what to retrieve only when the question is known. We call this asymmetric fidelity with deferred selectivity. A shared semantic index supports retrieval, while exact records retain observations and their historical relationships. Because models differ in vocabulary and tokenization, host-specific interfaces convert selected evidence into inputs the receiving model can use. Reversible encoding reduces archival storage while enabling exact recovery. The shared state and its access interfaces let successive hosts read and extend the accumulated experience. On matched LoCoMo states, MIGRO scores 7.3% higher and uses 99.3% less private memory than the strongest quality baseline. After handoff, continued writing retains 99.9% of full-history rebuilding quality by extending the inherited state. On million-token BEAM histories, MIGRO uses 57.5% less private storage on average than the most compact baseline. MIGRO combines high answer quality, continued memory growth across hosts, and low storage cost. Code is available at https://github.com/zOUMINGXI/MIGRO.

cs.AI

版本更新 · v3arXiv:2510.10982

Catch-Only-One: Non-Transferable Examples for Model-Specific Authorization

Zihan Wang, Ethan Ma, Zhongkui Ma, Shuofeng Liu, Akide Liu, Derui Wang, Minhui Xue, Guangdong Bai

原文摘要

Recent AI regulations increasingly emphasize the need for mechanisms that preserve the utility of data for AI innovation while preventing misuse, particularly by enforcing purpose limitation in downstream AI applications. In practice, enforcing this principle remains challenging, as released data can be trivially fed into arbitrary models beyond its declared intent. Existing approaches attempt to mitigate this risk by either perturbing data or retraining models to limit unintended use. These strategies, however, offer no protection against inference by unknown or externally trained models, or fundamentally rely on control over the training or deployment. In this work, we introduce non-transferable examples (NTEs), recoded data that act as a task-level "ciphertext" decodable only by a designated model. Whereas adversarial examples exploit directions of high model sensitivity, NTEs leverage the complementary insensitive subspace. We propose a training-free, data-agnostic method that recodes data within a model-specific low-sensitivity subspace, preserving outputs for the authorized model while degrading unauthorized ones through subspace misalignment. We establish formal bounds certifying authorized-model fidelity and showing that unauthorized degradation scales with measurable spectral misalignment between models. Empirically, NTEs preserve performance across diverse vision backbones and state-of-the-art vision-language models under common preprocessing, while unauthorized models collapse even under adaptive reconstruction attacks. These results establish NTEs as a practical means to preserve intended data utility while preventing unauthorized exploitation. Our project is available at https://trusted-system-lab.github.io/model-specificity.

cs.LG · cs.AI

新论文 · v1arXiv:2611.12466

Prediction with Foundation Models under Deployment Budgets

Zhaohui Xu, Xiaofei Xu, Doudou Zhou

原文摘要

Foundation models can supply predictions for a downstream task that draw on knowledge from extensive pretraining, but they differ in task-specific accuracy and query cost. We develop a framework that improves prediction for the task under a budget on the average query cost per input, using training data to learn which models to query and how to combine their outputs with features available without querying. The first method queries the same models for every input; the second lets them vary across inputs and uses Q-learning to decide which model to query next and when to stop. For the first method, we show that several individually inaccurate outputs can jointly outperform the best single one. It attains the minimax rate when the regression function depends on the outputs only through models within the budget. Otherwise, any method within the budget can be forced to a slower rate, which the first method attains. For the second method, we show when varying the queried models across inputs strictly improves on querying the same models for every input, and bound its prediction risk and average cost in finite samples. In simulations and a multimodal laptop price application, both methods outperform existing methods.

