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

+1084新增论文
+883其中 AI 相关
+130已有论文修订
+94新增 AI 导读

arXiv 最近完成扫描:2026-11-18 04:28 · 全站最近入库:2026-11-18 19:45各信源进度 →

统计口径与数据新鲜度

统计 2026-11-18 06:25 — 2026-11-19 06:25(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:2026-11-18 04:28最近发现新增版本:2026-11-18 19:45库内论文最新更新时间:2026-11-18 08:00

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。

26,870 已收录论文22,374 AI 相关8,156 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 11 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-18 04:27(北京时间)
新论文 · v1arXiv:2611.17825

Beyond Endpoints: Bézier Paths for Vision Foundation Model Adaptation

Chenzhi Zheng, Xi Xiao, Chih-Ting Liao, Qizhen Lan, Jingwen Lin, Tianyang Wang, Hao Xu, Xin Cao

原文摘要

Frozen vision foundation models can be adapted by learning a few prompt tokens shared across all images. When prompts are updated as data shift or tasks change, the distribution of image features moves from one endpoint to another. Prior theory asks which endpoints prompts can reach; we ask whether they can travel between them along the shortest path. Because shared prompts compete with each image's own content for attention, they often cannot. We quantify this restriction as the extra transport cost that shared attention imposes over the shortest path. In an exactly solvable attention model, the reachable features form a cone; unrolling it into a plane gives the exact minimum action for every pair of endpoints and explicit optimal prompts on short arcs. The analysis reveals a sharp threshold in content normalization, which measures how much attention each image's own tokens absorb: for same-sign endpoints, at most two distinct levels allow prompts to follow the shortest path, whereas three or more force a strictly positive extra cost unless the endpoints differ only by rescaling. This gap persists with more prompts, small content perturbations, and aligned multihead outputs. On ViT-B/16 with VTAB-1k, a quadratic Bézier prompt path that minimizes feature action reduces held-out transport action relative to linear and cross-entropy-trained paths between the same endpoints and improves intermediate and online accuracy under unlabeled distribution shift, while cross-entropy-trained paths retain higher minimum accuracy when labeled calibration data are available. Reachable endpoints do not guarantee a reachable shortest path: how prompts move matters, not only where they end. Project page: https://xixiaouab.github.io/Beyond-Endpoints/.

cs.CV

版本更新 · v2arXiv:2506.02864

BNPO: Beta Normalization Policy Optimization

Changyi Xiao, Mengdi Zhang, Yixin Cao

原文摘要

Reinforcement learning with rule-based, binary-valued reward functions has been shown to significantly enhance the reasoning capabilities of large language models. These models typically rely on REINFORCE-based policy optimization methods, such as REINFORCE with baseline and group relative policy optimization (GRPO). However, a key limitation is that existing methods either ignore reward normalization or use static normalization schemes, which fail to adapt to the evolving policy during training. This mismatch can lead to unstable gradient estimates and hinder training stability. To address this issue, we propose Beta Normalization Policy Optimization (BNPO), a novel method that adaptively normalizes rewards using a Beta distribution with dynamically updated parameters. By aligning normalization with the changing policy distribution, BNPO enables more accurate, lower-variance gradient estimation and promotes stable training dynamics. We provide theoretical analysis showing that BNPO reduces variance and generalizes both REINFORCE and GRPO under binary-valued rewards. In addition, we introduce an advantage decomposition mechanism to extend BNPO to more complex reward settings. Experimental results demonstrate that BNPO achieves state-of-the-art performance among policy optimization methods on reasoning tasks.

cs.LG · cs.AI

新论文 · v1arXiv:2611.07566

Guidance Is a Field, Not a Scalar: Interval-Integrated Classifier-Free Guidance for One-Step Video Generation

