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

+197新增论文
+151其中 AI 相关
+16已有论文修订
+95新增 AI 导读

arXiv 最近完成扫描:2026-11-21 12:19 · 全站最近入库:2026-11-21 12:19各信源进度 →

统计口径与数据新鲜度

统计 2026-11-20 21:37 — 2026-11-21 21:37(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:2026-11-21 12:19最近发现新增版本:2026-11-21 12:19库内论文最新更新时间:2026-11-20 02:59

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

29,158 已收录论文24,181 AI 相关8,992 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 18 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-21 12:17(北京时间)
版本更新 · v3arXiv:2611.06120

ReactWorld: Structuring Interaction Responses for Embodied Video Generation

Hanwei Liu, Huijia Zhang, Hanzhang Wang, Guangyu Liu, Hanbei Zhan, Linjie Hu, Bin Xiao, Xuan Zhou 等 9 位作者

原文摘要

Embodied video world models predict scene changes caused by robot actions, yet generated object responses may remain inconsistent with task intent and contact dynamics. Critical responses in embodied scenes often concentrate in local regions and specific time intervals. Inspired by chain-of-thought reasoning, we introduce ReactWorld, an embodied video generation model guided by structured interaction anticipation. In the generation backbone's feature space, an interaction response field predictor anticipates how the interaction will unfold, explicitly describing participant roles, future motion response intensity, and local motion changes. A response feedback adapter forms increments to object features from the difference between the object branch's outputs with predicted and zeroed local dynamics inputs. It also converts predicted displacements into positional biases for temporal attention. These two mechanisms guide object feature updates and information retrieval across frames, respectively. Through interaction anticipation and generative feedback, ReactWorld achieves an average score of 77.51 on RBench, outperforming PhysisForcing by 3.29 points with improvements in all nine categories. On EZS-Bench, it achieves a Domain score of 86.98 and an average score of 81.87.

cs.CV · cs.RO

版本更新 · v3arXiv:2609.27964

Learning from Transients: Linear RNN Scaling Laws in Stationary Prediction

Ziyan Chen, Linhe Zheng, Peilin Liu, Zhongzhu Zhou, Ding-Xuan Zhou

原文摘要

Does the organization of a fixed number of transitions into independent trajectories affect a fixed stationary prediction task? We study a Gaussian-sketched linear recurrent teacher--student model, where independently initialized training trajectories create transients that can excite weak spectral directions. For full-batch gradient descent with stable deterministic learning rates, we establish matching stationary approximation, bias, variance, and total-risk scaling laws from dependent trajectory data. When initialization decays at least as fast spectrally as the innovations, trajectory count and length enter the leading variance rate through their product. With a heavier initialization tail, transient excitation can improve the trajectory-length-dependent finite-time bias and, at fixed transition count, variance, while the stationary oracle shift remains controlled by representation approximation. Under token and compute caps, constant-length trajectories attain the optimal risk scaling order within the theorem-controlled range. Synthetic experiments provide finite-scale evidence in this stylized model.

cs.LG

新论文 · v1arXiv:2611.24236

From a finite subgroup of $SO(3)$:$C_2$ leads to a Holant dichotomy

Yuxuan Zhou

原文摘要

Counting complexity within the Holant framework has been extensively studied through complexity dichotomy theorems. While substantial progress has been achieved for various restricted settings, the general complex-valued even-arity case without auxiliary unary signatures remains largely open. This paper studies a fundamental case in this program where the binary signatures induce a group structure satisfying \(M(B(\mathcal{F}))\cong C_2\). It is observed that projective equivalence classes of binary signature matrices naturally embed into finite subgroups of \(SO(3)\), providing an algebraic viewpoint for structural classification. This paper further identifies two fundamentally different regimes for the \(C_2\) case, namely the normal and isotropic cases, and analyzes their complexity behavior separately. Combining gadget constructions, decomposition techniques, interpolation arguments, and a realnumberization approach that transforms the complex-valued setting into an equivalent real-structured framework, this paper establishes a complete dichotomy theorem: for every even-arity signature set \(\mathcal{F}\) satisfying \(M(\B(\mathcal{F}))\cong C_2\), the problem \(\mathrm{Holant}(\mathcal{F})\) is either polynomial-time computable or \#P-hard. A Pauli-basis representation for signatures is further introduced as an algebraically natural language for organizing realizable signatures and revealing structural relations that are less transparent in the standard basis.

