cs.CL（2025-07-10）

📊 共 25 篇论文 | 🔗 5 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (15 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (10 🔗3)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (15 篇)

#题目一句话要点标签🔗⭐
1 Single-to-mix Modality Alignment with Multimodal Large Language Model for Document Image Machine Translation M4Doc：利用多模态大语言模型进行单模态到混合模态对齐的文档图像机器翻译 large language model multimodal
2 When Large Language Models Meet Law: Dual-Lens Taxonomy, Technical Advances, and Ethical Governance 构建法律领域LLM双重视角分类体系，提升法律任务泛化与伦理治理能力 large language model multimodal ✅
3 Automating Expert-Level Medical Reasoning Evaluation of Large Language Models 提出MedThink-Bench基准与LLM-w-Ref评估框架，用于自动化评估LLM的医学推理能力。 large language model
4 Automating MD simulations for Proteins using Large language Models: NAMD-Agent 提出NAMD-Agent，利用大语言模型自动化蛋白质分子动力学模拟流程。 large language model
5 Compactor: Calibrated Query-Agnostic KV Cache Compression with Approximate Leverage Scores Compactor：基于近似杠杆分数的免训练KV缓存压缩，提升长上下文LLM性能。 large language model
6 MIRIX: Multi-Agent Memory System for LLM-Based Agents MIRIX：面向LLM Agent的多Agent记忆系统，解决长期记忆和多模态理解难题 multimodal
7 TruthTorchLM: A Comprehensive Library for Predicting Truthfulness in LLM Outputs TruthTorchLM：一个全面的LLM输出真实性预测开源库 large language model ✅
8 GRASP: Generic Reasoning And SPARQL Generation across Knowledge Graphs 提出GRASP，利用大语言模型零样本生成SPARQL查询，提升知识图谱问答性能。 large language model
9 Understanding and Controlling Repetition Neurons and Induction Heads in In-Context Learning 探讨重复神经元与诱导头在上下文学习中的作用 large language model
10 StreamUni: Achieving Streaming Speech Translation with a Unified Large Speech-Language Model StreamUni：利用统一大型语音语言模型实现流式语音翻译 chain-of-thought
11 Lost in Pronunciation: Detecting Chinese Offensive Language Disguised by Phonetic Cloaking Replacement 提出音韵隐蔽替换检测方法以解决中文攻击性语言识别问题 chain-of-thought
12 CEA-LIST at CheckThat! 2025: Evaluating LLMs as Detectors of Bias and Opinion in Text 利用LLM和少量样本提示，CEA-LIST在CheckThat! 2025中实现多语言主观性检测 large language model
13 Toward Real-World Chinese Psychological Support Dialogues: CPsDD Dataset and a Co-Evolving Multi-Agent System 提出CPsDD数据集与协同演化多智能体系统，用于真实中文心理支持对话 large language model
14 SAND: Boosting LLM Agents with Self-Taught Action Deliberation 提出SAND框架，通过自学习行动审议提升LLM Agent性能 large language model
15 Krul: Efficient State Restoration for Multi-turn Conversations with Dynamic Cross-layer KV Sharing Krul：一种高效的状态恢复系统，通过动态跨层KV共享优化多轮对话。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (10 篇)

#题目一句话要点标签🔗⭐
16 Machine Bullshit: Characterizing the Emergent Disregard for Truth in Large Language Models 提出“机器胡扯”框架与指标，揭示大语言模型中涌现的对真理的漠视现象 reinforcement learning RLHF large language model
17 Towards Interpretable Time Series Foundation Models 提出一种基于指令调优的小型语言模型，用于时间序列的可解释性分析。 distillation foundation model multimodal
18 Distilling Empathy from Large Language Models 提出一种基于提示工程的两阶段微调方法，用于将大型语言模型的共情能力蒸馏到小型语言模型中。 distillation large language model
19 Exploring the Limits of Model Compression in LLMs: A Knowledge Distillation Study on QA Tasks 通过知识蒸馏压缩LLM：在QA任务中探索模型压缩的极限。 distillation large language model
20 The Synergy Dilemma of Long-CoT SFT and RL: Investigating Post-Training Techniques for Reasoning VLMs 揭示长CoT SFT与RL在视觉语言模型推理中的协同困境，探究后训练技术瓶颈 reinforcement learning multimodal chain-of-thought
21 RLEP: Reinforcement Learning with Experience Replay for LLM Reasoning RLEP：通过经验回放增强LLM推理的强化学习方法 reinforcement learning large language model ✅
22 SAGE: A Visual Language Model for Anomaly Detection via Fact Enhancement and Entropy-aware Alignment 提出SAGE，通过事实增强和熵感知对齐解决VLM在工业异常检测中的难题。 DPO direct preference optimization multimodal ✅
23 Not All Preferences are What You Need for Post-Training: Selective Alignment Strategy for Preference Optimization 提出选择性对齐策略Selective-DPO，提升LLM偏好优化效率与准确性 DPO distillation large language model ✅
24 Teaching LLM to Reason: Reinforcement Learning from Algorithmic Problems without Code TeaR：通过算法问题强化学习，提升LLM的推理能力，无需编写代码。 reinforcement learning distillation
25 PLAN-TUNING: Post-Training Language Models to Learn Step-by-Step Planning for Complex Problem Solving 提出PLAN-TUNING，通过模仿规划过程提升小模型在复杂问题求解上的能力。 reinforcement learning large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页