stat.ML · cs.LG

新论文 · v1arXiv:2611.10423

Mental Imagery That Matters: Curing Latent Collapse in Multimodal Reasoning

Mi Luo, Rohan Dugad, Zihui Xue, Alex Dimakis, Kristen Grauman

原文摘要

Mental simulation, the ability to reason by constructing and updating internal visual states, is central to flexible human cognition, but remains elusive in current multimodal large language models. Latent multimodal reasoning offers a promising route toward this capability by interleaving text with continuous visual latent blocks, allowing models to maintain internal visual workspaces without decoding them into pixels. In this paper, we ask whether these latent visual states actually matter. Through a direct perturbation analysis across representative latent-reasoning MLLMs, we find that they often do not: dropping, replacing, or shuffling generated visual latents has only minor effects on the model's own final-answer probability, while equivalent perturbations to the linguistic reasoning trace are substantially more damaging. We call this failure mode Mental Simulation Collapse (MSC): the model appears to perform interleaved visual reasoning, but routes most answer-relevant computation through text. We further diagnose that MSC follows naturally from existing training objectives: visual alignment can make generated latents resemble helper-image embeddings, but it does not require the decoder to rely on them; subsequent text-only supervision further leaves the semantic role of the latent pathway unidentifiable. To address this, we introduce AnchorSim (Anchored Mental Simulation), a contrastive distillation framework that trains models to make their generated visual latents consequential. Across multiple visual reasoning benchmarks, AnchorSim improves performance, while diagnostic re-evaluation shows that visual perturbations now meaningfully affect predictions. Our results expose a fundamental failure mode of latent multimodal reasoning and demonstrate that mental simulation must be anchored to model behavior in order to matter.

cs.CV · cs.AI

版本更新 · v2arXiv:2604.01025

Fast and Accurate Probing of In-Training LLMs' Downstream Performances

Zhichen Liu, Tianle Lun, Yang Xu, Zhibin Wen, Hao An, Yulin Ou, Jianhui Xu, Hao Zhang 等 10 位作者

原文摘要

The paradigm of scaling Large Language Models (LLMs) in both parameter size and test time has pushed the boundaries of AI capabilities, but at the cost of making the traditional generative evaluation paradigm prohibitively expensive, therefore making the latency of LLM's in-training downstream performance evaluation unbearable. However, simple metrics like training loss (perplexity) are not always correlated with downstream performance, as sometimes their trends diverge from the actual task outcomes. This dilemma calls for a method that is computationally efficient and sufficiently accurate in measuring model capabilities. To address this challenge, we introduce a new in-training evaluation paradigm that uses a lightweight probe for monitoring downstream performance. The probes take the internal representations of LLM checkpoints (during training) as input and directly predict the checkpoint's performance on downstream tasks measured by success probability (i.e., pass@1). We design several probe architectures, validating their effectiveness using the OLMo3-7B's, OLMo3-32B's, and Apertus-8B's checkpoints across a diverse set of downstream tasks. The probes can accurately predict a checkpoint's performance (with avg. AUROC$>$0.75), have decent generalizability across checkpoints (earlier predicts later), and reduce the computation latency from $\sim$1 hr (using conventional generative evaluation method) to $\sim$3 min. In sum, this work presents a practical and scalable in-training downstream evaluation paradigm, enabling a more agile, informed, and efficient LLM development process.