cs.CC · cs.CL · cs.DM

版本更新 · v2arXiv:2609.37118

Three-Color Free-Flood-It on Fixed-Height Grids Is Polynomial-Time Solvable

Yuxuan Zhou

原文摘要

We give a deterministic algorithm for \textsc{Free-Flood-It} on rectangular grids $P_k\square P_n$ with at most three colors. For every fixed height $k$, it computes the minimum number of moves and an optimal sequence in $N^{O(k^2)}$ time, where $N=kn$. This resolves the previously open three-color case on complete \(3\times n\) boards. The proof uses a representation of flooding strategies as paintings by connected regions. We show that an optimal painting can be chosen so that its regions form a rooted tree with strong restrictions on the colors along ancestral paths. These restrictions make the part of the tree visible in any fixed-size connected window admit only polynomially many descriptions. The remaining common ancestors may form an arbitrarily long chain. We retain that chain on a stack and use a finite context-free recurrence to minimize the total painting cost without enumerating all possible stack contents. Connectivity information at the boundary ensures that the resulting local descriptions assemble into one valid global painting. The argument also applies to graphs supplied with an ordering into connected layers of bounded size, provided every edge lies within one layer or joins consecutive layers, and consecutive layers are joined by an edge. For a family of three-row boards, every optimal admissible painting tree has a common-ancestor chain of depth linear in the board width. For any fixed height and palette size, we give a deterministic EPTAS via a separate budget-parameterized algorithm, and we show that no FPTAS exists for $k\geq 3$ with at least four colors unless P = NP. We also provide a solver for \(3\times n\) boards, implemented with AI assistance based on the algorithm described in this paper. The implementation is available at GitHub

cs.CC · cs.DS

新论文 · v1arXiv:2611.22971

SEAM: Computer-Use Agents Know the Harm Between Screens, and Commit It Anyway

Yuxuan Zhou, Chen Xiong, Haitong Liu, Kuofeng Gao, Yang Bai, Zhiyuan He, Tao Dai, Shu-Tao Xia

原文摘要

Safety benchmarks for computer-use agents (CUAs) place the harm either in the intent of the task or on a single screen: a malicious instruction, an injected command, a phishing page. An agent that checks the task and each screen in isolation can pass them, so they do not test whether it combines facts seen on different screens. Yet in delegated work, harm often arises exactly this way: a fact about a person on one screen, a fact about an item on another, and a legitimate task that binds them. We introduce SEAM (Safety Execution Across Multiple screens), a benchmark of such cross-screen compositional harm: 600 multi-step, cross-application OSWorld tasks in which the instruction and every single observation are benign and only their combination is harmful. Minimal-pair variants that differ in one place, scored from the server log rather than by an LLM judge, locate where an agent fails. General-purpose open-weight and flagship agents carry out the harm in up to 92% of runs, yet decline when the user states it in one sentence. Agents open the decisive pages, so exposure is not the primary bottleneck. Asked about those pages, they flag the harm, and with thinking they often write the conflict into their reasoning and confirm anyway. Safety recognition thus does not reliably translate into safe execution, and the translation depends on the information channel: the same warning stops agents far more often in the user's instruction than on screen. Proprietary models are markedly more cautious. We release SEAM with its generator and quality gates at https://yuxuan2003.github.io/SEAM/ .

