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

+86新增论文
+61其中 AI 相关
+14已有论文修订
+93新增 AI 导读

arXiv 最近完成扫描:2026-11-16 01:34 · 全站最近入库:2026-11-16 10:40各信源进度 →

统计口径与数据新鲜度

统计 2026-11-15 10:41 — 2026-11-16 10:41(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:2026-11-16 01:34最近发现新增版本:2026-11-16 10:40库内论文最新更新时间:2026-11-13 16:48

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。

23,434 已收录论文19,641 AI 相关6,765 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 58 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-16 01:33(北京时间)
版本更新 · v2arXiv:2611.12305

HumanSightBench: A Benchmark for Human Visual Perception in MLLMs

Zixin Su, Yangguang Lin, Zegang Peng, Yuntong Sun, Xuanyi Wang, Jiahao Wang, Qiankai Xu, Bing Yuan 等 13 位作者

原文摘要

Multimodal large language models (MLLMs) have demonstrated strong performance on knowledge-intensive and reasoning-based vision-language benchmark datasets, but still struggle to solve the visual questions that people can solve easily. Evaluating the differences between MLLMs and human-level visual perception is challenging due to end-to-end evaluation which combines various skills, or because we are uncertain if individual items are valid. Specifically, observed errors may reflect visual perception, language priors, background knowledge, question ambiguity, and underspecified answers. Our guideline for HumanSightBench is that we construct items that are human-easy by construction and model-hard by evaluation. HumanSightBench contains 2,876 visual questions with short answers across 7 top-level perceptual categories and 21 subcategories. We only include items that pass five inclusion gates: question clarity, answer uniqueness, visual grounding, human solvability, and text-only shortcut filtering. According to the protocol, errors are defined as incorrect answers on human-validated visual tasks with the evaluated interface. The inclusion gates are intended to reduce confounds from ambiguity and external knowledge. The best pass@3 score is 65.6% across eight evaluated MLLMs. No evaluated model exceeds 27.5% on Counting. These results identify capability gaps that are hidden by overall scores and can help future training experiments and held-out evaluation.

cs.CV · cs.AI · cs.CL

版本更新 · v2arXiv:2604.06330

STDec: Spatio-Temporal Stability Guided Decoding for dLLMs

Yuzhe Chen, Jiale Cao, Xuyang Liu, Jin Xie, Aiping Yang, Yanwei Pang

原文摘要

Diffusion Large Language Models (dLLMs) have achieved rapid progress, viewed as a promising alternative to the autoregressive paradigm. However, most dLLM decoders still adopt a global confidence threshold, and do not explicitly model local context from neighboring decoded states or temporal consistency of predicted token IDs across steps. To address this issue, we propose a spatio-temporal stability guided decoding approach, named STDec. We observe strong spatio-temporal stability in dLLM decoding: newly decoded tokens tend to lie near decoded neighbors, and their predicted IDs often remain consistent across several denoising steps. Inspired by this stability, our STDec introduces spatial-aware decoding and temporal-aware decoding. The spatial-aware decoding dynamically generates the token-adaptive thresholds by aggregating the decoded states of nearby tokens. The temporal-aware decoding relaxes the decoding thresholds for tokens whose predicted token IDs remain consistent over denoising steps. Our STDec is training-free and remains compatible with cache-based acceleration methods. Across textual reasoning and multimodal understanding benchmarks, STDec achieves improved inference throughput while maintaining comparable generation score. Notably, on MBPP with LLaDA, STDec achieves up to 14.17$\times$ speedup with a comparable score. The homepage is available at https://yzchen02.github.io/STDec.

cs.CL

新论文 · v1arXiv:2611.13855

Beyond Passive Imitation: On-Policy Mid-Training Distillation for Language Models

Ruize Zheng, Xuyang Liu, Siyu Ren, Xiaobing Tu, Yinggui Wang, Jinkui Ren, Xiantao Zhang, Linfeng Zhang

原文摘要

Continual pre-training (CPT) adapts large language models (LLMs) to specialized domains, yet its standard next-token prediction (NTP) objective treats every corpus token as a fixed target, regardless of its utility to the current model. In a controlled study on the same 1B-token mathematical corpus, we observe sharply different outcomes across base models: NTP improves Llama-3.2-3B's performance on mathematical benchmarks but substantially degrades Qwen3-1.7B-Base's performance, indicating that, under a fixed training recipe, the same training sequences can have model-dependent effects. To address this problem, we introduce On-Policy Mid-Training Distillation (OPMD), which reformulates mid-training from offline text imitation into active, query-driven knowledge acquisition. At selected anchor positions, OPMD samples trajectories from the evolving student policy and applies dense token-level reverse-KL rewards from a frozen teacher, training the model on states it actually visits rather than only on ground-truth prefixes. This on-policy objective reduces the training-generation mismatch of teacher-forced NTP. On MATH-500 and Minerva, OPMD improves the Qwen student over its base performance and outperforms the evaluated NTP baseline, which degrades on both benchmarks. OPMD integrates into CPT pipelines without added inference-time overhead and offers a model-aware alternative to passive domain adaptation.

cs.CL · cs.LG

新论文 · v1arXiv:2611.10414

Track or Follow? A Jev-Compatible Multi-Agent Trading System When Web News Lags the Market

Yebo Feng, Guangyan Gan, Liangrui Pan, Ruichao Liang, Cong Wu, Jiahua Xu, Yang Liu