cs.LG · cs.AI

新论文 · v1arXiv:2611.04701

Proving the Wrong Program: Miscompilations as a Threat to zkVM Guarantees

Hui Xuan Tan, Stefanos Chaliasos, Alastair F. Donaldson

原文摘要

A zero-knowledge virtual machine (zkVM) lets an untrusted prover convince a verifier that a program, typically Rust compiled to RISC-V, ran correctly. The proof covers the compiled binary, so a miscompilation can make a correct zkVM produce a valid proof of a computation that differs from the developer's source. We present the first empirical study of miscompilation as a threat to zkVM guarantees, on the production zkVMs SP1 and Jolt. We assemble 88 known rustc and LLVM miscompilations with Rust reproducers, using an LLM agent to recover Rust triggers for 60 of 124 LLVM bugs reported only as LLVM IR. Many of the bugs carry over to zkVMs: 48 of the 88 make a guest commit a wrong output on SP1 and 46 on Jolt, including the same 15 rustc bugs on both. Most bugs that do not reproduce never reach the faulty code on the zkVM's target, e.g. because they need an instruction-set extension. Existing guest programs are rarely affected by accident. Across 13 guests of 10 popular SP1 projects, the buggy and the fixed compiler produce byte-identical binaries in 447 of the 552 guest-bug pairs that build, and the other 105 have the same code or agree on every input we ran. An attacker who contributes code to a guest, however, can trigger these bugs on purpose. For about $40 per project, an LLM agent found patches that look like ordinary optimisations, preserve the behaviour of the guest's source, and expose a known bug in 12 of 366 project-bug pairs, across 11 of 20 projects. In four cases the guest commits a wrong output, including an Ethereum client that rejects a valid block, and in the other eight it fails on a valid input. Reviewing or formally verifying the zkVM or the guest's source cannot catch such a patch, because both are correct. zkVM projects should treat their compiler as part of their trusted base and track the known bugs of their toolchain.

cs.PL · cs.CR

版本更新 · v2arXiv:2611.02791

PGuard: A Native, Streaming, and Plug-in Guardrail for Multimodal Large Language Models

Junyu Lin, Mengjie Wu, Yonghao Dong, Yuchuan Fu, Jinfeng Li, Xiaodan Li, Xiaohan Yuan, Yicheng Liu 等 14 位作者

原文摘要

As streaming generation becomes the dominant serving paradigm for both text-only and multimodal large language models, ensuring safety throughout the generation process has become a critical challenge. We introduce PGuard, a native, streaming, and plug-in guardrail that transforms safety monitoring from an external serving dependency into an integrated capability of the backbone model. PGuard directly leverages the backbone's intermediate representations for token-level safety detection, without altering its original inference pathway or generation behavior. Specifically, dynamic layer gating selectively aggregates residual-stream representations across layers to extract safety-relevant signals during both prefill and autoregressive decoding. To balance detection accuracy and computational efficiency, PGuard supports multiple representation-modeling architectures, including a lightweight stateless variant and sequence-aware variants built on multi-scale causal EMA-attention and GRU-based recurrent modules. Its design also accounts for the execution characteristics of LLM serving frameworks, such as SGLang, optimizing computation and data access to minimize additional runtime and synchronization overhead. As a result, PGuard incurs only 0.82-2.26% end-to-end overhead on the hundred-billion-parameter backbone Qwen3.5-122B-A10B served by SGLang. Extensive evaluations across more than 30 benchmarks spanning text and multimodal safety, jailbreaks, cyber and CBRN risks, multilingual scenarios, and real-world traffic demonstrate state-of-the-art performance among native guardrails. On text queries with Qwen3.5-9B, PGuard achieves an average F1 score of 0.836, approaching dedicated external guardrails, while supporting deployment-time configuration of category-specific thresholds and enforcement actions.

cs.CL

版本更新 · v5arXiv:2607.22877

Towards Trustworthy Physical Intelligence: From Theory to Practice Across Life Cycle

Yang Wang, Hongxuan Liu, Xinghui Xu, Arjun Menon, Xiaoran Cai, Yunyu He, Alex Tarvo, Jingzong Zhou 等 43 位作者

原文摘要

Physical intelligence refers to intelligence systems that understand, reason about and act in accordance with the physical world and its underlying laws, dynamics, and constraints. Unlike conventional AI systems, physical intelligence interacts continuously with uncertain physical environments, and its actions produce consequences that are physically irreversible. As existing trustworthy AI frameworks have been developed primarily for digital AI systems, they do not fully capture the distinctive challenges of Physical Intelligence, such as physical safety, cyber-physical security, and physical manufacturing process. To address this gap, our paper develop a foundation for understanding, operationalizing, and governing trustworthy physical intelligence. First, we characterize the fundamental capabilities of physical intelligence. Second, we investigate the role of physical laws and principles in AI and analyze the associated challenges of physical intelligence. Third, we provide a unified structure of the end-to-end physical intelligence life cycle across five core stages and introduce the Trustworthy Physical Intelligence Operationalization (T-PAIO) framework. Fourth, we formulate the Trustworthy Physical Intelligence (T-PAI) Principles as a theoretical framework that synthesizes key trustworthiness dimensions and establishes a conceptual foundation for the governance of physical intelligence systems.