cs.CR · cs.AI

新论文 · v1arXiv:2611.22883

Fly as Instructed: Refining World-Action Policies for UAV Vision-Language Navigation

Zhexuan Zhou, Yanke Sun, Yuhan Zong, Youmin Gong, Jie Mei

原文摘要

Natural language-guided UAV navigation enables intuitive human-robot interaction and flexible autonomous flight in complex environments. However, reliable navigation requires accurate instruction following, continuous task progress tracking, and safe obstacle avoidance, which remain challenging for existing world-action models that primarily focus on action prediction. To address these limitations, we propose a world-action learning framework that integrates three complementary mechanisms: a language-conditioned action interface with targeted post-training for instruction-responsive motion, recurrent instruction-consumption prediction to track task progress and determine termination, and Gaussian-scene post-training to enhance obstacle avoidance through trajectory imitation and differentiable geometric feedback. Closed-loop evaluations and ablation studies demonstrate improvements in sequential task completion, safe goal reaching, and timely termination. Furthermore, direct one-step denoising reduces inference overhead, and real-world indoor and outdoor flights validate the framework's effectiveness.

cs.RO

新论文 · v1arXiv:2611.22650

Must You See Fakes to Catch Fakes? Language Says What to Look For, Real Images What to Ignore

Tianchi Xu, Yuxuan Zhou, Yuheng Zhang, Yanan Guan, Tao Dai, Shu-Tao Xia

原文摘要

Must a detector see generated images in order to catch them? Detection of AI-generated images has long faced two problems. The cues a detector learns from the few generators in its training set carry traces specific to those generators and transfer poorly to new ones, which limits generalization. The real and generated training images also differ in format and content, so the detector takes these differences as evidence of authenticity, and its decisions change with factors that have nothing to do with authenticity, which harms robustness as well as generalization. Both problems stem from training on generated images. A detector never trained on them can largely avoid both at the source. Vision foundation models have already learned the concept of "AI-generated" during pretraining, and text can point to it directly. Building on this, we propose SaySee, a training-free detector: Say builds a generation direction from sets of paired descriptions that differ only in authenticity and scores an image by the projection of its embedding onto it, and See uses real images to correct the score offsets caused by format, image type and content. On 24 evaluation sets that span standard benchmarks, in-the-wild data and the latest generators, SaySee reaches a mean AUC of 98.2% with MoE-ViE, 4.2 points above the best existing detector. When format is decoupled from authenticity, it remains the best of all methods. Code will be released.

cs.CV

新论文 · v1arXiv:2611.14929

Speech Tokenizers are Vulnerable: Transferable Semantic Attack and Robust Tokenizer

Zhisheng Zhang, Yifan Mi, Yixuan Zhou, Haiyun Li, Shengbo Cai, Dongfu Song, Ziang Xu, Jie Hao 等 9 位作者

原文摘要

Speech tokenizers serve as speech front ends, encoding input speech signals into representations for downstream models, such as automatic speech recognition (ASR) systems and large audio-language models (LALMs). In this paper, we find that speech tokenizers are fragile at the semantic level: small perturbations can disrupt their semantic encoding process and consequently cause severe degradation in downstream ASR and LALM performance. To study this vulnerability, we propose T-SemAttack, a transferable semantic attack based on a semantic encoder ensemble. By jointly drifting the representation spaces of multiple semantic encoders, T-SemAttack effectively disrupts the semantic content of speech while preserving perceptual quality, thereby inducing error features. We further analyze layer-wise representation drift and cross-model gradient alignment. Our findings show that perturbations can induce substantial shifts in deep semantic representations and that the input gradients of different surrogate encoders exhibit directional alignment. These observations suggest that shared adversarially sensitive directions in the input space may be a key factor underlying cross-model attack transferability. Building on these, we introduce ROSETok, a robust speech tokenizer that combines noisy training with robust semantic distillation to improve reconstruction fidelity and downstream SpeechLLM robustness. Extensive experiments on multiple datasets, 9 open-source and black-box ASR systems, and 4 LALMs demonstrate that T-SemAttack achieves strong transferable attack performance, while the proposed Robust Speech Tokenizer exhibits robustness under high-fidelity reconstruction and downstream tasks. Our code and demo are available at https://wxzyd123.github.io/t-semattack.

