cs.CL（2025-10-07）

📊 共 47 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (31) 支柱二：RL算法与架构 (RL & Architecture) (13 🔗3) 支柱五：交互与反应 (Interaction & Reaction) (1) 支柱八：物理动画 (Physics-based Animation) (1 🔗1) 支柱四：生成式动作 (Generative Motion) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (31 篇)

#题目一句话要点标签🔗⭐
1 EvalMORAAL: Interpretable Chain-of-Thought and LLM-as-Judge Evaluation for Moral Alignment in Large Language Models 提出EvalMORAAL框架以评估大型语言模型的道德一致性 large language model chain-of-thought
2 MMA-ASIA: A Multilingual and Multimodal Alignment Framework for Culturally-Grounded Evaluation 提出MMA-ASIA框架，用于多语言多模态文化背景下的大语言模型评测 large language model multimodal
3 Prototype-Based Dynamic Steering for Large Language Models 提出基于原型动态引导（PDS）方法，无需指令即可增强LLM的推理能力 large language model chain-of-thought
4 DACP: Domain-Adaptive Continual Pre-Training of Large Language Models for Phone Conversation Summarization 提出DACP：领域自适应持续预训练LLM，提升电话对话摘要效果 large language model
5 LatentBreak: Jailbreaking Large Language Models through Latent Space Feedback LatentBreak：通过隐空间反馈绕过大语言模型的安全机制 large language model
6 H1B-KV: Hybrid One-Bit Caches for Memory-Efficient Large Language Model Inference H1B-KV：提出混合一位缓存，解决LLM长文本推理中的内存瓶颈问题。 large language model
7 EverydayMMQA: A Multilingual and Multimodal Framework for Culturally Grounded Spoken Visual QA 提出EverydayMMQA框架与OASIS数据集，解决多语言多模态VQA中文化常识不足问题 multimodal
8 CreditDecoding: Accelerating Parallel Decoding in Diffusion Large Language Models with Trace Credits 提出CreditDecoding以加速扩散大语言模型的并行解码 large language model
9 Distributional Semantics Tracing: A Framework for Explaining Hallucinations in Large Language Models 提出分布语义追踪框架，解释大语言模型中的幻觉现象 large language model
10 Centering Emotion Hotspots: Multimodal Local-Global Fusion and Cross-Modal Alignment for Emotion Recognition in Conversations 提出基于情感热点的多模态局部-全局融合与跨模态对齐的对话情感识别模型 multimodal
11 Code-Switching In-Context Learning for Cross-Lingual Transfer of Large Language Models 提出代码切换上下文学习(CSICL)，提升大语言模型跨语言迁移能力。 large language model
12 GraphGhost: Tracing Structures Behind Large Language Models GraphGhost：通过图结构追踪大语言模型内部推理机制 large language model
13 MADIAVE: Multi-Agent Debate for Implicit Attribute Value Extraction 提出MADIAVE，利用多智能体辩论框架解决电商中隐式属性值抽取难题 large language model multimodal
14 Toward a Safer Web: Multilingual Multi-Agent LLMs for Mitigating Adversarial Misinformation Attacks 提出一种多语言多Agent LLM框架，用于缓解对抗性虚假信息攻击。 large language model
15 Mnemosyne: An Unsupervised, Human-Inspired Long-Term Memory Architecture for Edge-Based LLMs Mnemosyne：面向边缘LLM的、受人类启发的无监督长期记忆架构 large language model
16 Activation-Informed Pareto-Guided Low-Rank Compression for Efficient LLM/VLM 提出PGSVD，通过激活感知的Pareto引导低秩压缩高效压缩LLM/VLM large language model
17 RECODE-H: A Benchmark for Research Code Development with Interactive Human Feedback RECODE-H：一个通过人机交互反馈改进科研代码生成的基准。 large language model
18 Spectrum Tuning: Post-Training for Distributional Coverage and In-Context Steerability 提出Spectrum Tuning，提升语言模型在上下文引导下的分布覆盖能力 instruction following
19 LexiCon: a Benchmark for Planning under Temporal Constraints in Natural Language LexiCon：一个用于评估LLM在自然语言时序约束下规划能力的基准 large language model
20 Instructional Goal-Aligned Question Generation for Student Evaluation in Virtual Lab Settings: How Closely Do LLMs Actually Align? 提出教学目标对齐的问题生成框架，利用LLM辅助虚拟实验室学生评估。 large language model
21 Semantic Regexes: Auto-Interpreting LLM Features with a Structured Language 提出语义正则表达式以解决LLM特征自动解释问题 large language model
22 EVALUESTEER: Measuring Reward Model Steerability Towards Values and Preferences EVALUESTEER：衡量奖励模型在价值观和偏好方面的可操纵性 large language model
23 VecInfer: Efficient LLM Inference with Low-Bit KV Cache via Outlier-Suppressed Vector Quantization VecInfer：通过抑制异常值的向量量化实现低比特KV缓存的高效LLM推理 large language model
24 Evaluating The Impact of Stimulus Quality in Investigations of LLM Language Performance 通过优化刺激质量，提升LLM在句法预测任务中的性能评估 large language model
25 MASA: Rethinking the Representational Bottleneck in LoRA with Multi-A Shared Adaptation MASA：通过多A共享适配重塑LoRA中的表征瓶颈 large language model
26 Exploring Gaps in the APS: Direct Minimal Pair Analysis in LLM Syntactic Assessments 通过直接最小对分析揭示LLM句法评估中的差距，提升评估透明度。 large language model
27 Hire Your Anthropologist! Rethinking Culture Benchmarks Through an Anthropological Lens 通过人类学视角重塑文化基准评测，提升大语言模型文化理解能力 large language model
28 The fragility of "cultural tendencies" in LLMs 通过更广泛的实验验证，揭示大语言模型中“文化倾向”的脆弱性 large language model
29 Luth: Efficient French Specialization for Small Language Models and Cross-Lingual Transfer Luth：面向法语的小型语言模型高效特化与跨语言迁移 large language model
30 Adaptive and Multi-Source Entity Matching for Name Standardization of Astronomical Observation Facilities 提出一种自适应多源实体匹配方法，用于天文观测设施名称标准化。 large language model
31 KEO: Knowledge Extraction on OMIn via Knowledge Graphs and RAG for Safety-Critical Aviation Maintenance KEO：面向航空维护，利用知识图谱与RAG进行OMIn知识抽取 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (13 篇)

