标题:Language Models Learn to Improve Language Models
来源:arXiv
链接:https://arxiv.org/abs/2610.23344
摘要:Large language models (LLMs) are increasingly used to automate AI research, including language-model (LM) development, offering a potential path toward recursive self-improvement. However, turning experimental feedback into a learned ability to propose effective improvements across models faces two challenges. First, training such an ability with real feedback is expensive: reinforcement learning (RL) requires evaluating tens of thousands of proposals, each potentially requiring a full model-training run. Even at ten GPU-hours per run, reward collection alone would consume hundreds of thousands of GPU-hours, motivating a cheaper surrogate. Second, surrogate feedback can become unreliable as policy optimization favors overestimated proposals when the proposal distribution shifts. To address
标题:Reward Engine: Reward-Centric Self-Evolution for Open-Ended Tool Agents
来源:arXiv
链接:https://arxiv.org/abs/2610.22915
摘要:Open-ended tool agents can complete the same objective through different tool paths, and their quality depends jointly on the interaction trace, environment state, and final artifacts, so such tasks rarely admit an executable ground-truth reward. The central challenge is therefore to construct a reward from existing experience that exposes the current policy's failures while remaining stationary during optimization. We introduce Reward Engine, which, like a self-motivated student, first works out what success requires and then optimizes toward it. From the task and initial trajectories, it builds a RewardPack of testcases for required correctness and rubrics for graded quality, each executed as a check that returns a judgment with evidence. The frozen RewardPack then guides the agent to re
标题:Self-Building Labs
来源:arXiv
链接:https://arxiv.org/abs/2610.23826
摘要:Self-driving laboratories (SDLs) accelerate scientific discovery by automating experimental execution and optimization. AI agents can increasingly propose new experiments, but existing SDLs can only run the procedures they were built for. Adapting them to a new experiment still takes extensive manual engineering, which makes laboratory flexibility the bottleneck for scaling discovery. We introduce the self-building laboratory (SBL), which designs, configures, and operates its own laboratory setup given a high-level scientific goal. SBL starts from whatever already exists, from an empty room to a working SDL, and builds the missing capabilities: the lab layout, its digital twin, robot skills, and experimental workflows. We demonstrate the system working across different robots and setups, s
标题:DiscoveryRSI: Recursively Learning to Discover Valuable Hypotheses
来源:arXiv
链接:https://arxiv.org/abs/2610.23839
摘要:Scientific discovery requires developing worthwhile hypotheses into practical investigations, yet learning from this evolving research process remains challenging. We introduce an operational definition of valuable and testable research hypotheses: evidence-grounded claims about consequential uncertainties, coupled with feasible investigations, explicit decision criteria, and alternative explanations. Our DiscoveryRSI framework develops hypothesis trees through multi-agent evolutionary search, compiles selected investigations into executable Harbor tasks, and recursively learns through three mechanisms: hypothesis credit optimization attributes scientific assessments to documented research contributions; harness replay identifies productive strategies and recurring failures to refine resea
[合并自 2026-10-20 discoveryrsi-agent]
标题:直播中|Claude垫底,GPT表现不佳,PhAI Labs用DiscoveryRSI探索科研Agent自我进化
来源:机器之心(微信公众号)
链接:http://mp.weixin.qq.com/s?__biz=MzA3MzI4MjgzMw==&mid=2651068408&idx=1&sn=3d762a23bad4d8ea4a1ddd3d23e6ae50&chksm=850941ff4e8ad6238d752c8c8c5585441fa85e50b4a91a1d86b0d0eb6bc08e87e902db77512c&scene=0&xtrack=1#rd
摘要:
[合并自 2026-10-21 discoveryrsi]
https://mp.weixin.qq.com/s/y6xAtJyVbI0uGQ_o2zZ74g
标题:Darwin-Claw: Environment-Adaptive Experience Consolidation for Recursively Self-Evolving Agents
来源:arXiv
链接:https://arxiv.org/abs/2610.23574
摘要:While Large Language Model (LLM) agents have demonstrated remarkable capabilities across complex tasks, they predominantly operate in isolated sessions, lacking the ability to continually learn from historical interactions or adapt to user-specific environments. Existing self-evolution paradigms often indiscriminately inject raw interaction trajectories into memory or skill modules, leading to obscured behavioral patterns and compromised update controllability. Furthermore, as foundational LLM reasoning improves, the bottleneck for agentic evolution shifts from accumulating generic knowledge to effectively managing situational experience. To address this, we propose Darwin-Claw, an environment-adaptive self-evolving framework that systematically distills episodic experiences to drive the c
标题:RL-Guided Temporal Localization for Dual-Channel Retrieval in Long-Horizon Agent Memory
来源:arXiv
链接:https://arxiv.org/abs/2610.23598
摘要:Time-aware retrieval is crucial for long-horizon LLM agents to effectively leverage ever-growing long-term memory. Existing methods focus on lexical or semantic similarity, leading to target memories being crowded out by temporally mismatched yet semantically similar candidates. Even time-aware approaches inject temporal cues only during memory organization or answer generation, overlooking temporal constraints at the retrieval stage, where candidate selection occurs - the root cause of the displacement. Inspired by the temporal contiguity effect in human episodic memory, we propose TIDER (Temporal Interval-driven Dual-channel Evidence Retrieval), a time-aware agentic retrieval framework. Central to TIDER is a reinforcement learning (RL)-trained temporal interval locator that infers the te
标题:GLOBE: Evolving Structured Agentic Memory for E-commerce Product Optimization in Generative Search
来源:arXiv
链接:https://arxiv.org/abs/2610.23709
摘要:Optimizing product listings for generative search requires language agents not only to rewrite effectively but also to continually self-improve by retaining successful strategies across diverse categories. However, existing approaches often treat structured listings as free-form text, leaving effective strategies implicit in prompts, search trajectories, or rewritten outputs rather than accumulating them as a reusable agentic memory. To address this limitation, we introduce GLOBE, an agentic framework that drives continual self-improvement by storing optimization knowledge in an evolving handbook of explicit, field-scoped behaviors. In each episode, an Editor applies relevant behaviors and receives rank-based feedback, which a Reflector converts into behavior updates. Crucially for dependa
标题:WebCoder: Scaling General Frontend Development with Artifact-Grounded Agentic Reinforcement Learning
来源:arXiv
链接:https://arxiv.org/abs/2610.23104
摘要:Real-world frontend development requires agents to compose reproduction, generation, modification, and repair over an evolving application state. Learning a general policy for this process requires translating heterogeneous execution feedback into task-aligned rewards and effective credit assignment across long sequences of interdependent actions. We introduce WebCoder, a general frontend development agent trained with artifact-grounded agentic reinforcement learning. Our key idea is to assess diverse development activities through their effects on a shared executable artifact and its satisfaction of task requirements. We build Webdev-World, an executable environment that integrates code editing, application execution, browser inspection, and on-demand skills while preserving project state
标题:Tree-Aware Posterior Distillation for Long-Horizon Agentic Reinforcement Learning
来源:arXiv
链接:https://arxiv.org/abs/2610.23181
摘要:Reinforcement learning with verifiable rewards has become a scalable post-training paradigm for large language models. However, training agents over long-horizon interactions presents two challenges: using only terminal rewards makes turn-level credit assignment sparse and high-variance, while sequential mini-batch updates on a fixed rollout batch introduce policy staleness as the current policy drifts from the behavior policy. Existing approaches obtain denser credit via learned critics, process reward models, or on-policy distillation, but they rely on a value network, step-level annotations, or a stronger external teacher that increase training complexity. Meanwhile, importance sampling and clipping mitigate policy drift, but they do not specify a finite target that stops sequential upd