原文摘要

LLM agents increasingly trade on the Web news they read, yet much Web news about listed firms is published after, and about, the price move it reports, so a reader can agree with the market by following it instead of reading the event. We ask whether LLM judgments of Web news track the market's reaction to the reported event or follow it. To answer, we built a multi-agent trading system whose agents share only point-in-time stores and answer typed questions through the interface of Jev, a hosted typed decision model; code sets every position under a proved bound on how far text can move it. In a hash-committed test of a Jev-compatible reader built from a pinned LLM's option probabilities, item judgments track announcement-window abnormal returns (weekly Spearman 0.119), also after the LLM's release, and event types agree with the issuers' 8-K filings (AUC 0.833); Jev's judgments of news published after its own release track the reaction as well (descriptive). Exploratory analyses show where the association comes from: 72% of earnings news items appear after the issuer's filing, contributor-written items often report the price move, and from the first trade after publication the association falls to 0.023. At a weekly horizon, the confirmatory test detects no rank information over code-only features. Under attack, decision cards traced 88% of successful attacks to the planted item. We propose that market-graded tests of Web-text readers report the criterion from publication, not only from the event.

cs.CL · cs.AI · cs.CR · cs.IR · q-fin.TR

版本更新 · v2arXiv:2611.02540

PixelBench: Can LLMs Understand and Generate Images Pixel by Pixel?

Haoquan Zhang, Feiyang Liu, Juanxi Tian, Wanhua Li, Haodong Duan, Weiyang Liu

原文摘要

Large language models (LLMs) can generate visual assets through text-based formats such as SVG and CAD, but these symbolic representations describe images using high-level primitives rather than explicit raster values. We ask whether LLMs can instead generate pixel images directly through their existing vocabulary, without a learned image tokenizer or decoder. We introduce PixelBench, a benchmark for text-to-pixel (T2P) and image-to-pixel (I2P) generation across multiple resolutions, together with a complementary pixel-understanding evaluation. Models generate an RGB color palette followed by a grid of indices specifying the palette color of each output pixel. In T2P, the strongest models produce recognizable images directly from text. In I2P, the strongest models outperform palette-matched downsampling and specialized pixelization algorithms at low resolutions. Pixel-understanding performance correlates strongly with generation performance. A resolution-scaling study shows that pixel-generation performance often improves initially before declining at high resolutions. Finally, we introduce Pixel Reconstruction Tuning (PRT), which augments visual instruction tuning with caption-palette-grid trajectories. We post-train Molmo2-4B from its pretraining checkpoint by mixing general instruction data with our curated PRT data, improving performance across 16 visual-centric benchmarks, indicating that explicit pixel reconstruction can provide useful auxiliary supervision.

cs.CV · cs.CL · cs.LG

新论文 · v1arXiv:2611.07461

The Last Mile to the Real World: Agent-Native Primitives as Reusable Building Blocks for Solving Real-World Tasks

Haibo Jin, Peng Kuang, Xiaopeng Yuan, Lucheng Fu, Jerry Wang, Xucheng Yu, Yaoning Yu, Dehao Wu 等 14 位作者

原文摘要

How can an agent handle a real-world task end to end? Frontier models with their own harnesses still fall short on real-world benchmarks: on Terminal-Bench 4.0, GPT-5.6 Sol with Codex resolves 37.3% of the tasks and Claude Opus 5 with Claude Code 51.8%. These tasks fail at the step from reaching a resource to applying it, which we call the last mile to the real world. Existing approaches put the resource within reach but leave the competence to apply it with the caller, so each harness rebuilds that competence. We introduce agent-native primitives, reusable units that equip a resource with a resident LLM that applies it in context and communicates with other primitives, and HEELO (Harness Engineering around Evolving Agent-Native Primitives for the Last Mile to Real-WOrld Tasks), a harness engineering framework that builds each task's harness from a library of such primitives. HEELO composes the primitives a task needs into a task-specific graph, verifies the result against evidence from the environment, revises the primitives a failure implicates, and admits new or improved primitives after validation. We instantiate HEELO with eight primitive families, each built for one kind of task, and evaluate them on more than twenty benchmarks. Dev-Primitives raise Terminal-Bench 4.0 resolution from 37.3% under Codex to 55.5% on the same backbone, Harness Primitives composed per task raise Terminal-Bench 2.0 Pass@1 from 60.0% for the seed harness to 72.2%, Tool Primitives raise ToolBench pass rate from 55.8% to 75.1%, and Navigation Primitives raise GAIA-Text-103 accuracy from 45.6% to 58.3%. On the 46 real-world tasks of our heelo.dev platform, Claude Opus 5.5 with Claude Code and GPT-6.1 Sol with Codex complete 9 each on their own and 39 and 35 when connected to HEELO through our MCP server, with 1.9$\times$ and 2.3$\times$ less time and 1.55$\times$ lower cost per completed task.

cs.CL · cs.AI · cs.LG · cs.MA · cs.SE

新论文 · v1arXiv:2611.07228

Source-balanced Capped Fisher: Bounded Evidence Aggregation for User Inference in Language Models

Yang Liu, Ran Zou, Wanrong Zhu

原文摘要

User inference asks whether a source contributed documents to a language-model update, even when the auditor observes only fresh documents from that source. Document-level scores can provide evidence of participation, but combining this evidence is challenging: documents differ in baseline predictability and in how strongly they reflect information learned from other documents of the same source. We propose source-balanced capped Fisher (SCF), an aggregation method that uses known nonparticipating sources to assess how unusual each document score is. It transforms the resulting reference-tail frequencies into negative-log evidence, caps individual contributions, and averages them within each source. This limits the influence of isolated extreme scores while retaining evidence across multiple documents. We complete the inference procedure with a decision rule that tests source participation using a p-value calibrated against a separate set of known nonparticipating sources. The rule provides finite-sample Type-I error control under source exchangeability. Experiments across four real-world datasets show that SCF achieves the highest AUROC among the evaluated aggregation methods in four domains. With the calibration-based decision rule, SCF delivers consistently competitive detection power across the tested settings, achieving the highest power in many cases, while empirical false-positive rates remain generally close to or below the nominal levels.