Zijie Meng, Xixin Cao

原文摘要

One-step generators built on MeanFlow synthesize a video in a single function evaluation, yet inherit a rigid treatment of classifier-free guidance (CFG): the guidance scale is frozen during training or, at best, exposed as a single scalar. This conflicts with mounting evidence that the optimal guidance is not a constant but a field that varies with the noise level, with spatial content, and, for video, with frame position. Every existing remedy presupposes a multi-step sampler along which guidance can be re-scheduled--an assumption one-step models violate by construction. We resolve this tension through a single observation: because the MeanFlow target is an integral over a noise interval, an arbitrary guidance field can be absorbed into the target itself and compiled into one evaluation. We formalize this as the interval-integrated CFG identity, prove that the compiled guidance carries no schedule-discretization error and can thus match or exceed any finite multi-step sampler, and instantiate it as AGAVE, which equips the average-velocity field with a learnable temporal schedule, a free motion-saliency spatial modulation, and a frame-position asymmetry. Adding fewer than ten scalars at zero inference cost, AGAVE surpasses all one-step baselines on VBench, UCF-101 and MSR-VTT while matching multi-step guided sampling. Project repository: https://github.com/YMlinfeng/AGAVE.

cs.CV

新论文 · v1arXiv:2611.04784

Optimal Approximation for Constrained Correlation Clustering via Repulsion

Yixin Cao, Ying Xu

原文摘要

The best-known approximation factor for constrained correlation clustering stood at three for nearly two decades, until a recent improvement to below $16/7$. We give a randomized polynomial-time $(2+\varepsilon)$-approximation using \emph{Repulsion}, a recursive operation that separates a metric ball from its complement. Repulsion naturally respects hard constraints and admits a short potential-based analysis; as a warm-up, radius-$2/3$ Repulsion rounds the standard metric LP into a feasible partition of cost at most three times its objective. The $(2+\varepsilon)$-approximation takes a two-stage approach, first removing must-links with arbitrarily small loss in the approximation ratio. The second stage combines lossless, constraint-aware preclustering with the bounded sub-cluster LP framework of Cao, Cohen-Addad, Lee, Li, Newman, and Vogl (STOC 2024). Repeated positive conditioning and sampling produce candidate sets, which radius-one Repulsion partitions into feasible clusters at controlled cost. We also give an approximation-preserving reduction from vertex cover to cluster deletion, a special case with only cannot-links. Apart from establishing a matching factor-two hardness threshold under the Unique Games Conjecture, it motivates our two-stage approach and Repulsion.

cs.DS

新论文 · v1arXiv:2610.39761

GPU-Resident Speculative Decoding for In-Browser LLM Inference with WebGPU

Wei Song, Yuxin Cao, Ziqi Ding, Xiao Cheng

原文摘要

WebGPU enables private, offline LLM inference inside web pages, but browser dispatch, inter-process communication, synchronous readback and single-token kernels slow decoding and make conventional speculative decoding inefficient. We present a WebGPU runtime that addresses these browser-specific costs. Deterministic small-batch integer kernels accelerate multi-token verification while preserving exact agreement with greedy decoding in the same quantized runtime. A GPU-resident loop performs drafting, verification, acceptance and KV-cache maintenance without a per-token CPU decision. A draft-free context-mixing drafter reuses page content, conversation history and session memory, while a device-aware controller adapts the speculation width and the submission window to measured costs. Across seven models from 0.5B to 8B parameters, 11 tasks and three client GPUs, our runtime decodes 3.5 to 5.4$\times$ faster than WebLLM on a laptop GPU, 4.8 to 6.2$\times$ faster on an integrated GPU and 4.8 to 8.6$\times$ faster on a workstation GPU. The GPU-resident loop and speculation contribute a further 1.8 to 3.2$\times$ over the optimized kernels on the workstation. Outputs match the quantized greedy decoding of the runtime exactly, and the mean GSM8K accuracy matches that of WebLLM. Optional draft heads raise the peak speedup to 10.2$\times$, while session memory provides additional gains without downloading extra weights.

cs.CY

新论文 · v1arXiv:2610.39358

Before the Next Rollout: Diagnosing Skill Utility Shifts from Policy Updates

Yifan Wang, Zhenyi Gai, Yiwen Zhao, Mingxuan Jiang, Keyang Chen, Zeping Li, Hongfeng Chai, Yixin Cao