eess.AS · cs.AI · cs.CR · cs.SD

版本更新 · v2arXiv:2502.13297

Understanding and Tackling Label Errors in Individual-Level Nature Language Understanding

Yunpeng Xiao, Kyrie Zhixuan Zhou, Youpeng Zhao, Kai Shu

原文摘要

Natural language understanding (NLU) is a task that enables machines to understand human language. Some tasks, such as stance detection and sentiment analysis, are closely related to individual subjective perspectives, thus termed individual-level NLU. Previously, these tasks are often simplified to text-level NLU tasks, ignoring individual factors. This not only makes inference difficult and unexplainable but often results in a large number of label errors when creating datasets. To address the above limitations, we propose a new NLU annotation guideline based on individual-level factors. Specifically, we incorporate other posts by the same individual and then annotate individual subjective perspectives after considering all individual posts. We use this guideline to expand and re-annotate the stance detection and topic-based sentiment analysis datasets. We find that error rates in the samples were as high as 31.7\% and 23.3\%. We further use large language models to conduct experiments on the re-annotation datasets and find that the large language models perform well on both datasets after adding individual factors. Both GPT-4o and Llama3-70B can achieve an accuracy greater than 87\% on the re-annotation datasets. We also verify the effectiveness of individual factors through ablation studies. We call on future researchers to add individual factors when creating such datasets. Our re-annotation dataset can be found at https://github.com/24yearsoldstudent/Individual-NLU

cs.CL · cs.AI

新论文 · v1arXiv:2611.06067

Offline Learning of Reserve Prices from Censored Bids

Lavanya Marla, Yunzong Xu, Boxuan Zhou

原文摘要

We study how a seller can learn a reserve price for a second-price auction from censored historical bid data. Because values below past reserves are unobserved, even unlimited data may not reveal which reserve earns the most revenue. Under a monotone hazard rate (MHR) assumption, we characterize exactly when an optimal reserve is identified, even though the value distribution may not be: either revenue has not yet begun to fall at the lowest historical reserve, so the best reserve lies among the observed bids, or the censored fraction is as large as MHR allows, which pins down the entire distribution. When identification fails, we derive the reserve that minimizes worst-case revenue regret over distributions consistent with the data, and prove that this unique minimax reserve and its regret depend on just two observable quantities: the censored fraction and the hazard just above the lowest historical reserve. With $N$ historical bidders, the optimal reserve is learned with expected worst-case regret $O(N^{-1/2})$ when it lies among the observed bids, and $O(N^{-1})$ from the censored count alone when the censored fraction is as large as MHR allows and the optimum lies at or below the lowest reserve. In the nonidentifiable regime, plugging estimates of the two quantities into the minimax reserve gives expected worst-case regret within $O(N^{-2/5})$ of the population minimum under local density smoothness, with a matching minimax lower bound on a smooth class. Beyond MHR, we solve the minimax problem for regular and unrestricted distributions and prove an $O(N^{-2/5})$ learning bound in the regular nonidentifiable regime, given known bounds on the boundary parameters. Experiments on synthetic and eBay data show gains from exploiting MHR and the benefits of the robust rule when parametric models are misspecified.

cs.GT · cs.LG · econ.EM · stat.ML

新论文 · v1arXiv:2610.40492

Agentic University:面向个性化学习的协作智能体

Agentic University: Collaborative Agents for Personalized Learning

Jingru Fan, Tianle Zhou, Yunjie Zhou, Zihao He, Yixuan Zhou, Kezhao Wu, Xinyu Zheng, Xiyuan Wang 等 13 位作者