标题:OfficeTown: Generative Organizations for Evolving Agent Intelligence
来源:arXiv
链接:https://arxiv.org/abs/2610.23194
摘要:AI agents increasingly use files and software tools to produce professional deliverables. Such work draws on materials, decisions, and feedback accumulated across assignments. We introduce OfficeTown, a generative simulation framework for professional organizations in which tasks and their context arise from continuing organizational work. From a natural-language description of an organization's purpose and activities, OfficeTown initializes an organization whose members coordinate work, delegate bounded assignments to assistant agents, and review their deliverables. Artifacts, decisions, and unresolved matters persist and shape subsequent work, while responsibilities and permissions determine what each member can see and do. We curate work requests, their context, and review feedback into
标题:Verbalising Model's Computation for Chain-of-Thought Faithfulness Detection
来源:arXiv
链接:https://arxiv.org/abs/2610.23275
摘要:Current Chain-of-Thought (CoT) verification methods detect reasoning faithfulness from model outputs or static activations, but are limited in explaining how unfaithfulness develops during generation. We introduce VERITAS to verbalise the model's computation: how different reasoning anchors emerge, transfer across layers, and surface in the output. VERITAS adopts the layer-wise Jacobian projection to produce a verbalised account of the model's internal reasoning evolution at each generated CoT token. By training a classifier on the extracted features along the traces, we show that these structured features provide insightful perspectives on how and where reasoning diverges toward unfaithfulness. Experiments on four datasets from FaithCoT-Bench and BonaFide show that VERITAS achieves a sign
标题:Response-Informed Value Estimation for Cost-Aware Multi-Turn LLM Routing
来源:arXiv
链接:https://arxiv.org/abs/2610.23239
摘要:Multi-turn LLM routing selects a model at every turn before its response is available, yet task success is revealed only when the interaction ends, leaving each intermediate choice without a label of its own. Logged trajectories, however, retain the response each call produced, which exposes what each turn required and how well the chosen model served it, evidence that outcomes alone hide. We introduce RIVET (Response-Informed Value Estimation for Turn-Level Routing), which exploits this post-call evidence during training while keeping routing strictly pre-call at deployment. RIVET jointly trains two value estimators over a shared state-model representation: a deployable Routing Value Estimator that scores candidate models from pre-call information, and a training-only Response-Informed Va
标题:MetaCurate: Evolving Curation Filters for High-Quality Agentic Coding Data
来源:arXiv
链接:https://arxiv.org/abs/2610.23134
摘要:Supervised fine-tuning on agent trajectories is an effective way to improve coding agents, but selecting useful training examples requires more than identifying successful runs. A resolved patch may follow inefficient reasoning or unnecessary tool use, while a failed run may still contain useful intermediate steps. Training pipelines therefore rely on rules and qualitative assessments to filter trajectories, yet designing these filters typically requires substantial human expertise, and validating improvements through repeated training and evaluation is expensive. We introduce MetaCurate, a training-free framework that evolves executable trajectory curation filters using pairwise LLM judgments. Disagreements between the filter's rankings and the judge's preferences provide targeted feedbac
标题:DrugHarness: An Evolvable Agent Harness for Computational Drug Discovery
来源:arXiv
链接:https://arxiv.org/abs/2610.23791
摘要:Computational drug discovery requires agents to compose scientific operations and adapt to changing evidence and constraints. Here we introduce DrugHarness, an evolvable agent harness coupling composable scientific execution with conditional skill induction. Its environment integrates over 80 scientific operations, biomedical resources and task-specific code, preserving associations among candidates, results and artifacts. From execution trajectories and outcome feedback, DrugHarness induces action hypotheses with observable applicability conditions, recommended actions and supporting evidence. An independent review agent evaluates candidate skills derived from these hypotheses through source verification and, when needed, targeted execution trials. Admitted skills guide subsequent tasks a
标题:Algorithm Modelization: Transforming Optimization Algorithms into Evolvable Intelligence
来源:arXiv
链接:https://arxiv.org/abs/2610.23161
摘要:Can algorithms be learned and evolved as models, rather than repeatedly redesigned as programs? We explore this question by introducing Algorithm Modelization, a new paradigm that turns iterative algorithms into trainable models. Its central idea is to regard algorithm execution trajectories as computational experience: by learning how algorithmic states evolve over iterations, a model can internalize the knowledge embodied in existing algorithms, reproduce their computational behaviors, and further evolve beyond them through training. As a first realization of this paradigm, we modelize iterative optimization as an autoregressive process, where preceding generations constitute the computational context for generating the next population. We develop a dimension-independent population-tempo
标题:Beyond Success Rate: Interventional Evaluation of Memory in Vision-Language-Action Systems
来源:arXiv
链接:https://arxiv.org/abs/2610.23372
摘要:When we say that a robot uses memory, does that memory actually play a role? Success-rate ablations alone cannot establish the use of task-relevant facts: they change information and computation together, while aggregate rates can hide offsetting gains and losses. We propose an Interventional Memory Evaluation Protocol (IMEP) for vision-language-action (VLA) systems to address four questions: (1) Does memory provide a net benefit? (2) Under what conditions does memory help? (3) Do changes to memory alter task outcomes? (4) Are these effects specific to task-relevant facts? At the retrieval interface, we compare target-fact substitutions with fact-preserving replacements, minimal numerical perturbations, and exact replicas, while assessing decision correctness separately from motor executio
标题:From Memory Hits to Answerable Evidence: Query-Conditioned Memory Compilation for Long-Term Agent Memory
来源:arXiv
链接:https://arxiv.org/abs/2610.23313
摘要:Long-term agent memory requires recovering answer-relevant evidence dispersed across turns and sessions. Existing systems increasingly use structured memories and adaptive retrieval, yet often pass retrieved or expanded units directly to the reader, coupling memory addressing with evidence presentation and leaving the final context redundant or incomplete. To address this gap, we propose MemCompile, a query-conditioned evidence compilation framework for long-term conversational memory. Given a query, the Address Plane locates relevant support through compact source-linked descriptors and capacity-aware Cells, while a normalization boundary resolves heterogeneous retrieval hits to stable SourcePacket identities. Its Evidence Plane then allocates a bounded evidence budget across complementar
标题:Beyond Marginal Influence: Interaction-Aware Memory Selection for LLM Agents
来源:arXiv
链接:https://arxiv.org/abs/2610.23266
摘要:LLM-agent memory selection often uses retrieval rankings based on record relevance or individual influence. Such rankings cannot identify sufficient subsets under redundant or jointly necessary evidence. We study per-decision memory sufficiency: selecting a small subset that preserves full-memory behavior. To separate search error from model error, we introduce JointCoreBench, a benchmark with nine interaction families, six workflow domains, natural-language decisions, and exact sufficient-set oracles, and JointCore, a relevance cost-ordered group-deletion baseline that tests blocks before individual records. On 6,912 exact conditions, cost-ordered JointCore recovers a minimum-cost sufficient set in 100.0% of cases, reduces subset evaluations from 30.00 to 11.13 (62.9%), and achieves 87.8%
同专题其余 3 条