原文摘要

Skill-evolution methods typically analyze execution trajectories to guide agent skill refinement. However, policy updates can change the behavioral effects and marginal utility of an unchanged skill. Attribution based on earlier trajectories may therefore no longer reflect the skill's contribution under the updated policy. Moreover, trajectory failures can arise from multiple factors, making individual skill contributions difficult to isolate. We evaluate whether policy-update signals contain predictive information about changes in skill marginal utility. At fixed contexts from pre-update training rollouts, we measure how policy updates change the contrast in token preferences between skill and control conditions. Combining these interaction shifts with advantage estimates yields reward-directed readouts for diagnosing utility changes. The readouts reuse existing training rollouts and require only forward passes for single-step action scoring, without additional post-update rollout evaluation. Our experiments show that policy updates can alter skill marginal utility and that policy-update readouts carry diagnostic information for identifying skills with declining utility. Integrating readout-guided prioritization into skill evolution improves downstream performance while reducing policy and editor token consumption.

cs.AI

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.1288ACL Anthology · ACL 2026

Rethinking the Role of Entropy in Optimizing Tool-Use Behaviors for Large Language Model Agents

Zeping Li, Hongru Wang, Yiwen Zhao, Guanhua Chen Southern University of Science and Technology, Yixia Li, Keyang Chen Fudan University, Yixin Cao Fudan University, Guangnan Ye Fudan University 等 10 位作者

原文摘要

Tool-using agents based on Large Language Models (LLMs) excel in tasks such as mathematical reasoning and multi-hop question answering. However, in long trajectories, agents often trigger excessive and low-quality tool calls, increasing latency and degrading inference performance, making managing tool-use behavior challenging. In this work, we conduct entropy-based pilot experiments and observe a strong positive correlation between entropy reduction and high-quality tool calls. Building on this finding, we propose using entropy reduction as a supervisory signal and design two reward strategies to address the differing needs of optimizing tool-use behavior. Sparse outcome rewards provide coarse, trajectory-level guidance to improve efficiency, while dense process rewards offer fine-grained supervision to enhance performance. Experiments across diverse domains show that both reward designs improve tool-use behavior: the former reduces tool calls by 72.07% compared to the average of baselines, while the latter improves performance by 22.27%. These results position entropy reduction as a key mechanism for enhancing tool-use behavior, enabling agents to be more adaptive in real-world applications.

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.139ACL Anthology · ACL 2026

CriticLean: Critic-Guided Reinforcement Learning for Mathematical Formalization

Zhongyuan Peng, Yifan Yao, Kaijing Ma, Shuyue Guo, Yizhe Li, Yichi Zhang, Chenchen Zhang, Yifan Zhang Princeton University 等 19 位作者

原文摘要

Translating natural language mathematical statements into formal, executable code is a fundamental challenge in automated theorem proving. While prior work has focused on generation and compilation success, little attention has been paid to the critic phase—the evaluation of whether generated formalizations truly capture the semantic intent of the original problem. In this paper, we introduce CriticLean, a novel critic-guided reinforcement learning framework that elevates the role of the critic from a passive validator to an active learning component. Specifically, first, we propose the CriticLeanGPT, trained via supervised fine-tuning and reinforcement learning, to rigorously assess the semantic fidelity of Lean 4 formalizations. Then, we introduce CriticLeanBench, a benchmark designed to measure models’ ability to distinguish semantically correct from incorrect formalizations, and demonstrate that our trained CriticLeanGPT models can significantly outperform strong open- and closed-source baselines. Building on the CriticLean framework, we construct FineLeanCorpus, a dataset comprising over 509K problems that exhibits rich domain diversity, broad difficulty coverage, and high correctness based on human evaluation.Overall, our findings highlight that optimizing the critic phase is essential for producing reliable formalizations and we hope our CriticLean will provide valuable insights for future advances in formal mathematical reasoning.