cs.CL

版本更新 · v2arXiv:2610.40726

Li$^{2}$SAO: Lifting Light Single-Rollout Asynchronous RL Optimization via Group-Relative Actor-Critic Bootstrapping

Zichen Bai, Yanli Wang, Yang Liu, Junhao Shi, Xueqin Liu, Jinguo Sun, Muye Li, Shaoqiang Zheng 等 12 位作者

原文摘要

Asynchronous reinforcement learning (RL) is becoming a mainstream paradigm for post-training large language models (LLMs), as it improves training efficiency by updating the model as rollouts arrive. Recent work argues that group-wise sampling in GRPO does not naturally fit asynchronous RL training because rollout lengths follow a long-tailed distribution, and proposes Single-rollout Asynchronous Optimization (SAO) to improve stability and off-policy behavior, especially on agentic tasks. However, SAO relies on a closed-source value model obtained from scaling value pretraining. In this paper, we analyze GRPO and LiSAO, our lightweight reproduction of SAO without the pretrained value model, on single-turn reasoning and multi-turn agentic tasks. We find that GRPO outperforms LiSAO on reasoning tasks, while the reverse holds on agentic tasks. Since the value model of LiSAO is not yet well trained , LiSAO lags behind GRPO substantially in early training steps. To address this issue, we propose Li$^{2}$SAO, which Lifts Light Single-rollout Asynchronous Optimization with group-relative actor-critic bootstrapping: while the critic is cold, a group-relative advantage trains the actor and a group-centered return trains the critic, after which training switches to single-rollout GAE. Experiments show that Li$^{2}$SAO catches up with or surpasses GRPO in early training and further raises the later performance ceiling. Code is available at https://github.com/jd-cc-coder/Li2SAO.

cs.CL · cs.LG

新论文 · v1arXiv:2611.02017

Switch-SWE: Do Coding Agents Hold Focus When Switching Between Unfinished Tasks?

Shuochen Chang, Qianli Ma, Yibo Miao, Qingyang Liu, Bingjie Gao, Haonan Zhao, Siyu Sun, Yulin Sun 等 10 位作者

原文摘要

Software development sessions often involve several unfinished tasks rather than one issue carried through to completion. As coding agents take on repository-level development, they need to recover an earlier task's requirements and partial progress after intervening work. Existing repository-level benchmarks typically score one issue-to-patch episode and do not directly measure whether an agent can resume a task after advancing other unfinished work in the same repository. We introduce \textbf{\ourmethod}, a benchmark for focus-task resumption in sessions that interleave unfinished tasks drawn from the same source repository, with a persistent checkout for each task. Each focus task is evaluated in matched uninterrupted and switched sessions with identical focus requirements and executable verifiers. The switched session interleaves ordered stages from other unfinished tasks in the same repository, while only the final focus patch is scored. \ourmethod~contains 200 focus tasks from 19 Python repositories, reconstructed from issue, pull-request, and review histories and rendered as 400 matched session specifications. Across 7 coding agents, interleaved unfinished work produces an average focus-task resolution gap of 11.5 percentage points, ranging from 6.0 to 15.0 percentage points. Trajectory analysis identifies cross-task requirement confusion and loss of task state as recurrent resumption behaviors. We introduce a plug-in context-management module that selects task-relevant history and maintains concise task-state summaries. On OpenCode, the module improves focus-task completion under same-repository task switching by 5.5 to 15.5 percentage points across the evaluated models. Code is available at github.com/Miraclecsc/Switch-SWE.

cs.CL

新论文 · v1arXiv:2610.38282

Human Evaluation with Automatic Scores: Separating Sampling and Estimation Gains

Yang Liu, Ran Zou, Zhengtao Gui, Chao Xue

原文摘要

Automatic judges offer inexpensive information about model outputs, but an evaluation team must decide how to turn that information into fewer human annotations. Scores can guide sampling, correct an estimate, or support adaptive acquisition, and these uses may capture overlapping gains. We compare their incremental value on fixed evaluation pools from three preference datasets, and 19 attributed-question-answering systems. Matched comparisons use the same scores, account for pilot labels and budget rules, and separate acquisition from estimation. Score stratification is a strong reference: after ten-bin stratification, the best fixed local linear correction offers only 1.00--5.27% further MSE reduction in the primary configurations. Unequal allocation offers larger gains in some pools, but can increase error when confident scores conceal variable human outcomes. Comparison on AttributedQA, shrinking learned quotas toward score-only quotas lowers pooled MSE for 17 of 19 systems relative to the score-only procedure, but ordinary pooled 95% intervals cover only 88.09--93.54%. The same shrinkage improves on the score-only reference in only one of eight preference configurations. Ranking experiments further show that lower individual MSE need not improve system selection. These findings support a simple starting procedure, proportional sampling within score strata followed by a stratified human mean, and identify what additional evidence is needed to justify correction or adaptation for the intended evaluation output.

cs.CL

版本更新 · v4arXiv:2606.04874

Agent Planning Benchmark: A Diagnostic Framework for Planning Capabilities in LLM Agents

Haoyu Sun, Wenxuan Wang, Mingyang Song, Jujie He, Weinan Zhang, Yang Liu, Yang Yang, Yu Cheng