★3腾讯混元开源智能体记忆框架 Hy Memory 2.0AI工具集日报 · 长期记忆 · 代码 · 原文腾讯混元开源 Hy Memory 2.0、Hy-Memory-30B-A3B 专用模型及 MemHub 本地应用,为聊天、工作和代码场景提供可跨工具同步的长期记忆能力。
研究针对有限记忆库无法保存所有事实、共享门控策略难以适应不同用户复用模式的问题,探索事实级个性化记忆筛选。作者提出基于检索的标注策略,根据事实是否在后续交互中被检索并用于生成回答来构造反馈标签。在线门控被形式化为带延迟单侧反馈的上下文 Apple Tasting 问题,Warm-Start UCB 利用历史数据离线初始化,并持续适应当前用户。理论上,该方法在延迟窗口不超过总轮数平方根量级时具有高概率遗憾上界;实验中在不同记忆预算和历史长度下均优于基线。
用途与启示
为容量受限的智能体长期记忆提供按用户动态保留事实的策略,减少低价值信息占用。
展示如何利用事实的后续检索与回答使用情况构造可观测反馈,降低记忆监督数据获取成本。
将离线历史数据与在线用户适应结合,适用于用户行为反馈稀疏且存在延迟的个性化智能体。
理论遗憾界可为记忆预算、反馈延迟窗口及特征设计提供分析依据。
📝 原始笔记(逐字保留)
标题:Learning Personalized Memory Gating from Delayed One-Sided Feedback
来源:arXiv
链接:https://arxiv.org/abs/2610.23812
摘要:Long-term memory enables language-model agents to personalize future interactions, but a finite memory bank cannot preserve every extracted fact. Existing memory gating policies are typically shared across users and therefore fail to account for user-specific patterns of future reuse. In this paper, we study memory gating at the fact level. We first propose a retrieval-based labeling strategy that labels each fact by whether it is later retrieved and incorporated into responses. We then formulate online memory gating as contextual apple tasting with delayed one-sided feedback, and propose Warm-Start UCB, which combines offline learning from existing data with online adaptation to the current user. Our theoretical analysis establishes a high-probability regret upper bound of $\tilde{O}(Tε\w
标题:SAFE-Mem: Secure Agentic Memory with Fine-Grained Authority and Governance-Aware Retrieval
来源:arXiv
链接:https://arxiv.org/abs/2610.22959
摘要:Large language model (LLM) agents increasingly rely on persistent memory to support long-term reasoning and personalized interactions. However, information collected from external sources can introduce security risks when untrusted claims (i.e., atomic statements) are later treated as authorized preferences, policies, or permissions. Existing defenses typically rely on content-based filtering or coarse-grained provenance tracking, which either fail to prevent unauthorized memory use or block legitimate information, sacrificing memory utility. To address these limitations, we propose SAFE-Mem, a fine-grained memory governance framework that improves security while preserving legitimate memory use. First, SAFE-Mem annotates each claim with three governance labels (i.e., kind, authority, and
★2MARS:面向会话级 LLM 助手的记忆感知路由arXiv · 模型路由 · 论文MARS 以会话状态统一协调记忆构建与模型选择,在长期交互中实现回答质量、记忆利用和 API 成本之间的权衡。
MARS 将会话状态作为记忆策略与在线模型路由器之间的共享控制接口,避免二者相互独立决策。Query and Topic Manager(QTM)提取紧凑且与决策相关的状态,据此为每轮交互选择适当的记忆视图和回答模型。该框架在 RealMem 与 MultiChallenge 上以每项 16 种系统配置进行评估,覆盖长历史记忆交互和结构化上下文干扰场景。在 RealMem 上,MARS 仅使用全上下文 GPT-4o 生成 API 成本的 14.5%,便保留其 98.2% 的问答表现,并取得最强的记忆召回与检索成绩;在 MultiChallenge 上则以较低成本接近最强静态系统。
用途与启示
为长期运行的助手提供记忆检索与模型路由联合优化方案,减少无关历史信息造成的干扰。
表明会话状态可作为统一控制信号,动态平衡回答质量、记忆暴露范围与调用成本。
可用于构建按任务难度和上下文需求选择不同能力、不同成本模型的会话级智能体系统。
提示模型路由评估应计入记忆构建及控制器调用成本,而不能只比较回答模型的费用。
📝 原始笔记(逐字保留)
标题:MARS: Memory-Aware Routing for Session-Level LLM Assistants
来源:arXiv
链接:https://arxiv.org/abs/2610.23583
摘要:Long-term LLM assistants must repeatedly decide what prior context to expose and which model should answer as an interaction evolves. Existing memory systems and model routers usually make these decisions separately, although memory construction changes the evidence, distractors, and difficulty of the resulting input. We introduce MARS, a memory-aware routing framework that uses session state as a shared control interface between memory construction and model selection. A Query and Topic Manager (QTM) estimates compact decision-relevant state that guides a fixed state-conditioned memory policy and an online router, enabling each turn to receive an appropriate memory view and a suitable model under a quality-memory-cost trade-off. We evaluate MARS on RealMem and MultiChallenge with 16 syste
标题:The Cognitive Cost of LLM Writing: A Reading-Time Comparison with Human Text
来源:arXiv
链接:https://arxiv.org/abs/2610.23715
摘要:Although linguistic differences between human-written and Large Language Model (LLM)-generated texts have been extensively studied, their implications for the cognitive effort involved in reading remain less explored. In psycholinguistics, reading time is commonly used as an indirect measure of cognitive processing effort. Building on this perspective, we investigate whether human-written texts and their LLM-generated rewrites exhibit differences in predicted reading time that may reflect differences in processing demands for readers. To this end, we introduce RoBERTa-RT, a model trained on human eye-tracking data to predict word-level reading times. We evaluate the model on both internal and external test sets, construct a paired dataset of 496 human-written articles and their LLM-generat
★2AI 主导与人类主导的聊天访谈质量比较arXiv · 交互式评测 · 论文一项随机实验表明,本地部署的 7B 开放权重模型可在相近访谈时长内提出更多问题并获取更多文本,初步验证了 AI 规模化开展定性访谈的可行性。
研究对 AI 主导和人类主导的同步文本访谈进行了直接实证比较,41 名大学生被随机分配给本地 AI 系统或经过培训的非专家访谈员。两组访谈时长中位数相近,但 AI 访谈员提出的问题数量超过人类的两倍,并使每场访谈的受访者文本量增加约 65%。人类访谈员获得的单题回答更长,但该差异未通过多重检验校正;两组回答几乎都切题,具体性方面未检测到显著差异。实验采用本地部署的 7B 开放权重模型,有利于数据安全、GDPR 合规和科学可复现性,但较小样本量限制了结论的确定性。
用途与启示
为大规模定性数据采集提供 AI 访谈员的初步实证依据。
提示评估 AI 访谈系统时应同时考察问题数量、回答长度、具体性、相关性和跳题行为。
本地开放权重模型可降低敏感访谈数据外泄风险,并增强合规性与实验可复现性。
后续需要在更大、更具代表性的样本和不同访谈主题上验证结果。
📝 原始笔记(逐字保留)
标题:Can AI interview? Comparing AI-led to Human-led Chat Interviews
来源:arXiv
链接:https://arxiv.org/abs/2610.23149
摘要:Large language model chatbots have been proposed as AI-interviewers for collecting qualitative interview data at scale, potentially bridging social science's longstanding divide between the breadth of surveys and the depth of qualitative interviews. This paper provides the first direct empirical comparison of data quality in AI-led and human-led interviews. In a randomized experiment (N = 41), university students were interviewed via synchronous text chat by either trained non-expert interviewers or a locally hosted AI system; we evaluate exposure, specificity, and relevance. Median interview duration was similar across conditions, but AI-interviewers asked more than twice as many questions per interview and elicited roughly 65 percent more respondent text per interview. Per-question respo
标题:NestedVLA: Learning to Consolidate and Generate Skills for Vision-Language-Action Models
来源:arXiv
链接:https://arxiv.org/abs/2610.23271
摘要:Vision--Language--Action (VLA) models promise generalist robotic agents, yet consolidating skills learned across heterogeneous tasks and environments remains difficult: training data is fragmented across platforms, joint multi-task training suffers from interference, and model merging tends to introduce parameter conflicts or require architectural changes. We instead advocate a paradigm in which skills are first learned independently and then consolidated post hoc. To this end, we propose NestedVLA, which replaces parameter aggregation with parameter generation: a nested hypernetwork, conditioned on the target task, synthesizes task-adaptive action-expert parameters that compose with a frozen base VLA. Task-specific compressed VLM residuals are retained separately. This formulation enables
标题:Who Pays for Efficient Inference? Confidence-Based Small-to-Large LLM Routing Across Tasks and Languages
来源:arXiv
链接:https://arxiv.org/abs/2610.23249
摘要:Small-to-large language model routing can reduce inference cost by escalating difficult inputs while retaining easier ones on a smaller model. Its success, however, depends on identifying examples that benefit from escalation, and the reliability of confidence across multilingual tasks remains unclear. We present a systematic evaluation of confidence-based routing across three model families, four languages, and three task types, covering 72 model, dataset, language, and scale configurations and 42,288 predictions. We compare global confidence routing with random routing, equal-quota confidence routing, and an oracle upper bound. Confidence identifies beneficial escalations on the Belebele reading comprehension and SIB-200 topic classification datasets, with mean correction-detection AUCs
标题:ScoutRoute: Efficient Mixture-of-Experts Routing Using Internal Language Model Signals
来源:arXiv
链接:https://arxiv.org/abs/2610.23443