cs.AI · cs.HC · cs.RO

新论文 · v1arXiv:2611.02310

Detect, Repair and Audit: A Hierarchical Multi-Agent Defense Against Adversarial Attacks in Energy Forecasting

Zewei Shi, Ruoxi Sun, Minhui Xue, Feng Liu, Hua Ma, Xingliang Yuan, Surya Nepal

原文摘要

Accurate electricity-demand forecasting is essential for the modern energy infrastructure. Recently, Times-Series Large Language Models (TS-LLMs) have emerged as a promising approach to forecasting more accurate demand from historical consumption data in zero- and few-shot settings, helping energy operators effectively plan supply and allocate energy resources. However, these models are vulnerable to compromised sensor inputs that can distort forecasts and lead to erroneous market orders or unnecessary load shedding. Existing defenses focus primarily on end-device forecasting, overlooking the distinct data-drift, latency, and audit requirements across energy infrastructure. To address this gap, we extract from major energy standards and regulatory frameworks a three-layer conceptual model that captures the security postures for each layer. Guided by this model, we develop a Multi-Agent Reinforcement Learning (MARL) detector coupled with signal repair at local control centers, designed to mitigate distributional data drift through complementary feature-based detection and clean-reference-calibrated repair while meeting latency budgets. We further deploy an evidence-grounded LLM auditing agent with human-in-the-loop validation for audit-critical Virtual Power Plants (VPPs). In our evaluation, the best MARL configuration achieves an F1 score above 0.91 across all dataset-attack pairs, while signal repair reduces adversarial-to-clean input Relative Squared Error (RSE) by up to 83%. The few-shot auditor detects 77.78% of attacks in the Sherlock dataset. On the evaluated hardware, detection and repair require 7.95 ms on average, with a worst-observed latency of 14.50 ms, remaining within the local-center latency budget informed by IEEE 1646.

cs.CR

新论文 · v1arXiv:2610.35188

V$^3$AST-Bench: Verifiable, Valid, and Versatile Safety Evaluation for Coding Agents

Weitao Feng, Hengheng Xiong, Zihao Zeng, Jun Chen, Jiyang Guan, Xiaofeng Wang, Jialing Tao, Hui Xue 等 9 位作者

原文摘要

Large language models increasingly act through code, modifying repositories, executing programs, and interacting with services in agent form. Their actions introduce risks that cannot be assessed from generated responses alone. Safety evaluation must verify actual consequences, distinguish harm avoidance from task incapability, and cover realistic workflows. We introduce V$^3$AST-Bench, a benchmark for Verifiable results, Valid interpretation, and Versatile risk coverage of Agent Safety. It comprises 212 risk tasks spanning 18 categories. Dedicated runtime environments verify harm and task outcomes through executable checks over states, records, and artifact behavior, while paired capability twin tasks verify workflow competence. Our experiments span 51 combinations of 11 models and five harnesses, spanning 86,496 trials and over 9 billion tokens of model interaction. Results show that V$^3$AST-Bench remains challenging even for frontier coding agents, as Claude Opus 4.8 with Claude Code still causes verified harm on 20.9% of tasks. Across model-harness combinations, agents with comparable competence exhibit markedly different safety outcomes, showing that task capability alone does not guarantee safe execution. More importantly, even explicit recognition of risks does not reliably prevent harmful actions, revealing a critical gap between risk awareness and safe execution. These findings highlight the need to move beyond risk recognition toward enforcing safety throughout execution.

cs.SE · cs.AI

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.1438ACL Anthology · ACL 2026