原文摘要

Planning is central to LLM agents: before acting, an agent must decompose goals, select tools, reason over constraints, and decide when a task is infeasible. Yet existing agent evaluations often report only end-to-end success, making it difficult to determine whether failures stem from planning or execution. We introduce Agent Planning Benchmark (APB), a planning-specific diagnostic benchmark with 4,209 multimodal cases across 22 domains and five settings, covering holistic planning, feedback-conditioned step-wise planning, and robustness under extraneous tools, broken tools, and unsolvable tasks. Across 12 MLLMs, APB reveals systematic weaknesses in long-horizon planning, tool-noise robustness, calibrated refusal, and inference-time refinement. We further validate APB on 200 ToolSandbox tasks and 200 $τ^2$-bench tasks, where APB-guided refinement consistently improves plan correctness, plan grade, and downstream execution metrics across three representative models. APB thus serves as an upstream diagnostic complement to execution benchmarks. The APB benchmark and code are available in \href{https://github.com/Mikivishy/AgentPlanningBenchmark}{this URL}.

cs.CL

版本更新 · v5arXiv:2610.04906

Scaling Verifiable Environments for Long-horizon Work Agents

Jiazheng Zhang, Long Ma, Yunxian Yang, Zhiheng Xi, Zhikai Lei, Yajie Yang, Chenyang Liao, Enyu Zhou 等 26 位作者

原文摘要

Work agents operate over digital artifacts to execute professional knowledge-intensive work, requiring training environments that support long-horizon interaction and trustworthy verification. However, hand-crafted environments incur prohibitive engineering overhead that prevents environment scaling, whereas synthesis methods sacrifice workspace complexity, realism, or grounded verifiability. To bridge this gap, we introduce WorkForge, a scalable synthesis framework for constructing verifiable work-agent environments from real-world resources. Starting from expert workflows, WorkForge first identifies the resources, decisions, and deliverables required by each workflow. It then retrieves relevant real-world files and organizes them into a workspace. WorkForge inspects the workspace to extract concrete, checkable facts about its content. These factual anchors fix which task types the workspace can support and how their outcomes can be verified. Therefore, WorkForge derives each task's instructions, solution plan, and complementary programmatic and semantic verifiers directly from these factual anchors, keeping verification traceable to observable workspace evidence. Furthermore, we construct 16.7K verifiable environments across 40 professional domains, with workspaces collectively covering 60 file types. Post-training Qwen3.5-35B-A3B-Base improves GDPVal from 45.5 to 73.6 and APEX Score from 5.0 to 21.3, while enabling Qwen3.5-27B to achieve highly competitive performance and outperform strong competitors. Our analyses confirm the efficacy of the proposed method and reveal consistent scaling behaviors across both data volume and interaction horizons.

cs.CL · cs.AI

新论文 · v1arXiv:2610.34887

NumTrace: Auditing the Number Provenance of Financial Analyst Agents

Tianlei Zhu, Matt White, Xue Liu, Xiao-Yang Liu

原文摘要

A numerical claim generated by a financial analyst agent can match a number in a company filing but refer to the wrong entity, concept, period, unit or scope. A check based on reference answers can mark such an answer as wrong but does not explain why it fails. In this paper, we present NumTrace, a new benchmark that audits the provenance of each numerical claim made by financial analyst agents: its source and financial semantics. NumTrace currently has 540 questions over 2,163 SEC EDGAR documents. To the best of our knowledge, it is the first numerical benchmark about who holds a company's shares (Form 13F) and what its executives are owed (Item 402(i)). On a frozen snapshot of these documents, we build the reference answers according to accounting rules. NumTrace then gives each claim three separate verdicts: value, binding and support. By retaining separate value and binding verdicts, NumTrace detects right number, wrong account (RNWA) errors: a claim has an RNWA error when it passes the value check but fails the binding check. The RNWA rate quantifies the gap between value-only and joint value-and-binding pass rates. Finally, we audit three agents, Claude Opus 5.5, GPT-6 Astra and DeepSeek-V4.1-Flash. RNWA errors make up 67.5-81.8% of each agent's failing claims. In particular, on 13F co-filing group questions, about 91% of claims pass the value check, but fewer than half pass the binding check.

cs.CL · cs.AI

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-demo.17ACL Anthology · ACL 2026

RiskLab: A Controlled Toolkit for Probing Emergent Risks in LLM-Based Multi-Agent Systems

Yu Jiang, Wenjie Wang, Yue Huang University of Notre Dame, Yanbo Wang Mohamed bin Zayed University of Artificial Intelligence, Zhenhong Zhou, Xiuying Chen Mohamed bin Zayed University of Artificial Intelligence, Yang Liu Nanyang Technological University, Pin-Yu Chen International Business Machines 等 10 位作者

原文摘要

Large language model (LLM) agents increasingly operate in multi-agent settings where failures emerge from interaction dynamics rather than isolated model errors. We introduce RiskLab, an open-source toolkit for instantiating, probing, and measuring emergent risks in LLM-based multi-agent systems under controlled conditions. Each experiment is defined as a structured topology–environment–protocol–agent–task quintuple, enabling reproducible studies of how communication structure, coordination mechanisms, and incentives shape system-level risks. RiskLab provides flexible communication topologies, swappable interaction protocols, trajectory-grounded evaluation, and extensible registries for risk detectors and agent backends. We demonstrate the toolkit across representative risks, including collusion, resource overreach, semantic drift, and strategic misreporting, and support one-file reproducibility via configuration.

cs.CL · cs.MA

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-demo.35ACL Anthology · ACL 2026

MASFactory: A Graph-centric Framework for Orchestrating LLM-Based Multi-Agent Systems with Vibe Graphing