摘要:Efficient expert routing is a fundamental challenge in Mixture-of-Experts (MoE) systems, particularly for heterogeneous collections of independently trained Large Language Mod els (LLMs). Existing routing approaches typically rely on jointly trained gating networks, prompt-based predictors, or computationally expensive multi-model inference, limiting their flexibility and efficiency. This paper presents ScoutRoute, a lightweight post-hoc routing framework that predicts expert competence using internal language model signals extracted from a single forward pass through an inexpensive scout model. Hidden-state representations are used to estimate cross-model competence through lightweight logistic probes, while logit-derived confidence measures are incorporated within a domain-aware routing
标题:Chinese Decision: A Family of Language Models for Single-Pass Probabilistic Decision-Making
来源:arXiv
链接:https://arxiv.org/abs/2610.23170
摘要:Many language-based applications require probabilities over a finite set of decisions rather than open-ended responses. We present Chinese Decision, a family of Chinese-language models for single-pass probabilistic decision-making. Given context, an instruction, and candidate descriptions, the model extracts final-position candidate-letter logits in one forward pass and normalizes them over valid options, without autoregressive answer generation. A shared interface supports Boolean judgments, categorical choices, and ordinal scores. We report a completed 0.8B model trained with distribution cross-entropy on 13.048 million decisions and preliminary 2B checkpoints trained with an additional RLCD objective. On the full CJ-Bench of 307,900 decisions, the uncalibrated 0.8B model achieves 72.24%
标题:Before Retrieval Begins: Safety and Tool-Use Decisions in LLM Agents
来源:arXiv
链接:https://arxiv.org/abs/2610.23071
摘要:Large language models increasingly act as agents that call external tools, yet most tool-safety research examines only what follows a tool call. The prior decision of whether to call the tool remains largely unexamined and can influence downstream safety. Across the seven open-weight models, the unweighted model-average harm rate after admitted harmful requests is 74.5%. To identify what governs this decision, we apply a factorial intervention to the input request. Within our generated construction, neutralizing the requested-operation component produces the largest admission contrast (19 pp). A linear direction estimated from this contrast influences held-out admission in four of seven models; effects on operation-neutralized requests in some models show that it is not an operation-specif
标题:MGE-PO: Entropy-Aware Credit Assignment for Mathematical Reasoning
来源:arXiv
链接:https://arxiv.org/abs/2610.23245
摘要:GRPO assigns the same group-relative advantage to equally rewarded responses within a question, leaving their generation uncertainty unused. We introduce MGE-PO, centered on four-quadrant (4Q) advantage shaping that combines verified correctness with rollout generation entropy. Correctness determines the sign of an additional correction, while entropy controls its magnitude: lower-entropy successes receive larger bonuses and higher-entropy failures receive larger penalties. An online entropy-based question curriculum supports this credit rule using cached rollout measurements. At the final checkpoint after eight training epochs on Qwen3-8B-Base, adding 4Q across three policy optimization settings yields absolute gains of 3.23-3.69% in six-benchmark macro avg@32 and 4.24-11.06% in macro pas
标题:Process Reward Modeling with Reliability- and Conservatism-Aware Bayesian Belief
来源:arXiv
链接:https://arxiv.org/abs/2610.23223
摘要:Process Reward Models (PRMs) play a central role in test-time scaling for large language model (LLM) reasoning. However, PRMs may provide inaccurate process rewards for test-time scaling, which we identify as arising from two sources of uncertainty: (1) \textit{PRM uncertainty} that arises from insufficient supervision data and imperfect PRM training; and (2) \textit{policy uncertainty} that results from the mismatch between the policy to collect supervision data and the policy for test-time scaling, leading to systematic reward overestimation. Previous studies fail to jointly consider both uncertainties. To bridge this gap, we propose BayesianPRM, which treats the process reward as a random variable within a reward hypothesis space and maintains a Bayesian belief over it. Since exact Baye
标题:Not All Test-Time Compute Is Evidence: Model-Specific Bias in LLM Reasoning
来源:arXiv
链接:https://arxiv.org/abs/2610.23213
摘要:Test-time scaling methods implicitly assume that additional computation from the same model provides independent evidence. We show that this assumption fails within a single reasoning model: errors are correlated across repeated samples, across reasoning depth, and in the model's own self-evaluation, so more compute from the same model is not more evidence. We introduce DIA (Disagreement-triggered Independent Arbitration), which treats disagreement between an intermediate-stage and the final-stage prediction as a per-question signal of reasoning instability and resolves it by asking a small panel of heterogeneous models to solve the question independently, without seeing either candidate, and aggregating their answers directly. Across seven benchmarks and two base models, DIA outperforms f
标题:Beyond the Answer: Student-Conditioned Trace Construction for Reasoning Distillation
来源:arXiv
链接:https://arxiv.org/abs/2610.23831
摘要:A reasoning trace can accompany a correct answer without providing effective supervision for a particular student. We study how to construct traces that improve learning beyond answer-only supervision by considering both their intermediate content and their accessibility to the target learner. We introduce ReDiST, a student-conditioned trace-construction and distillation framework for evidence-rich reasoning. Starting from the student's own attempts, ReDiST revises answer-correct drafts against the supplied evidence and repairs unsuccessful drafts with additional guidance from the verified answer. A frozen student-matched reader screens and ranks original and edited candidates using answer recovery and verified-answer probability, given the question, evidence, and candidate trace. We disti
标题:OwlSight: A Panoramic Monitor for Chain-of-Thought Reasoning
来源:arXiv
链接:https://arxiv.org/abs/2610.23670
摘要:Chain-of-Thought (CoT) monitoring offers actionable evidence for oversight AI systems, with the potential to enable early detection and intervention. However, existing approaches are focused on specific tasks, whereas the regulation of AI, particularly reasoning models, requires monitoring a broader range of tasks. We introduce Factor for CoT monitoring, a strategy that supports customizing monitoring requirements in the prompt, which enables a single monitor to cover different tasks by relying on different factors. Furthermore, we introduce two types of factors, fine-grained and coarse-grained (e.g., monitoring whether a specific email address has been leaked versus monitoring privacy leakage), to accommodate diverse user monitoring needs. Based on this, we develop a CoT monitor named Owl
标题:Reasoning Models Can Control Their Chains of Thought When They Tell Themselves To
来源:arXiv
链接:https://arxiv.org/abs/2610.23257
摘要:Chain-of-thought (CoT) controllability, the ability of reasoning models to control properties of their CoT, is consequential for both their usefulness and reliable oversight. Recent work evaluates this capability by measuring whether models follow CoT constraints given in the user prompt, and reports very low controllability across models. We argue that this low controllability may not reflect an inherent inability of reasoning models to control their CoT, but instead result from naively specifying constraints in the user prompt. To test this, we introduce Self-Instructed CoT Controllability: a complementary evaluation that places the constraint as a self-instruction inside the model's CoT and measures its ability to follow it. We show that, under self-instruction, models achieve significa
标题:ReLaDrift: Recurrent Language Generation via Drifting for Test-Time Scaling
来源:arXiv
链接:https://arxiv.org/abs/2610.23225
摘要:Drifting Models generate a sample with a single function evaluation, making them attractive for fast parallel language generation, but their one-step formulation does not support iterative refinement. We propose Recurrent Language Generation via Drifting (ReLaDrift), which uses a shared Drifting Model to generate an initial draft from noise and iteratively refine it in a continuous latent space. We train this recurrence by sampling the number of rounds and applying the drifting loss only at the final round, stopping gradients through preceding rounds so that only one round requires backpropagation. Unlike the one-step Drifting Model, whose accuracy does not improve when iterated, ReLaDrift enables a trade-off between test-time computation and generation quality, as diffusion and flow langu