AI 速读 基于摘要

提出一种以学习者参与为核心的多智能体组织框架,并在 Agentic University 中实现,用可修订的“能力项目”把个人目标、学习活动与分层评估串联起来。

cs.AI · cs.CL

查看原文摘要

Large language models support personalized tutoring, content generation, and adaptive learning paths. However, capability development requires these resources and interactions to form a coherent learning process that evolves with the learner. A central challenge is to align goals, activities, and assessment while using learning evidence to decide what should follow. We introduce Human-Participatory Multi-Agent Organization and instantiate it in Agentic University (AU). In AU, learners participate as organization members and work with specialized agents to maintain a revisable Capability Program. The program connects personal goals, learning activities across modalities, and layered assessment, while learner feedback and performance guide subsequent support. Our evaluation shows advantages in program organization, assessment design, and revision quality. Users achieved larger mean assessment gains (12.19 versus 5.61 points on a 50-point scale) and reported stronger cognitive support. This work contributes an organizational approach to adaptive learning that makes learner participation a continuing basis for multi-agent coordination. The Agentic University project is available at \href{https://github.com/AgenticOrgLab/AgenticUniversity.git}.

新论文 · v1arXiv:2610.38262

Rosemary: Efficient and Adaptable In-Storage Acceleration for Fully Homomorphic Encryption

Xuan Wang, Tianqi Zhang, Gabrielle De Micheli, Yujin Nam, Keming Fan, Augusto Vega, Minxuan Zhou, Tajana Rosing

原文摘要

Fully Homomorphic Encryption (FHE) enables computation on encrypted data and is promising for privacy-preserving computing. However, FHE deployment remains constrained due to massive computational, memory, and data-movement overheads. Existing acceleration strategies often overlook the storage I/O bottleneck caused by the enormous size of FHE data in large-scale applications, as most prior works focus on small benchmarks that fit in main memory. Our studies show that state-of-the-art FHE systems spend 71\% to 85\% of their execution time moving data between storage and compute units rather than performing useful computation. To address the I/O bottleneck, we present Rosemary, a heterogeneous FHE acceleration system that incorporates a novel in-storage processing (ISP) architecture with compiler-level and algorithmic optimizations. Our key innovations include 1) an adaptable hardware-aware FHE engine inside a NAND flash-based solid-state drive (SSD) for efficient and cost-effective in-storage FHE processing, 2) a Throughput-Data-Aware Program Allocator (TDAP) that partitions FHE workloads into heterogeneous programs, and 3) a Compiler-Level Ciphertext-Aware Optimizer (CLCAO) that exploits ISP-friendly FHE kernels for low-cost ISP execution. Across a diverse set of FHE workloads, including encrypted deep learning inference and cloud database queries, Rosemary achieves geometric-mean speedups of $185.7\times$ over state-of-the-art CPU/GPU implementations and $3.7\times$ over prior FHE accelerators, while improving energy efficiency by up to $3.9\times$ and Energy-Delay-Area-Product (EDAP) by up to $16.5\times$.

cs.CR · cs.AR

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-demo.56ACL Anthology · ACL 2026

UltraEval-Audio: A Unified Framework for Comprehensive Evaluation of Audio Foundation Models

Qundong Shi ModelBest, Jie Zhou, Biyuan Lin, Junbo Cui ModelBest, Guoyang Zeng ModelBest, Yixuan Zhou Tsinghua University, Ziyang Wang, Xin Liu 等 11 位作者