Yang Liu, Jinxuan Cai, Yishen Li, Qi Meng Beijing University of Posts and Telecommunications, Zedi Liu, Xin Li Free Org., Chen Qian Shanghai Jiaotong University, Chuan Shi Beijing University of Post and Telecommunication 等 9 位作者

原文摘要

Large language model-based (LLM-based) multi-agent systems (MAS) are increasingly used to extend agentic problem solving via role specialization and collaboration. MAS workflows can be naturally modeled as directed computation graphs, where nodes execute agents or sub-workflows and edges encode dependencies and message passing. However, implementing complex graph workflows in current frameworks still requires substantial manual effort, offers limited reuse, and makes it difficult to integrate heterogeneous external context sources. To overcome these limitations, we present MASFactory, a graph-centric framework for orchestrating LLM-based MAS. It introduces Vibe Graphing, a human-in-the-loop approach that compiles natural-language intent into an editable workflow specification and then into an executable graph. In addition, the framework provides reusable components, skill support, multimodal message handling, and pluggable context integration, as well as a visualizer for topology preview, runtime tracing, and human-in-the-loop interaction. We evaluate MASFactory on seven public benchmarks, validating both reproduction consistency for representative MAS methods and the effectiveness of Vibe Graphing. Our code (https://github.com/BUPT-GAMMA/MASFactory, licensed under Apache-2.0) and video demonstration (https://youtu.be/ANynzVfY32k) are publicly available.

cs.CL · cs.AI · cs.MA · cs.CV

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-industry.65ACL Anthology · ACL 2026

ACRM: Multi-Agent Trajectory Learning for Automated Credit Risk Model Refreshing in Production

Liangzu Liu, Mengzhe Ruan, Xiaotian Chen, HaonanChen, XudongNiu, Wendi Yuan, YuechenLi, Yang Liu 等 9 位作者

原文摘要

Credit risk models suffer from rapid performance decay due to distribution shifts, requiring frequent updates to meet strict operational guardrails. However, manual refreshing takes weeks of trial-and-error across upstream data engineering and downstream training. We present ACRM, a deployed multi-agent framework that automates the end-to-end credit modeling workflow by treating it as a learnable trajectory of agent interactions. Unlike AutoML, which optimizes hyperparameters on fixed datasets, ACRM’s action space extends to upstream data semantics—cohort selection, observation windowing, feature screening—where the majority of performance recovery occurs. A central Orchestrator coordinates specialist agents through a three-stream decision stack: rule-based safety guardrails, retrieval-augmented grounding from historical workflows, and preference alignment via DPO on expert-labeled trajectories. Deployed at a major fintech institution for three months across six business scenarios, ACRM reduced the average model refresh cycle from weeks to 1.1 days and iteration rounds by 65%, while maintaining superior stability metrics.

cs.CL · cs.MA

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-industry.83ACL Anthology · ACL 2026

FlowHN: Adaptive Token Routing for Efficient Parallel Hybrid Networks

Mohammad Mahdi Moradi, Walid Ahmed, Shuangyue Wen, Sudhir Mudur, Weiwei Zhang, Yang Liu

原文摘要

Production LLMs must balance modeling quality with predictable latency, stable accelerator utilization, and cost-efficient scaling—constraints that remain difficult for existing architectures. Transformers provide strong reasoning but incur quadratic complexity, while state-space models (SSMs) scale efficiently yet lack fine-grained interactions; prior hybrids either introduce sequential bottlenecks or rely on learned routing that complicates deployment. We present FlowHN, a deployment-oriented parallel hybrid architecture that enables deterministic conditional computation via FLOP-aware token circulation across attention and SSM branches. Instead of dynamic expert routing, FlowHN performs hardware-aligned token scheduling that balances workloads, reduces synchronization stalls, and preserves full parameter utilization. Across 135M–1B models, FlowHN achieves up to 4× higher throughput and 15% higher MFU than strong Transformer, SSM, and hybrid baselines while maintaining competitive accuracy on reasoning, coding, and long-context tasks up to 32K tokens. FlowHN is designed to integrate directly into existing Hybrid pipelines without changes to optimizers, training stacks, or inference serving infrastructure, making it practical for real-world deployment.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.1021ACL Anthology · ACL 2026

OS-Symphony: A Holistic Framework for Robust and Generalist Computer-Using Agents

Bowen Yang, Kaiming Jin national university of singaore, National University of Singapore, Zhenyu Wu, Zhaoyang Liu Hong Kong University of Science and Technology and Shanghai AI Laboratory, Qiushi Sun University of Hong Kong, Zehao Li University of Science and Technology of China, JingJing Xie, Zhoumianze Liu 等 15 位作者

原文摘要

While Vision-Language Models (VLMs) have significantly advanced Computer-Using Agents (CUAs), current agentic frameworks struggle with robustness in novel domains and long-horizon workflows due to the absence of visual-aware tutorial retrieval and the lack of granular control over historical visual context curation and pruning. To bridge these gaps, we introduce OS-Symphony, a holistic framework that comprises an Orchestrator coordinating two key innovations for robust automation: (1) a Reflection-Memory Agent that utilizes milestone-driven long-term memory to enable trajectory-level self-correction, effectively mitigating visual context loss in long-horizon tasks; (2) Versatile Tool Agents featuring a Multimodal Searcher that adopts a “SeeAct” paradigm to navigate a browser-based sandbox to synthesize live, visually aligned tutorials, thereby resolving fidelity issues in unseen scenarios. Experimental results demonstrate that OS-Symphony delivers substantial performance gains across varying model scales, establishing new state-of-the-art results on three online benchmarks, notably achieving 65.84% on OSWorld. All research assets will be made publicly available.