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.1404ACL Anthology · ACL 2026

Evidence-Augmented Policy Optimization with Reward Co-Evolution for Long-Context Reasoning

Xin Guan, Zijian Li The Hong Kong University of Science and Technology, Shen Huang Alibaba Group, Pengjun Xie Alibaba Group, Jingren Zhou Alibaba Group, Jiuxin Cao Southeast University

原文摘要

While Reinforcement Learning (RL) has advanced LLM reasoning, applying it to long-context scenarios is hindered by sparsity of outcome rewards. This limitation fails to penalize ungrounded “lucky guesses,” leaving the critical process of needle-in-a-haystack evidence retrieval largely unsupervised. To address this, we propose EAPO (Evidence-Augmented Policy Optimization). We first establish the Evidence-Augmented Reasoning paradigm, validating via Tree-Structured Evidence Sampling that precise evidence extraction is the decisive bottleneck for long-context reasoning. Guided by this insight, EAPO introduces a specialized RL algorithm where a reward model computes a Group-Relative Evidence Reward, providing dense process supervision to explicitly improve evidence quality. To sustain accurate supervision throughout training, we further incorporate an Adaptive Reward-Policy Co-Evolution mechanism. This mechanism iteratively refines the reward model using outcome-consistent rollouts, sharpening its discriminative capability to ensure precise process guidance. Comprehensive evaluations across eight benchmarks demonstrate that EAPO significantly enhances long-context reasoning performance compared to SOTA 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.302ACL Anthology · ACL 2026

Don’t Be Misled by Style: A Style-Adaptive Reranker for Capturing Effective Knowledge in Retrieval-Augmented Generation

Ruwen Zhang, Bo Liu, Zhang Sheng Xiang, Yida Chen Southeast University, Hantao Zhao Southeast University, Ding Ding Southeast University, Jiahui Jin Southeast University, Jiuxin Cao Southeast University

原文摘要

Rerankers are critical in Retrieval-Augmented Generation (RAG) for filtering evidence that enhances the accurate generation of LLMs. With the extension to open-domain scenarios, rerankers are inevitably deployed on mixed-style corpora, whereas most existing rerankers are mainly trained on well-edited texts. A rarely explored issue lies in enabling rerankers to maximally capture the effective knowledge for downstream LLMs without being misled by stylistic features. To address this issue, we propose SARK (Style-Adaptive Reranker with Knowledge Prioritization), a style-augmented multi-task framework that prioritizes effective knowledge over stylistic perturbations. SARK performs multi-granular knowledge mining by using an LLM to derive passage-level supervision on whether a passage helps or harms answer correctness, and list-level relative ranking preferences over candidate passages. It then jointly optimizes the reranker model with passage-level classification and list-level ranking objectives via style-augmented multi-task learning, encouraging the model to focus on the information needed for answering under mixed-style scenarios. Extensive experiments demonstrate that SARK improves generation performance across multiple LLMs under mixed-style conditions.

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.563ACL Anthology · ACL 2026

PRISM: Probabilistic Reward Model with Inherent Structural Modeling

Yuhang Zhou, Yixin Cao Fudan University, Yuchen Ni, Shihan Dou, Xutian Chen, Ge Zhang ByteDance Inc., Xiang Liu, Guangnan Ye Fudan University

原文摘要

Standard evaluators, such as reward models, compress diverse human judgments into a single scalar, conflating valid Subjective Preference with Cognitive Uncertainty. This structural mismatch often leads to brittle alignment and reward hacking. To address this, we propose PRISM which reinterprets reward evaluation as a conditional distribution parameterized by a Mixture of Gaussians. PRISM structurally disentangles these factors: distinct Gaussian experts emerge to capture conflicting preference dimensions, while their variance estimates quantify uncertainty, acting as a dynamic reliability gate during optimization. We introduce a two-stage training strategy to learn these disentangled representations from scalable pairwise comparisons without requiring massive fine-grained annotations. Empirical results show that PRISM significantly outperforms scalar baselines in both accuracy and generalization. Furthermore, in downstream Reinforcement Learning, PRISM effectively mitigates reward hacking, yielding policies that are more robust and resilient to distribution shifts.

cs.CL

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