标题:Rethinking GRPO for Large Reasoning Model Safety Alignment
来源:arXiv
链接:https://arxiv.org/abs/2610.23085
摘要:Large reasoning models (LRMs) exhibit strong reasoning capabilities but remain vulnerable to unsafe outputs. While Group Relative Policy Optimization (GRPO) is widely used for safety alignment, its group-relative optimization may provide insufficient safety-directed learning signals. In this paper, we identify two failure modes: advantage collapse, where homogeneous rollout groups yield vanishing advantages, and unsafe diversification, where unsafe behaviors persist and become increasingly diverse. Our theoretical analysis reveals that behavioral consistency weakens within-group safety contrasts, while entropy regularization can diversify unsafe responses when relative signals vanish. To address these challenges, we propose Beacon, comprising Safety Signal Recovery (SSR) and Safety-Utility
标题:Sample, Mix, or Verify? Allocating Test-Time Compute under Imperfect Verifiers
来源:arXiv
链接:https://arxiv.org/abs/2610.23255
摘要:A common way to make language models and agents more accurate is to spend more compute at inference time: generate many candidate solutions and submit the one a verifier prefers. Success then requires that some candidate is correct and that the verifier picks it. The first grows predictably with the number of samples; the second does not, because real verifiers (unit tests, majority votes, LLM judges, reward models) accept wrong answers, reject right ones, and make correlated mistakes. With a fixed budget, a practitioner must decide how many samples to draw, which models to draw them from, and how much to spend on verification, and today these choices are made by rules of thumb. We introduce SPLIT, which makes them from a small labelled pilot. It rests on an exact formula for the probabili
标题:MultiAgent-OPD: End-to-End Training of Hierarchical Multi-Agent Systems with On-Policy Distillation
来源:arXiv
链接:https://arxiv.org/abs/2610.23234
摘要:In a hierarchical multi-agent system, a Leader assigns tasks and subagents carry them out. Jointly training the Leader and the subagents is difficult because a final answer gives only indirect feedback to the subagents. An LLM judge can score the performance of the subagents, but these scores depend on the judge's own assessment of the quality of subagents' responses. We present MultiAgent-OPD, which combines on-policy distillation with reinforcement learning to jointly train a shared-parameter Leader-Searcher-Answerer team end to end. We first train a Teacher in a Searcher-Answerer workflow using answer correctness. During team training, the frozen Teacher scores student-generated search and summary tokens to provide local process supervision for the Searchers, while the Leader learns fro
标题:GR-OPD: Group-Relative Sequence Credit for On-Policy Distillation
来源:arXiv
链接:https://arxiv.org/abs/2610.23042
摘要:On-policy distillation (OPD) provides token-level teacher feedback, yet common sampled-token updates assign credit locally and omit the effects of earlier actions on subsequent feedback. We introduce \method{} (Group-Relative On-Policy Distillation), which decouples feedback computation from credit assignment by aggregating token feedback into calibrated sequence-level credit. We establish a theoretical basis for this approach by showing that, under ideal on-policy assumptions, broadcasting raw summed feedback to every response token yields an unbiased estimate of the negative gradient of the sequence-level reverse-KL objective. To support effective learning in practice, \method{} discounts later feedback that may become less reliable under distribution shift and uses same-prompt rollout g
标题:Grounded Credit Assignment in Reinforcement Learning for Multi-Turn Search Agents
来源:arXiv
链接:https://arxiv.org/abs/2610.23148
摘要:Outcome-reward reinforcement learning such as GRPO gives every token of a search-agent rollout the same advantage, although the rollout exercises two capabilities: finding evidence and using it. We show that accuracy improves without grounding: Search-R1 raises accuracy on all seven QA benchmarks, yet on Bamboogle the evidence-supported share of correct answers falls from 84.0% to 42.9%. We attribute this to GRPO's credit assignment: rollouts that retrieve answer-bearing evidence are favoured in aggregate only in proportion to their accuracy gain over the rest, a grounding gap that answering from memory or misusing evidence shrinks or reverses. To address this, we propose Grounded Credit Assignment (GCA), which uses verifiable sub-outcomes (evidence retrieval and answer correctness) to cor
标题:Distill Until Divergence: Error-Truncated Hindsight Distillation for Agentic RL
来源:arXiv
链接:https://arxiv.org/abs/2610.23048
摘要:Long-horizon interaction requires agents to make multiple decisions in a changing environment. Outcome-based GRPO lacks fine-grained guidance for intermediate decisions. Recent methods distill completed trajectories into reusable skills and use hindsight-conditioned OPD to provide dense token-level credit. However, skill-induced probability shifts do not necessarily represent reward-aligned credit; especially after an erroneous action changes subsequent interaction states, skill-conditioned supervision may become mismatched with the current state. To address this mismatch, we propose \methodname{}, which routes two complementary hindsight skills according to trajectory outcomes and error boundaries: Positive skills capture successful workflows to reinforce successful trajectories and pre-e
标题:Adaptation of Computer Use Agents to New Software with User Manuals
来源:arXiv
链接:https://arxiv.org/abs/2610.23012
摘要:Recent advances in multimodal foundation models have significantly improved the ability to operate real-world software through graphical user interfaces (GUIs) using visual observations and natural language instructions. However, maintaining strong performance on unseen applications remains challenging, as practical tasks often require long-horizon procedural knowledge and application-specific operation patterns. In addition, constructing executable tasks and collecting corresponding trajectories for training still heavily rely on human-annotated data, incurring substantial cost. In this paper, we propose MAGE (Manual-augmented computer use AGent Execution and learning), a framework that leverages user manuals as a scalable source of procedural knowledge to improve both inference and train
标题:Self-Evolving Agents Can Repair Failures but Struggle to Generalize from Them
来源:arXiv
链接:https://arxiv.org/abs/2610.22984
摘要:Experience-driven self-evolution offers a promising path for large language model (LLM) agents to continually improve through interaction. Yet existing approaches often rely on successful trajectories or external evaluation signals, leaving unclear whether agents can evolve when such positive anchors are unavailable. We study this question through Cold-Start Failure-Only Evolution (CFOE), where agents start without accumulated experience and learn only from failed interactions without privileged guidance. Across three self-evolving paradigms, multiple LLM backbones, and four interactive environments, we uncover a consistent repair-generalization gap: failure-derived experience can help repair previously failed tasks, but rarely yields sustained improvements on unseen tasks. Further analysi
标题:SkillBrowser: Selective Skill Exposure for Scalable Skill-Using Agents
来源:arXiv
链接:https://arxiv.org/abs/2610.23554
摘要:Large language model agents increasingly rely on skills to extend their capabilities in complex tasks. However, scalable skill use requires Selective Skill Exposure (SSE): providing relevant procedural content when needed while avoiding unnecessary exposure that may disrupt reasoning. Existing paradigms fall short in complementary ways. Progressive disclosure exposes metadata for a growing skill library, introducing context pollution; retrieval limits catalog exposure, but static retrieval from the initial task query may be misaligned with capability needs that emerge during reasoning and may still inject unnecessary skills. We propose SkillBrowser, a Meta-Skill-managed framework that enables an agent to identify current capability needs, formulate targeted queries, and selectively load sk
标题:Agentic In-Context Learning: What Algorithms Can Transformers Learn in Closed-Loop Interaction?