原文摘要

The development of audio foundation models has accelerated rapidly since the emergence of GPT-4o. However, the lack of comprehensive evaluation has become a critical bottleneck for further progress in the field, particularly in audio generation. Current audio evaluation faces three major challenges: (1) audio evaluation lacks a unified framework, with datasets and code scattered across various sources;(2) audio codec, as a key component of audio foundation models, lacks a widely accepted and holistic evaluation methodology; (3) existing speech benchmarks are heavily reliant on English, making it challenging to objectively assess models’ performance on Chinese.We introduce UltraEval-Audio, a unified framework addressing these challenges through a modular architecture supporting 10 languages, 14 task categories, 24 models, and 36 benchmarks with one-command evaluation and real-time leaderboards. For audio codec, we propose a three-dimensional evaluation scheme covering semantic accuracy, timbre fidelity, and acoustic quality. For Chinese evaluation, we introduce two new benchmarks: SpeechCMMLU and SpeechHSK. Our code, benchmarks, and leaderboards are available at https://github.com/OpenBMB/UltraEval-Audio.

cs.CL

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-long.1531ACL Anthology · ACL 2026

RedCoder: Automated Multi-Turn Red Teaming for Code LLMs

Wenjie Jacky Mo, Qin Liu University of California, Davis, Xiaofei Wen University of California, Davis, Dongwon Jung, Hadi Askari, Wenxuan Zhou Google Deepmind, Zhe Zhao University of California, Davis, Muhao Chen University of California, Davis

原文摘要

Large Language Models (LLMs) for code generation (i.e., Code LLMs) have demonstrated impressive capabilities in AI-assisted software development and testing. However, recent studies have shown that these models are prone to generating vulnerable or even malicious code under adversarial settings. Existing red-teaming approaches rely on extensive human effort, limiting their scalability and practicality, and generally overlook the interactive nature of real-world AI-assisted programming, which often unfolds over multiple turns. To bridge these gaps, we present RedCoder, a red-teaming agent that engages victim models in multi-turn conversation to elicit vulnerable code. The pipeline to construct RedCoder begins with a multi-agent gaming process that simulates adversarial interactions, yielding a set of prototype conversations and an arsenal of reusable attack strategies. We then fine-tune an LLM on these prototype conversations to serve as the backbone of RedCoder. Once deployed, RedCoder autonomously engages Code LLMs in multi-turn conversations, dynamically retrieving relevant strategies from the arsenal to steer the dialogue toward vulnerability-inducing outputs. Experiments across multiple Code LLMs show that our approach outperforms prior single-turn and multi-turn red-team methods in inducing vulnerabilities in code generation, offering a scalable and effective tool for evaluating the security boundaries of modern code-generation systems.

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.403ACL Anthology · ACL 2026

SCVQ: Sparse-Compensated Vector Quantization for Large Language Models

Zixuan Zhou Beijing University of Posts and Telecommunications, Yujun Diao Beijing University of Post and Telecommunications, Zicheng Kong Beijing University of Posts and Telecommunications, Dehua Ma, Zhenbo Xu, Pei Pei Li, Zhaofeng He Beijing University of Post and Telecommunication

原文摘要

Large Language Models (LLMs) are primarily constrained by memory and bandwidth bottlenecks during deployment. Although Vector Quantization (VQ) has emerged as a promising solution, existing methods incur inference overhead due to massive codebook storage and intensive index lookups. Moreover, these methods typically suffer from non-negligible performance degradation under ultra-low bitwidth regimes. To bridge this gap, we propose Sparse-Compensated Vector Quantization (SCVQ), a novel framework designed for high-efficiency LLM vector quantization. SCVQ introduces a salience-aware weighted K-means clustering scheme with symmetry constraints to reduces codebook size and indexing costs. Central to our approach is a unified structured representation that consolidates outliers, salient weights, and quantization residuals into a single sparse compensation matrix. This design effectively preserves critical model information while leveraging VQ-specific properties to enable efficient custom kernels. Extensive experiments across multiple benchmarks demonstrate SCVQ’s superior performance. Specifically, SCVQ achieves a perplexity of 5.78 on WikiText-2 for LLaMA-2-7B at 2-bit quantization, while delivering a 1.4\times end-to-end inference speedup over existing 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.569ACL Anthology · ACL 2026