#题目一句话要点标签🔗⭐
32 LANTERN: Scalable Distillation of Large Language Models for Job-Person Fit and Explanation LANTERN：面向职位-人才匹配与解释的大语言模型可扩展蒸馏框架 distillation large language model
33 The Valley of Code Reasoning: Scaling Knowledge Distillation of Large Language Models 揭示代码推理的“山谷”现象：扩展大语言模型知识蒸馏 distillation large language model
34 Prompt reinforcing for long-term planning of large language models 提出基于强化学习的Prompt优化框架，提升LLM在长期规划任务中的性能 reinforcement learning large language model
35 Probing the Difficulty Perception Mechanism of Large Language Models 探究大语言模型难度感知机制，实现自动难度标注。 curriculum learning large language model ✅
36 Webscale-RL: Automated Data Pipeline for Scaling RL Data to Pretraining Levels Webscale-RL：构建自动化数据流水线，将强化学习数据扩展到预训练级别 reinforcement learning imitation learning large language model
37 On the Role of Difficult Prompts in Self-Play Preference Optimization 研究表明，在自博弈偏好优化中，困难提示词会降低大语言模型对齐效果，并提出缓解策略。 DPO direct preference optimization large language model
38 Peeking inside the Black-Box: Reinforcement Learning for Explainable and Accurate Relation Extraction CogRE：利用强化学习提升关系抽取准确性和可解释性 reinforcement learning
39 ASPO: Asymmetric Importance Sampling Policy Optimization ASPO：非对称重要性采样策略优化，提升LLM在强化学习中的性能 reinforcement learning large language model ✅
40 EEPO: Exploration-Enhanced Policy Optimization via Sample-Then-Forget 提出EEPO，通过采样后遗忘机制增强LLM在强化学习中的探索能力。 reinforcement learning large language model
41 Diversity Is All You Need for Contrastive Learning: Spectral Bounds on Gradient Magnitudes 提出谱约束的对比学习方法以提升梯度稳定性 contrastive learning
42 DecEx-RAG: Boosting Agentic Retrieval-Augmented Generation with Decision and Execution Optimization via Process Supervision DecEx-RAG：通过过程监督优化决策与执行，提升Agentic RAG性能 reinforcement learning large language model ✅
43 A Goal Without a Plan Is Just a Wish: Efficient and Effective Global Planner Training for Long-Horizon Agent Tasks EAGLET：高效训练全局规划器，提升LLM智能体长程任务表现 reinforcement learning large language model
44 Presenting a Paper is an Art: Self-Improvement Aesthetic Agents for Academic Presentations 提出EvoPresent框架，通过自提升美学智能体增强学术报告的呈现效果 reinforcement learning multimodal

🔬 支柱五：交互与反应 (Interaction & Reaction) (1 篇)

#题目一句话要点标签🔗⭐
45 MathRobust-LV: Evaluation of Large Language Models' Robustness to Linguistic Variations in Mathematical Reasoning MathRobust-LV：评估大语言模型在数学推理中对语言变异的鲁棒性 IMoS large language model

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
46 Sci-Phi: A Large Language Model Spatial Audio Descriptor Sci-Phi：首个能完整描述空间音频场景的大语言模型 PULSE large language model ✅

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
47 Controllable Stylistic Text Generation with Train-Time Attribute-Regularized Diffusion 提出RegDiff：一种训练时属性正则化的扩散模型，用于可控风格文本生成。 classifier-free guidance

⬅️ 返回 cs.CL 首页 · 🏠 返回主页