来源:arXiv
链接:https://arxiv.org/abs/2610.23051
摘要:The learning algorithm view of in-context learning (ICL) studies how fixed-parameter Transformers learn from supplied examples. Language model agents extend this setting through interaction: their actions determine what evidence enters context next. We call this regime Agentic In-Context Learning (Agentic ICL). Existing studies largely depend on externally supplied contexts or reward-bearing interaction histories, while work on general tool-using agents lacks a learning algorithm view. We represent an interactive learner by its state, update, and decision rule, covering active learning, bandits, and online reinforcement learning (RL). For expressivity, we construct causal Transformers that approximate these algorithms over a finite horizon, and bound the resulting loss in return. For train
标题:From Trajectories to Differentiated Skills: Cognitive Differentiation for Self-Improving Agents
来源:arXiv
链接:https://arxiv.org/abs/2610.23599
摘要:Large language model (LLM) agents tackle complex tasks through reasoning and interaction with external environments. Agent skills provide persistent, transferable capabilities that guide task planning, inform contextual decisions, support tool use, and enable executable procedures. Yet existing skill-evolution methods largely organize knowledge around skill artifacts, overlooking the functional heterogeneity within individual skills and its implications for knowledge maintenance and use. Inspired by cognitive science, we propose Cognitive Differentiation in Skill Evolution (CDSE), which evolves a unified skill through four functionally distinct yet coordinated components: Meta, Experience, Tool, and Script. CDSE extracts evidence-grounded semantic gradients from successful and failed traje
标题:Behavioral Couplings: From Formation to Behavioral Emergence in Language Models
来源:arXiv
链接:https://arxiv.org/abs/2610.23847
摘要:Fine-tuning large language models (LLMs) can induce behaviors beyond those explicitly targeted. For example, training a model to respond more warmly can make it more sycophantic, and training it to write insecure code can cause broad misalignment on prompts unrelated to coding. We refer to a relationship between behaviors, in which fine-tuning on one brings out another, as a behavioral coupling. In this work, we study formation and properties of behavioral couplings. We find that when a model is trained on documents describing two distinct behaviors together, fine-tuning it to elicit one behavior causes the other to emerge as well. For instance, when training documents describe a fictitious character that both explains scientific concepts in a playful, rubber-duck-like manner and discourag
标题:Path to Prior-Data Fitted Relational Models via Learned Latent Multi-table Representations
来源:arXiv
链接:https://arxiv.org/abs/2610.23073
摘要:Prior-data fitted networks (PFNs) show that pretraining on large populations of synthetic tasks can enable generalization to unseen datasets with little or no downstream fine-tuning. This paradigm is effective for transformer-based tabular and relational foundation models such as RDB-PFN, but remains largely unexplored for graph-based relational models built on message passing. We pretrain the graph-based relational foundation model Griffin on large collections of synthetic relational databases and ask whether curriculum learning, which benefits transformer-based relational learners, also improves message-passing representations. Rather than fine-tune on downstream tasks, we use the pretrained encoder as a general-purpose relational representation learner and pass its latent representation
标题:Out of Support, Not Out of Reach: In-Context Learning with Unsupported Instructions and Composed Tasks
来源:arXiv
链接:https://arxiv.org/abs/2610.23528
摘要:In-Context Learning (ICL) enables pretrained LLMs to adapt to downstream tasks from a small set of input-output demonstrations, without parameter updates. Yet existing theory primarily explains ICL as identifying a latent task already represented in pretraining, leaving unclear whether ICL can succeed when either the instruction or the target task falls outside the pretraining support. We develop a theoretical framework under mild assumptions that separates these two forms of distribution shift. For target tasks represented in pretraining, we show that demonstrations can identify the task and reduce sensitivity to pretrain-supported prompt templates as their number increases. For an unsupported instruction, we show that under a fixed model-induced reference extension, demonstration evidenc
标题:COMET: Decoupled Distillation, Routing, and Capacity Control for Task-Agnostic Continual Vision--Language Learning
来源:arXiv
链接:https://arxiv.org/abs/2610.22907
摘要:Vision-language models (VLMs) such as CLIP retain zero-shot recognition ability, but their open-world use often requires task-agnostic continual learning: tasks arrive sequentially, past data are unavailable, task identity is unknown, and the backbone cannot be retrained. We argue that this setting fails VLMs through three coupled pressures beyond the stability-plasticity tradeoff: (1) modal-consistency drift, (2) representation-space interference, and (3) per-task capacity shortfall. We propose COMET (Collaborative Mixture-of-Expert Transfer) to resolve these three issues respectively as follows. A frozen teacher provides only a feature reference on a shared image-text pool; Tri-Affinity Distillation preserves image-image, text-text, and image-text geometry; and Student-Aware Cross-Expert
标题:AI-Driven Reproducibility and Analytical Robustness in Scientific Discovery
来源:arXiv
链接:https://arxiv.org/abs/2610.23845
摘要:Reproducible and analytically robust research findings are essential to scientific progress, but evaluating these properties typically requires costly manual effort. We investigate whether LLM agents can help automate reproduction and robustness analysis, using a medical case study grounded in the Surveillance, Epidemiology, and End Results (SEER) database, a widely used U.S. cancer registry underlying tens of thousands of published studies. First, we introduce SEER-Repro, a benchmark of 20 published SEER-based papers together with the data and resources required to reproduce their findings. Second, we create an AI agent that successfully reproduces 77.4% of 2,190 atomic claims, while providing executable baselines for further analysis. Third, we develop an agentic analysis pipeline to tes
★2Optimist Deep Research:基于可变研究状态的迭代式文献研究智能体arXiv · 深度研究智能体 · 论文Optimist Deep Research 将文献研究建模为围绕显式可变状态持续获取、审查和组织证据的迭代过程,并以可追溯证据生成最终报告。
标题:Optimist Deep Research: An Iterative, State-Centric Agent for Autonomous Literature Research
来源:arXiv
链接:https://arxiv.org/abs/2610.23299
摘要:Most retrieval-augmented systems treat literature search as a single retrieval step followed by summarization. Real literature research works differently: the investigator's understanding of the question changes as they read, new subquestions appear, and evidence is weighed, not just collected. We present Optimist Deep Research, an open-source agent that runs literature research as an iterative evidence-acquisition process over an explicit, mutable research state. A research question is broken into a tree of subquestions, each served by search intents that are translated into source-specific queries over arXiv, OpenAlex, Crossref, PubMed, Semantic Scholar and the web. Results are merged into a canonical, provenance-preserving document registry through a conservative four-tier deduplication
标题:Where Research Ideation Begins: Forward-Citation Supervision for Actionable Limitation Generation
来源:arXiv
链接:https://arxiv.org/abs/2610.23038
摘要:Automated research ideation has advanced rapidly, with multi-agent workflows that generate whole research ideas end to end. These frameworks leave largely unexamined the step that seeds a good idea: identifying an actionable limitation from literature. In this paper, we focus on this first step in the research ideation pipeline. Starting from a pilot study, we find that as large language models (LLMs) scale from 0.6B to 32B parameters, they become far better at recognizing which limitation later work addressed, but no better at generating one, suggesting that the bottleneck lies in the supervision signal rather than model capacity. We therefore construct ActLim-Bench from forward citations, yielding 18,279 evidence-linked supervision triples with human verification. On this signal, we prop
标题:VISTA: Steering LLMs with Latent User Values Inferred from a Few Survey Answers
来源:arXiv
链接:https://arxiv.org/abs/2610.23702
摘要:Personalizing a language model requires knowing what a person values, which stays largely consistent across tasks. Yet most methods either learn task-specific preferences or place the user's profile in the prompt, where models struggle to act on it. We introduce VISTA (Value-Informed Steering and Targeted Acquisition), which infers a latent profile from users' (full or partial) survey answers, and maps the profile through a hypernetwork to a user-specific steering intervention in a frozen LLM to select answer choices or generate free-text responses. Given the same evidence, VISTA matches or beats in-context profiles on synthetic preferences (Food2D), the World Values Survey (WVS), and PrimeX, without adding prompt tokens. To reduce the cognitive burden of completing full surveys, VISTA als
标题:Reason-guided Human Reward Modeling for Preference-based Reinforcement Learning
来源:arXiv
链接:https://arxiv.org/abs/2610.23550
摘要:Reinforcement learning (RL) policies optimized for task rewards may produce behaviors that are efficient but misaligned with human expectations, safety preferences, or intent. This paper presents a reason-guided preference-based reinforcement learning (PbRL) framework that learns a human reward model from trajectory-level feedback. In addition to preference labels, the framework incorporates reason tags explaining why a behavior is preferred, enabling the learned reward to capture both human choices and their underlying rationale. Compared with preference-only reward learning, the reason-guided formulation provides additional supervision for learning human-aligned rewards and improves the interpretability of the resulting policy. The learned human reward is combined with the environment re