cs.CL · cs.AI · 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.1068ACL Anthology · ACL 2026

MMTutorBench: The First Multimodal Benchmark for AI Math Tutoring

Tengchao Yang, Sichen Guo, Mengzhao Jia, Jiaming Su, Yuanyang Liu, Zhihan Zhang Amazon, Meng Jiang University of Notre Dame

原文摘要

Effective math tutoring requires not only solving problems but also diagnosing students’ difficulties and guiding them step by step. While multimodal large language models (MLLMs) show promise, existing benchmarks largely overlook these tutoring skills. We introduce MMTutorBench, the first benchmark for AI math tutoring, consisting of 770 problems built around pedagogically significant key-steps. Each problem is paired with problem-specific rubrics that enable fine-grained evaluation across six dimensions, and structured into three tasks—Insight Discovery, Operation Formulation, and Operation Execution. We evaluate 12 leading MLLMs and find clear performance gaps between proprietary and open-source systems, substantial room compared to human tutors, and consistent trends across input variants: OCR pipelines degrade tutoring quality, few-shot prompting yields limited gains, and our rubric-based LLM-as-a-Judge proves highly reliable. These results highlight both the difficulty and diagnostic value of MMTutorBench for advancing AI tutoring.

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.1233ACL Anthology · ACL 2026

False Friends in the Shell: Unveiling the Emoticon Semantic Confusion in Large Language Models

Weipeng Jiang, Xiaoyu Zhang Nanyang Technological University, Juan Zhai University of Massachusetts at Amherst, Shiqing Ma University of Massachusetts at Amherst, Chao Shen Xi’an Jiaotong University, Yang Liu Nanyang Technological University

原文摘要

Emoticons are widely used in digital communication to convey affective intent, yet their safety implications for Large Language Models (LLMs) remain largely unexplored. In this paper, we identify emoticon semantic confusion, a vulnerability where LLMs misinterpret ASCII-based emoticons to perform unintended and even destructive actions. To systematically study this phenomenon, we develop an automated data generation pipeline and construct a dataset containing 3,757 code-oriented test cases spanning 21 meta-scenarios, four programming languages, and varying contextual complexities. Our study on six LLMs reveals that emoticon semantic confusion is pervasive, with an average confusion ratio exceeding 38%. More critically, over 90% of confused responses yield ’silent failures’, which are syntactically valid outputs but deviate from user intent, potentially leading to destructive security consequences.Furthermore, we observe that this vulnerability readily transfers to popular agent frameworks, while existing prompt-based mitigations remain largely ineffective. We call on the community to recognize this emerging vulnerability and develop effective mitigation methods to uphold the safety and reliability of human-LLM interactions.

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.1276ACL Anthology · ACL 2026

Trust Within? Seek Beyond? Knowledge Boundary Aware Policy Optimization for Agentic Search

Tao Feng Zhejiang University, Xinke Jiang, Xinyan Hu, Yonggang Zhang, Zhen Tao Harbin Institute of Technology, Shenzhen, Wentao Zhang, Boyang Liu, Wenhao Jiang Harbin Institute of Technology 等 9 位作者

原文摘要

Agentic search augments large language models (LLMs) with external knowledge through reinforcement learning. However, existing approaches suffer from blind reliance on noisy retrieval and hallucination when both parametric and external knowledge fail—reflecting a lack of calibration regarding the model’s knowledge boundary. We propose Knowledge boundary Policy Optimization (KbPO), a reinforcement learning framework that explicitly aligns retrieval decisions with quantified knowledge states. KbPO introduces: (1) a semantic stability metric to delineate reliable parametric knowledge; (2) a four-quadrant taxonomy synthesising internal certainty with retrieval quality; and (3) a quadrant-based reward mechanism incentivising calibrated behaviour. We further adopt an iterative query evolution pipeline to construct boundary-probing training samples. Experiments on ten benchmarks demonstrate that KbPO outperforms strong baselines while exhibiting reduced hallucination rates.

cs.CL · cs.AI

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.1283ACL Anthology · ACL 2026

OptiCo: Adaptive Distributed Training Optimization via Collaborative Agent Reasoning

Sheng Chen Zhejiang lab, Tang Zhe, Weixing Zhang Zhejiang Lab, Fei Yang Zhejiang Lab, Yuanyuan Wang, Tianlin Li Beihang University, Yang Liu Nanyang Technological University

原文摘要

Optimizing distributed training strategies for large-scale deep learning models remains a critical challenge in both industry and academia, demanding extensive domain expertise and manual tuning. Existing automated distributed training frameworks are plagued by over-reliance on prior profiling, poor generalization across models/hardware, and scalability constraints stemming from vast search spaces, impeding real-world applicability. To address these challenges, we propose OptiCo, a model-driven multi-agent framework that leverages Large Language Models (LLMs) to enable automatic and explainable distributed training strategy configuration. OptiCo orchestrates a team of reasoning-driven agents, through a shared Global Message Pool facilitating persistent memory and coordination. By employing inception prompting and Chain-Of-Thought (COT) reasoning, agents iteratively refine configurations, detect bottlenecks, analyze failures, and optimize resource utilization. Evaluated across 25+ configurations spanning diverse model architectures, GPU types and scales, OptiCo outperforms expert-designed strategies within 20 iterations, achieving an average performance improvement of 1.84%, with gains ranging from 0.08% to 8.65%. The source codes are avaiable at https://github.com/TangZhe96/OptiCo-public.

cs.CL · cs.MA

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.1329ACL Anthology · ACL 2026