GASim: A Graph-Accelerated Hybrid Framework for Social Simulation

Xuan Zhou, Yanhui Sun, Hantao Yao, Allen He, Yongdong Zhang University of Science and Technology of China, Wu Liu

原文摘要

Large-scale social simulators are essential for studying complex social patterns. Prior work explores hybrid methods to scale up simulations, combining large language models (LLM)-based agents with numerical agent-based models (ABM). However, this incurs high latency due to expensive memory retrieval and sequential ABM execution. To address this challenge, we propose GASim, a graph-accelerated hybrid multi-agent framework for large-scale social simulations. For core agents driven by LLM, GASim introduces Graph-Optimized Memory (GOM) to replace intensive LLM-based retrieval pipelines with lightweight propagation over a sparse memory graph. For the majority of ordinary agents, GASim employs Graph Message Passing (GMP), substituting sequential ABM execution with parallel updates by fine-grained feature aggregation and Graph Attention Network. We further introduce Entropy-Driven Grouping (EDG) that coordinates this hybrid partitioning, leveraging information entropy to dynamically identify emergent core agents situated in information-diverse neighborhoods. Extensive experiments show that GASim not only delivers a substantial 9.94× end-to-end speedup over the traditional hybrid framework but also consumes less than 20% of baseline tokens, significantly reducing costs while preserving strong alignment with real-world public opinion trends.

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.78ACL Anthology · ACL 2026

SFTMix: Elevating Language Model Instruction Tuning with Mixup Recipe

Yuxin Xiao, Shujian Zhang University of Texas, Austin, Marzyeh Ghassemi Massachusetts Institute of Technology, Wenxuan Zhou Google Deepmind

原文摘要

To acquire instruction-following capabilities, large language models (LLMs) undergo instruction tuning, where they are trained on instruction-response pairs using next-token prediction (NTP). Efforts to improve instruction tuning often focus on higher-quality supervised fine-tuning (SFT) datasets, typically requiring data filtering with proprietary LLMs or human annotation. In this paper, we take a different approach by proposing SFTMix, a novel Mixup-based recipe that elevates LLM instruction tuning without relying on well-curated datasets. We observe that LLMs exhibit uneven confidence across the semantic representation space. We argue that examples with different confidence levels should play distinct roles in instruction tuning: Confident data is prone to overfitting, while unconfident data is harder to generalize. Based on this insight, SFTMix leverages training dynamics to identify examples with varying confidence levels. We then interpolate them to bridge the confidence gap and apply a Mixup-based regularization to support learning on these additional, interpolated examples. We demonstrate the effectiveness of SFTMix in both instruction-following and healthcare-specific SFT tasks, with consistent improvements across LLM families and SFT datasets of varying sizes and qualities. Extensive analyses across six directions highlight SFTMix’s compatibility with data selection, adaptability to compute-constrained scenarios, and scalability to broader 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-short.45ACL Anthology · ACL 2026

Taming Extreme Tokens: Covariance-Aware GRPO with Gaussian-Kernel Advantage Reweighting

Cheng Wang, Qin Liu University of California, Davis, Wenxuan Zhou Google Deepmind, Muhao Chen University of California, Davis

原文摘要

Group Relative Policy Optimization (GRPO) has emerged as a promising approach for improving the reasoning capabilities of large language models. However, it struggles to effectively balance the trade-off between exploration and exploitation during training, often resulting in suboptimal performance. Motivated by the theoretical insight that changes in entropy are governed by the covariance between token probabilities and their corresponding advantages, we propose a hyperparameter-free, covariance-weighted optimization method that dynamically down-weights extreme token-level updates via a Gaussian kernel. This approach automatically reduces the instability caused by the exploration-exploitation trade-off while preserving informative learning signals. Extensive empirical evaluations show that our approach improves downstream performance across reasoning benchmarks compared with GRPO, and effectively stabilizes entropy as training progresses.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) · 2026 · 正式发表