Seeing but Not Thinking: Routing Distraction in Multimodal Mixture-of-Experts

Haolei Xu Zhejiang University, Haiwen Hong Alibaba Group, Hongxing Li, Rui Zhou, Yang Zhang, Longtao Huang Alibaba Group, Hui Xue, Yongliang Shen Zhejiang University 等 10 位作者

原文摘要

Multimodal Mixture-of-Experts (MoE) models have achieved remarkable performance on vision-language tasks. However, we identify a puzzling phenomenon termed Seeing but Not Thinking: models accurately perceive image content yet fail in subsequent reasoning, while correctly solving identical problems presented as pure text. Through systematic analysis, we first verify that cross-modal semantic sharing exists in MoE architectures, ruling out semantic alignment failure as the sole explanation. We then reveal that visual experts and domain experts exhibit layer-wise separation, with image inputs inducing significant routing divergence from text inputs in middle layers where domain experts concentrate. Based on these findings, we propose the Routing Distraction hypothesis: when processing visual inputs, the routing mechanism fails to adequately activate task-relevant reasoning experts. To validate this hypothesis, we design a routing-guided intervention method that enhances domain expert activation. Experiments on three multimodal MoE models across six benchmarks demonstrate consistent improvements, with gains of up to 3.17% on complex visual reasoning tasks. Our analysis further reveals that domain expert identification locates cognitive functions rather than sample-specific solutions, enabling effective transfer across tasks with different information structures.

cs.CL · cs.CV

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.1443ACL Anthology · ACL 2026

How Controllable Are Large Language Models? A Unified Evaluation across Behavioral Granularities

Ziwen Xu Zhejiang University, Kewei Xu Zhejiang University, Haoming Xu, Haiwen Hong Alibaba Group, Longtao Huang Alibaba Group, Hui Xue, Ningyu Zhang Zhejiang University, Yongliang Shen Zhejiang University 等 11 位作者

原文摘要

Large Language Models (LLMs) are increasingly deployed in socially sensitive domains, yet their unpredictable behaviors, ranging from misaligned intent to inconsistent personality, pose significant risks. We introduce SteerEval, a hierarchical benchmark for evaluating LLM controllability across three domains: language features, sentiment, and personality. Each domain is structured into three specification levels: L1 (what to express), L2 (how to express), and L3 (how to instantiate), connecting high-level behavioral intent to concrete textual output. Using SteerBench, we systematically evaluate contemporary steering methods, revealing that control often degrades at finer-grained levels. Our benchmark offers a principled and interpretable framework for safe and controllable LLM behavior, serving as a foundation for future research.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.1463ACL Anthology · ACL 2026

Why Steering Works: Toward a Unified View of Language Model Parameter Dynamics

Ziwen Xu Zhejiang University, Chenyan WU Zhejiang University of Technology, Hengyu Sun, Haiwen Hong Alibaba Group, Mengru Wang, Yunzhi Yao Zhejiang University, Longtao Huang Alibaba Group, Hui Xue 等 12 位作者

原文摘要

Methods for controlling large language models (LLMs), including local weight fine-tuning, LoRA-based adaptation, and activation-based interventions, are often studied in isolation, obscuring their connections and making comparison difficult. In this work, we present a unified view that frames these interventions as dynamic weight updates induced by a control signal, placing them within a single conceptual framework. Building on this view, we propose a unified preference-utility analysis that separates control effects into preference, defined as the tendency toward a target concept, and utility, defined as coherent and task-valid generation, and measures both on a shared log-odds scale using polarity-paired contrastive examples. Across methods, we observe a consistent trade-off between preference and utility: stronger control increases preference while predictably reducing utility. We further explain this behavior through an activation manifold perspective, in which control shifts representations along target-concept directions to enhance preference, while utility declines primarily when interventions push representations off the model’s valid-generation manifold. Finally, we introduce a new steering approach guided by this analysis that improves preference while better preserving utility.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.1585ACL Anthology · ACL 2026

GameplayQA: A Benchmarking Framework for Decision-Dense POV-Synced Multi-Video Understanding of 3D Virtual Agents