Learn Like Humans: Use Meta-cognitive Reflection for Efficient Self-Improvement

Xinmeng Hou, Bohao Qu A*STAR, Wuqi Wang Chang 'an University, Peiliang Gong Singapore University of Technology and Design and Nanyang Technological University, Qing Guo Nankai University and National University of Singapore, Yang Liu Nanyang Technological University

原文摘要

While Large Language Models (LLMs) enable complex autonomous behavior, current agents remain constrained by static, human-designed prompts that limit adaptability. Existing self-improving frameworks attempt to bridge this gap but typically rely on inefficient, multi-turn recursive loops that incur high computational costs. To address this, we propose Metacognitive Agent with Reflective Self-improvement (MARS), a framework that achieves efficient self-evolution within a single recurrence cycle. Inspired by educational psychology, MARS mimics human learning by integrating principle-based reflection (abstracting normative rules to avoid errors) and procedural reflection (deriving step-by-step strategies for success). By synthesizing these insights into optimized instructions, MARS allows agents to systematically refine their reasoning logic without continuous online feedback. Extensive experiments on six benchmarks demonstrate that MARS outperforms state-of-the-art self-evolving systems while significantly reducing computational overhead. Code is available at https://github.com/Paparare/MARS/tree/main

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.1349ACL Anthology · ACL 2026

MARS-RA: Rank Aggregation for Credit Assignment via Multimodal Comparisons in Embodied Multi-Agent Cooperation

Dawei Wang, Di Zhao University of Auckland, Xinyuan Liu, Marci Chi Ma Newcastle University, UK, Xiaoyang Liu Newcastle University, UK, Chengming Zhou, Gary Ushaw Newcastle University, UK, Richard Davison University of Newcastle-upon-Tyne

原文摘要

Credit assignment is a fundamental challenge in cooperative multi-agent reinforcement learning, particularly in embodied AI settings characterized by limited and delayed feedback as well as dynamically changing numbers of active agents. We propose MARS-RA, a framework that reformulates credit assignment as a rank aggregation problem using contribution-based pairwise comparisons among agents generated by large multimodal models. This shift from absolute to relative estimation ensures robustness against noise and dynamic agent participation, converting comparison results into contribution scores for potential-based reward shaping. We provide theoretical justification for the convergence and robustness of the proposed framework, and show that Shapley values can be used as an interpretive reference. Experimental results on challenging tasks of different types indicate that MARS-RA can guide agents toward effective cooperation.

cs.CL · 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.1393ACL Anthology · ACL 2026

Why Supervised Fine-Tuning Fails to Learn: A Systematic Study of Incomplete Learning in Large Language Models

Chao Xue, Yao Wang, Mengqiao Liu, Di Liang, Xingsheng Han, Peiyang Liu, Xianjie Wu Beijing Information Science & Technology University, Chenyao Lu Tencent AI Lab 等 14 位作者

原文摘要

Supervised Fine-Tuning (SFT) is the standard approach for adapting large language models (LLMs) to downstream tasks. However, we observe a persistent failure mode: even after convergence, models often fail to correctly reproduce a subset of their own supervised training data. We refer to this behavior as the Incomplete Learning Phenomenon (ILP). This paper presents the first systematic study of ILP in LLM fine-tuning. We formalize ILP as post-training failure to internalize supervised instances and demonstrate its prevalence across multiple model families, domains, and datasets. Through controlled analyses, we identify five recurrent sources of incomplete learning: (1) missing prerequisite knowledge in the pre-trained model, (2) conflicts between SFT supervision and pre-training knowledge, (3) internal inconsistencies within SFT data, (4) left-side forgetting during sequential fine-tuning, and (5) insufficient optimization for rare or complex patterns. We introduce a diagnostic-first framework that maps unlearned samples to these causes using observable training and inference signals, and study several targeted mitigation strategies as causal interventions. Experiments on Qwen, LLaMA, and OLMo2 show that incomplete learning is widespread and heterogeneous, and that improvements in aggregate metrics can mask persistent unlearned subsets. The findings highlight the need for fine-grained diagnosis of what supervised fine-tuning fails to learn, and why.

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.1436ACL Anthology · ACL 2026

Rethinking Entropy Interventions in RLVR: An Entropy Change Perspective

Zhezheng Hao, Hong Wang, Haoyang Liu, Jian Luo State University of New York at Stony Brook, Jiarui Yu Tencent, Hande Dong, Qiang Lin, Can Wang 等 9 位作者

原文摘要

Reinforcement Learning with Verifiable Rewards (RLVR) serves as a cornerstone technique for enhancing the reasoning capabilities of Large Language Models (LLMs). However, its training is often plagued by entropy collapse, a rapid decline in policy entropy that limits exploration and undermines training effectiveness. While recent works attempt to mitigate this issue via several heuristic entropy interventions, the underlying mechanisms remain poorly understood. In this work, we conduct comprehensive theoretical and empirical analyses of entropy dynamics in RLVR, offering two main insights: (1) We derive a tight approximation for token-level entropy change at each update step, revealing four governing factors and providing a unified theoretical framework of how existing methods influence entropy; (2) We reveal a fundamental limitation of recent approaches: they rely on heuristic adjustments to one or two of these factors, leaving other relevant factors unconsidered, thus inherently limiting their effectiveness. Motivated by these findings, we propose STEER, a principled entropy-modulation method that adaptively reweighs tokens based on theoretically-estimated entropy variations. Extensive experiments across six mathematical reasoning and three coding benchmarks demonstrate that STEER effectively mitigates entropy collapse and consistently outperforms state-of-the-art baselines.