★2Trapped in Your Story:叙事诱导的谄媚行为评测arXiv · 安全评测 · 论文研究构建受控基准,揭示用户叙事的因果框架、世界观综合及呈现形式会显著影响语言模型的谄媚性判断。
标题:Trapped in Your Story: Evaluating Narrative-Induced Sycophancy in Large Language Models
来源:arXiv
链接:https://arxiv.org/abs/2610.23525
摘要:Large language models (LLMs) are known to exhibit sycophancy by adjusting their responses toward users' expressed views. However, real users often justify their opinions through reasons, personal experiences, causal explanations, and broader narratives, raising the question of whether such narratives can further steer model judgments. We construct a controlled benchmark of 500 debatable questions across ten domains, where the user stance is held fixed while its justification is systematically varied without introducing new task-relevant evidence. Each question is instantiated under 11 matched conditions: one neutral baseline and five mirrored narrative levels for each of two opposing positions, yielding 5,500 evaluation instances. We evaluate seven LLMs and observe substantial differences
HardEval基于Red Hat Enterprise Linux 8的STIG控制项构建,任务覆盖SSH配置、身份认证策略、审计规则与内核参数等操作系统安全领域。每项任务仅向模型提供目标安全状态,不公开专家修复方案及作为隐藏真值的OVAL检查。模型生成的Bash脚本会在真实虚拟机中执行,并通过对应的OpenSCAP/OVAL合规检查判定修复是否成功。对14种前沿及开放权重模型配置的评测显示,不同安全领域的能力差异显著,常见失败是脚本看似合理却未精确满足合规条件。
用途与启示
为语言模型驱动的操作系统安全加固提供可执行、可复现且以最终系统状态为依据的评测方式。
揭示仅检查代码表面合理性不足以判断合规修复是否成功,应结合真实环境执行与隐藏检查器。
可用于比较模型在不同安全配置领域的可靠性,并定位持久化配置、精确条件满足等失效模式。
为安全运维智能体、自动合规修复工具及其验收测试提供基准基础。
📝 原始笔记(逐字保留)
标题:HardEval: An Execution-Verified Benchmark for AI-Driven Security Hardening and Compliance Remediation
来源:arXiv
链接:https://arxiv.org/abs/2610.23685
摘要:Implementing Security Technical Implementation Guide (STIG) controls remains a demanding operating-system administration task. Each natural-language security requirement must be translated into a precise and persistent configuration change. These changes span Secure Shell (SSH) configuration, authentication policy, audit rules, and kernel parameters. Existing benchmarks do not measure whether a model can perform this task on a live system. Code benchmarks evaluate isolated programs without persistent system state, while agent benchmarks largely focus on applications rather than operating-system configuration. In this paper, we introduce \textbf{HardEval}, an execution-verified benchmark for evaluating language-model-assisted security hardening. HardEval is built from Red Hat Enterprise Lin
标题:Robust Function Calling for LLM Agents Under Interface Drift
来源:arXiv
链接:https://arxiv.org/abs/2610.23342
摘要:Tool-augmented language-model agents rely on interface text: tool descriptions, argument schemas, and the candidate tool list shown at inference time. These interfaces often change after deployment even when the underlying tool semantics remain fixed, creating a practical reliability problem for function-calling systems. We study this failure mode as interface drift and evaluate whether a lightweight, training-free repair layer can recover robustness without retraining the base model. Our method, SchemaShield-Lite, combines schema validation, candidate-set retry for missing tool names, and one-shot constrained repair. On Qwen3.5-9B, it raises DICE-300 accuracy from 0.717 to 0.813 and BFCL-200 accuracy from 0.580 to 0.805, recovering 19/22 and 33/45 drift-induced failures with no observed r
标题:FLARE: Physics-Grounded Evaluation of Language-Model Agents for Feeder Restoration with Degraded Observations
来源:arXiv
链接:https://arxiv.org/abs/2610.23251
摘要:Restoring service after distribution-grid faults requires decisions from incomplete and conflicting information while respecting electrical constraints. Addressing this need, we introduce FLARE (Feeder-Level Agent Restoration Evaluation), a physics-grounded benchmark for post-isolation feeder restoration. FLARE con- tains 800 AC-feasible cases across five observation settings and allows agents to gather evidence, evaluate switches, and perform restoration actions under fixed operational limits. We evaluate five instruction-tuned models over 78,750 episodes, including topology, distributed-generation, and loading shifts. Mean normalized regret ranges from 0.082 to 0.148. Although every model has zero median regret, P95 regret ranges from 0.814 to 1.000, revealing an important failure tail.
标题:From Approval to Abuse: Privilege Escalation Attacks on Harness-Protected Coding Agents
来源:arXiv
链接:https://arxiv.org/abs/2610.22987
摘要:Coding agents use execution harnesses to constrain the consequences of prompt injection by enforcing permission boundaries. These systems deterministically intercept any out-of-sandbox action and prompt the user for approval. To reduce approval fatigue, users can approve allow-rules proposed by the agent, after which any future action matching the rule is allowed without a prompt. In this work, we show that this mechanism introduces a new risk: untrusted context can influence the permissions presented for approval, with consequences extending beyond the immediate task. We investigate privilege escalation targeting this permission-update process. Specifically, adversarial context planted in a repository can reliably steer an agent performing a routine task into requesting a plausibly justif
标题:Deconstructing Multi-Task Active Learning: The Paradox of Gradient Conflict and Orthogonal Decomposition
来源:arXiv
链接:https://arxiv.org/abs/2610.23623
摘要:Existing multi-task active learning (MTL-AL) methods predominantly rely on linear scalarization of heterogeneous objectives such as uncertainty, diversity, and gradient conflict. We argue that this entangled formulation is fundamentally limited, as linear combination fails to capture Pareto-optimal trade-offs under conflicting objectives. We propose a shift from combination to deconstruction and introduce a principled framework that rethinks MTL-AL along three orthogonal dimensions. First, we identify the Conflict Paradox: although the optimization literature treats gradient conflict as interference to be suppressed, penalizing conflict when ranking candidates is measurably worse than not scoring it at all. We therefore decouple the two concerns: the acquisition function does not score con
标题:LongForge: Measuring and Improving Long-Context Instruction Data via Evidence-Assertion Reasoning Graphs
来源:arXiv
链接:https://arxiv.org/abs/2610.23167
摘要:Existing quality control for long-context instruction-tuning data is a one-shot judgment inside the generation pipeline that leaves only a scalar, revealing neither which document parts an answer uses nor how many reasoning steps it writes down; the data can thus be neither audited nor controlled. We formalize data quality as a property of the alignment structure among document, question, and answer, and build LongForge, which pairs an analyzer that measures long-context data with a synthesizer that generates it. The analyzer reverse-reconstructs each sample into an evidence-assertion reasoning graph and programmatically computes four dimensions on it: breadth, depth, correctness, and completeness. Every reading traces back to specific evidence spans and reasoning steps, making data audita
标题:SIT-Eval: Situation-Based Interactive Turn-by-Turn Evaluation of Conversational Agents
来源:arXiv
链接:https://arxiv.org/abs/2610.23612
摘要:The rapid proliferation of conversational AI agents necessitates evaluation methods that assess performance across diverse dialogue scenarios, utilizing multiple situational criteria, as appropriate behavior depends on context and encompasses various aspects. Existing dialogue benchmarks typically use static natural conversations, which underrepresent rare situations---a critical safety gap, as low-probability events become frequent failures at production scale. However, limited sample sizes force a choice: either maintain natural dialogues for realism or turn to adaptive situation sampling for comprehensive yet statistically reliable coverage. We introduce SIT-Eval, an adjustable framework for evaluating situation-based conversational agents. We formalize an agent's desired behavior as si
标题:GatewayBench: Unified Capability and Security Evaluation Platform for MCP Gateways
来源:arXiv
链接:https://arxiv.org/abs/2610.23449
摘要:AI agents have increasingly interacted with external environments by invoking tools, with Model Context Protocol (MCP) emerging as a standard interface for connecting agents to heterogeneous tools and services. As organizations deploy agents at scale, MCP gateways have become a critical middleware layer for aggregating, governing, and shaping the tool surface exposed to agents. Despite the growing importance of MCP gateways in agentic systems, no prior work has systematically studied how gateway design affects agent capability, efficiency, and security. To address this gap, we introduce GatewayBench, the first unified evaluation framework and benchmark for MCP gateways. We characterize gateway architectures into two major paradigms: flattened gateways and hierarchical gateways. GatewayBenc
标题:Monitor-Validity Failures: LLM Safety Evidence Paths at Evolving Framework Boundaries
来源:arXiv
链接:https://arxiv.org/abs/2610.23362
摘要:Modern LLM applications rely on runtime monitors that inspect tool calls, traces, streams, detector outputs, and verifier inputs. This creates an interface-reliability problem: a framework path may remain operational while policy-relevant evidence never reaches its decision point. In one representative case, a regex detector returned a positive match, but the framework stranded the result before the block decision. This differs from jailbreaks and detector errors because the evidence is already recognized yet disconnected from enforcement. We study this repair-boundary signature with canary contracts: parent/fixed confirmation units that place a known payload only on the changed evidence surface and record monitor visibility together with the downstream decision for action-boundary cases.