Yunzhe Wang University of Southern California, Runhui Xu University of Southern California, Kexin Zheng University of Southern California, Tianyi Zhang, Jayavibhav Niranjan Kogundi, Soham Hans University of Southern California, Volkan Ustun USC Institute for Creative Technologies, University of Southern California

原文摘要

Multimodal LLMs are increasingly deployed as perceptual backbones for autonomous agents in 3D environments, from robotics to virtual worlds. These applications require agents to perceive rapid state changes, attribute actions to the correct entities, and reason about concurrent multi-agent behaviors from a first-person perspective, capabilities that existing benchmarks do not adequately evaluate. We introduce GameplayQA, a framework for evaluating agentic-centric perception and reasoning through video understanding. Specifically, we densely annotate multiplayer 3D gameplay videos at 1.22 labels/second, with time-synced, concurrent captions of states, actions, and events structured around a triadic system of Self, Other Agents, and the World, a natural decomposition for multi-agent environments. From these annotations, we refined 2.4K diagnostic QA pairs organized into three levels of cognitive complexity, accompanied by a structured distractor taxonomy that enables fine-grained analysis of where models hallucinate. Evaluation of frontier MLLMs reveals a substantial gap from human performance, with common failures in temporal and cross-video grounding, agent-role attribution, and handling the decision density of the game. We hope GameplayQA stimulates future research at the intersection of embodied AI, agentic perception, and world modeling.

cs.CL · cs.AI · cs.MA · cs.CV

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.1913ACL Anthology · ACL 2026

LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety

Junxiao Yang, Haoran Liu, Jinzhe Tu, Jiale Cheng, Zhexin Zhang, Shiyao Cui, Jiaqi Weng, Jialing Tao Alibaba Group 等 12 位作者

原文摘要

Large language models (LLMs) have demonstrated better safety performance in high-resource languages than in low-resource languages. We attribute this issue as a mismatch gap between language-agnostic semantic understanding ability and language dominant safety alignment biased toward high-resource languages. Based on above insights, we empirically identify the semantic bottleneck in LLMs: intermediate layers in which the geometry of model representations is governed primarily by shared semantic content rather than language identity. Then, we propose Language-Agnostic Semantic Alignment (LASA), which anchors safety alignment directly in semantic bottlenecks. Experiments show that LASA substantially improves safety across all languages: average attack success rate (ASR) drops from 24.7% to 2.8% on LLaMA-3.1-8B-Instruct and remains within 3–4% across Qwen2.5 and Qwen3 Instruct models (7B–32B). Besides, our analysis and method offer a representation-level perspective on LLM safety, suggesting that safety alignment requires anchoring safety understanding not in surface text, but in the model’s language-agnostic semantic space.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.506ACL Anthology · ACL 2026

Reinforcement Learning on Pre-Training Data

Siheng Li, Kejiao Li Tencent, Zenan Xu, Guanhua Huang Tencent, Kun Li, Haoyuan Wu, Wujiajia Tencent, Zihao Zheng 等 34 位作者

原文摘要

Recent progress in large language models (LLMs) is largely driven by scaling training compute through either pre-training with next-token prediction (NTP) or post-training with reinforcement learning (RL). The former contributes to learning broad knowledge and skills from general data, while struggling with data inefficiency and catastrophic forgetting in continual learning settings. The latter incentivizes reasoning capabilities with strong generalization, but is constrained by limited data availability due to its reliance on human annotation. To alleviate these issues, we propose Reinforcement Learning on Pre-Training data (RLPT), which combines the advantages of learning from general data and RL. In particular, RLPT derives reward signals directly from general text data through a next-segment reasoning objective, rewarding the policy for correctly predicting next text segments conditioned on the prefix text. Experiments across multiple benchmarks and models demonstrate the effectiveness of . For example, RLPT yields substantial improvements in continual pre-training (+4.6\%) and provides a strong foundation for post-training (+3.4\%) on Qwen3-8B-Base.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表