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.1516ACL Anthology · ACL 2026

Parameter Importance is Not Static: Evolving Parameter Isolation for Supervised Fine-Tuning

Zekai Lin University of Glasgow, Chao Xue, Di Liang, Xingsheng Han, Peiyang Liu, Xianjie Wu Beijing Information Science & Technology University, Lei Jiang Tencent AI Lab, Yu Lu 等 11 位作者

原文摘要

Supervised Fine-Tuning (SFT) of large language models often suffers from task interference and catastrophic forgetting. Recent approaches alleviate this issue by isolating task-critical parameters during training. However, these methods represent a static solution to a dynamic problem, assuming that parameter importance remains fixed once identified. In this work, we empirically demonstrate that parameter importance exhibits temporal drift over the course of training. To address this, we propose Evolving Parameter Isolation (EPI), a fine-tuning framework that adapts isolation decisions based on online estimates of parameter importance. Instead of freezing a fixed subset of parameters, EPI periodically updates isolation masks using gradient-based signals, enabling the model to protect emerging task-critical parameters while releasing outdated ones to recover plasticity. Experiments on diverse multi-task benchmarks demonstrate that EPI consistently reduces interference and forgetting compared to static isolation and standard fine-tuning, while improving overall generalization. Our analysis highlights the necessity of synchronizing isolation mechanisms with the evolving dynamics of learning diverse abilities.

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.1561ACL Anthology · ACL 2026

Observations and Remedies for Large Language Model Bias in Self-Consuming Performative Loop

Yaxuan Wang, Zhongteng Cai, Yujia Bao Accenture, Xueru Zhang Ohio State University, Yang Liu University of California, Santa Cruz

原文摘要

The rapid advancement of large language models (LLMs) has led to growing interest in using synthetic data to train future models. However, this creates a self-consuming retraining loop, where models are trained on their own outputs and may cause performance drops and induce emerging biases. In real-world applications, previously deployed LLMs may influence the data they generate, leading to a dynamic system driven by user feedback. For example, if a model continues to underserve users from a group, less query data will be collected from this particular demographic of users. In this study, we introduce the concept of Self-Consuming Performative Loop (SCPL) and investigate the role of synthetic data in shaping bias during these dynamic iterative training processes under controlled performative feedback. This controlled setting is motivated by the inaccessibility of real-world user preference data from dynamic production systems, and enables us to isolate and analyze feedback-driven bias evolution in a principled manner. We focus on two types of loops, including the typical retraining setting and the incremental fine-tuning setting, which is largely underexplored. Through experiments on three real-world tasks, we find that the performative loop increases preference bias and decreases disparate bias. We design a reward-based rejection sampling strategy to mitigate the bias, moving towards more trustworthy self-improving systems. The code is available at https://github.com/UCSC-REAL/SCPL.git.

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.1568ACL Anthology · ACL 2026

Unlocking the Black Box of Latent Reasoning: An Interpretability-Guided Approach to Intervention

Shuochen Chang Shanghai Jiaotong University and ByteDance Inc., Tong Bai, Xiaofeng Zhang, Qianli Ma Shanghai Jiao Tong University, Qingyang Liu, Zhaohe Liao Alibaba Group and Shanghai Jiao Tong University, Yibo Miao Moonshot AI, Li Niu Shanghai Jiao Tong University

原文摘要

Latent reasoning enables Large Language Models (LLMs) to perform multi-step inference within continuous hidden states, offering efficiency gains over explicit Chain-of-Thought (CoT). However, the opacity of these continuous thought vectors hinders their reliability and controllability. This paper bridges the gap between mechanistic interpretability and actionable control. We first present a systematic analysis using structural, causal, and geometric probes, revealing that latent vectors encode compressed, faithful representations of reasoning steps, with early vectors acting as critical causal hubs. Building on this, we operationalize these interpretability insights into a suite of training-free, decode-time interventions that refine the latent reasoning process by imposing the identified geometric and semantic priors. Extensive experiments across multiple model scales and diverse task domains demonstrate that our approaches consistently improve reasoning accuracy. Our interpretability-guided interventions consistently unlock latent capabilities and improve reasoning accuracy without any parameter updates.

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.1615ACL Anthology · ACL 2026

S2S-Arena: Evaluating Paralinguistic Instruction Following in Speech-to-Speech Models

Feng Jiang Shenzhen University of Advanced Technology, Zhiyu Lin, Yiyang Liu Shenzhen University of Advanced Technology, Liumeng Xue, Fan Bu, Yuhao Du, Xiangying Chen, Benyou Wang The Chinese University of Hong Kong, Shenzhen 等 9 位作者

原文摘要

Recent advances in large language models (LLMs) have fundamentally reshaped speech-to-speech (S2S) systems, enabling increasingly natural spoken interaction. However, existing benchmarks still rely heavily on text-based evaluation and largely ignore paralinguistic cues such as prosody, emotion, and speaker traits, which are central to expressive and human-like communication. We introduce S2S-Arena, a speech-native benchmark for evaluating instruction-following S2S models with explicit assessment of both semantic understanding and paralinguistic expression. S2S-Arena features a four-level interaction protocol that systematically probes models under increasing paralinguistic complexity, a two-stage data construction pipeline that produces 1,243 speech samples spanning 100+ real-world tasks, and an arena-style evaluation framework that enables reference-free, pairwise comparison directly in the speech modality. Benchmarking 10 state-of-the-art S2S systems over 1,000+ comparisons reveals substantial performance gaps (especially under complex paralinguistic demands) between current academic and industrial systems. Our analysis further identifies key design factors governing expressive instruction following, providing actionable insights for building more natural, robust, and human-aligned speech agents.

cs.CL

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