标题:Self-Reflective Chain-of-Thought (SRCoT): Evidence-Based Text Generation via a Closed-Loop Retrieve--Reflect--Revise Pipeline on Edge LLMs
来源:arXiv
链接:https://arxiv.org/abs/2610.23311
摘要:Large language models (LLMs) can generate fluent outputs that lack source support or conflict with local operational constraints. We introduce Self-Reflective Chain-of-Thought (SRCoT), a modular architecture that organizes generation as a closed Retrieve--Reflect--Revise loop. SRCoT separates environment state, persistent knowledge, accumulated experience, and short-term memory into provenance-labeled stores. BM25 retrieval supplies local context; subsequent passes expose intermediate drafts, missing-information markers, follow-up queries, and optional claim--evidence mappings. These artifacts inform retrieval, revision, clarification, or abstention decisions. We examine the architecture through qualitative, trace-based case studies in two fully synthetic settings: emergency medical servic
标题:Look, Reason, Draw a Region: Recursive Refinement for Spatial Context Prediction
来源:arXiv
链接:https://arxiv.org/abs/2610.23302
摘要:Spatial context prediction requires visual grounding and spatial reasoning to identify vacant regions relative to objects in a scene. Spatial relationships and physical constraints remain challenging for large vision-language models (VLMs). Recursive reasoning has demonstrated strong performance on discrete and abstract reasoning tasks, motivating its use for improving spatial context prediction in natural images. Weight sharing enables higher accuracy with a small trainable parameter count. We introduce Spatial-TRM, a compact recursive model with frozen vision-language features, a convex region decoder, and just 11.2M trainable parameters. This weight-tied reasoner supports six spatial relationships, three reference frames, and open-vocabulary anchor objects. On the RoboSpatial-Home bench
标题:Beyond Reward Scoring: Online Multi-Agent Preference Learning for Multimodal Response Generation
来源:arXiv
链接:https://arxiv.org/abs/2610.23256
摘要:Multimodal preference alignment often follows a score-based RLHF pipeline: a reward model is first trained to learn human preferences, then used to score on-policy responses and optimize the response model's generation. This turns preference supervision into a scalar signal for generation: a model may learn to produce answers with higher preference scores, but is rarely trained explicitly to learn the reasons underlying human preferences. We therefore ask whether making preference understanding an internal capability of the response model can improve its generation. To investigate this question, we propose \textbf{PrefMind}, an online dual-agent reinforcement learning framework with two trainable agents: a response Generator and a preference Critic. Given a pair of candidate responses, the
标题:NL2SQL with Iterative Self-Refinement
来源:arXiv
链接:https://arxiv.org/abs/2610.22968
摘要:Bridging the gap between human intent and machine logic remains a critical challenge in data access for non-technical users, yet Large Language Models (LLMs) remain unreliable: they hallucinate schema elements, generate syntactically invalid queries, or fail to capture user intent. These limitations are especially critical in domains such as insurance, where KPIs, regulatory constraints, and legacy relational systems demand high precision. We introduce a multi-agent, persona-driven NL2SQL framework that combines (1) domain relevance detection, (2) schema-grounded knowledge construction, (3) KPI-aware SQL synthesis, (4) syntax-level self-repair using sqlite execution feedback, and (5) semantic consistency checking via SQL-to-NL reverse engineering. Our system significantly reduces query hal
标题:A Stronger Teacher Within: Strengthening MoE Models via On-Policy Bidirectional Distillation
来源:arXiv
链接:https://arxiv.org/abs/2610.23185
摘要:Reinforcement Learning with Verifiable Rewards (RLVR) has substantially improved Mixture-of-Experts (MoE) LLMs, but its outcome-level rewards provide only sparse supervision. On-Policy Distillation offers dense token-level feedback, yet requires a stronger compatible teacher, while on-policy self-distillation relies on hand-designed privileged information. We observe that the MoE architecture itself offers a way to resolve this dilemma, as increasing the activation budget per token exposes latent expert capacity and yields a potentially stronger teacher within the model. However, naive expansion degrades performance because newly co-activated experts are not optimized to work together. We therefore propose MoEvolve, a bidirectional distillation framework following a recovery-then-improveme
标题:Annotrace: A Workbench for Step-Level Annotation of Agent Traces
来源:arXiv
链接:https://arxiv.org/abs/2610.23284
摘要:Every study of why LLM agents fail begins by labelling traces by hand, and every such study builds its own throwaway tool. Observability platforms record agent traces but offer no structured labelling; annotation tools label text but have no notion of a step, a delegation or a tool call, leaving it unclear what the smallest unit of execution to label should be. Annotrace reads six trace formats, including OpenTelemetry GenAI, LangSmith and Langfuse, and turns each run into an ordered, nested list of steps. The annotator writes their own labelling codebook; a frontier model proposes a starting set from a sample of their own traces, answers first, and is asked exactly the questions the human is asked, making annotation a joint human-AI process. Annotrace measures agreement between any two an
标题:Think More or Read More? A Token-Budget Theory of Reasoning with Shared Context
来源:arXiv
链接:https://arxiv.org/abs/2610.23246
摘要:Reasoning systems increasingly improve answers by incorporating additional context into the model's input, from multi-agent debate, where agents read their peers' responses, to self-refinement, where a model revisits its own drafts. In these systems, communication consumes context as one agent's output becomes another's input. We study the regime where token reading and generation draw on one fixed token window, so every token spent reading is a token that cannot be spent reasoning. We develop a \textbf{token-budget theory} that formalizes this trade-off, modeling accuracy as a function of the reasoning budget remaining after reading and aggregating votes with a Condorcet jury model that permits correlated errors. For multi-agent debate, the theory gives a closed-form prediction for the op
研究基于 ANES 2020—2022 Social Media Study 的代表性样本构造人物画像,并使用 LLM 生成相应的政治关系网络。作者同时评估网络结构、政治同配性,以及个体层面的同党和跨党派接触情况。结果显示,不同实验配置会产生不同程度的党派隔离,部分模型夸大民主党与共和党群体之间的分离,另一些则普遍低估对共和党人的接触。即使两个网络具有相近的整体政治同配性,它们也可能为同一人物生成显著不同的局部政治环境,并对特定社会群体产生系统性高估或低估。
用途与启示
为 LLM 社会仿真建立兼顾整体网络统计与个体局部环境的验证框架。
提醒研究者不要将结构合理或同配性接近真实值直接视为仿真有效。
可用于诊断生成式社会网络中的群体特异性偏差和党派接触失真。
为政治传播、社会网络分析及计算社会科学中的 LLM 仿真提供经验校准依据。
📝 原始笔记(逐字保留)
标题:Validating LLM-generated political networks with survey-reported partisan exposure
来源:arXiv
链接:https://arxiv.org/abs/2610.23199
摘要:Large language models (LLMs) are increasingly used to simulate human behavior in social science research, including decisions, interactions, and social relationships. However, in network formation tasks, existing evaluations often focus on aggregate network properties, while networks with plausible structural properties may still fail to reproduce individuals' social and political environments. Here, we generate LLM-based political networks using personas sampled from the nationally representative ANES 2020--2022 Social Media Study and use respondents' survey-reported partisan exposure as an empirical benchmark for the political environments generated for the corresponding personas. We first characterize the structural properties and political assortativity of the generated networks, and t
标题:JAXtari: High-Throughput and Easy-to-Modify Arcade Learning Environment
来源:arXiv
链接:https://arxiv.org/abs/2610.23650
摘要:The Arcade Learning Environment (ALE) has been a cornerstone of Deep Reinforcement Learning research, providing a standardized benchmark for evaluating general agents. However, its lack of native support for object-extraction and environment modification restricts research into evaluating whether agents learn robust causal models or rely on brittle visual heuristics. Furthermore, the ALE's dependence on CPU-based emulation creates a significant bottleneck in modern GPU-accelerated training pipelines, limiting both experimental scale and iteration frequency. In this work, we introduce JAXtari, an open-source, high-performance reimplementation of Atari 2600 environments. By leveraging massive GPU parallelization, JAXtari allows PPO agents to finish training on 100M environment steps on a sin