cs.CL（2025-04-04）

📊 共 33 篇论文 | 🔗 3 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (23 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (9 🔗1) 支柱六：视频提取与匹配 (Video Extraction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (23 篇)

#题目一句话要点标签🔗⭐
1 Explain with Visual Keypoints Like a Real Mentor! A Benchmark for Multimodal Solution Explanation 提出ME2基准，评估LLM在数学解题中基于视觉关键点的多模态解释能力 large language model multimodal visual grounding
2 Why Reasoning Matters? A Survey of Advancements in Multimodal Reasoning (v1) 综述多模态推理进展：应对视觉-文本融合挑战，探索后训练优化与推理方法。 large language model multimodal
3 Stance-Driven Multimodal Controlled Statement Generation: New Dataset and Task 提出StanceGen2024数据集与SDMG框架，用于多模态立场驱动的可控推文生成。 large language model multimodal
4 What Large Language Models Do Not Talk About: An Empirical Study of Moderation and Censorship Practices 研究表明大型语言模型存在审查和内容审核行为，且具有地域和意识形态倾向。 large language model
5 Think When You Need: Self-Adaptive Chain-of-Thought Learning 提出自适应思维链学习，解决语言模型在简单问题上过度推理的低效问题。 chain-of-thought
6 Metamorphic Testing for Fairness Evaluation in Large Language Models: Identifying Intersectional Bias in LLaMA and GPT 提出基于变形测试的公平性评估方法，用于识别LLaMA和GPT中的交叉偏差 large language model
7 NAACL2025 Tutorial: Adaptation of Large Language Models NAACL2025教程：面向领域自适应和动态更新的大语言模型调优技术综述 large language model
8 Can AI Master Construction Management (CM)? Benchmarking State-of-the-Art Large Language Models on CM Certification Exams 构建CMExamSet基准数据集，评估大型语言模型在建筑管理认证考试中的表现 large language model
9 Noise Augmented Fine Tuning for Mitigating Hallucinations in Large Language Models 提出NoiseFiT，通过噪声增强微调缓解大语言模型中的幻觉问题 large language model
10 Entropy-Based Block Pruning for Efficient Large Language Models 提出基于熵的Transformer模型块剪枝方法，提升大语言模型效率。 large language model
11 How Social is It? A Benchmark for LLMs' Capabilities in Multi-user Multi-turn Social Agent Tasks 提出HSII基准以评估LLMs在多用户社交任务中的能力 large language model chain-of-thought
12 Do LLM Evaluators Prefer Themselves for a Reason? 研究揭示LLM评估器自偏好现象，并探究其与模型质量及错误识别的关系 large language model chain-of-thought
13 Agentic Knowledgeable Self-awareness 提出KnowSelf，赋予LLM智能体情境自感知能力，提升规划效果。 large language model ✅
14 Language Models Are Implicitly Continuous 揭示Transformer语言模型将句子隐式地表示为连续时间函数 large language model
15 Evaluating Compact LLMs for Zero-Shot Iberian Language Tasks on End-User Devices 评估紧凑型LLM在终端设备上零样本伊比利亚语任务的表现 large language model
16 EnrichIndex: Using LLMs to Enrich Retrieval Indices Offline EnrichIndex：利用LLM离线增强检索索引，提升复杂语义检索性能。 large language model
17 Structured Extraction of Process Structure Properties Relationships in Materials Science 提出一种新标注模式，用于从材料科学文献中结构化提取工艺-结构-性能关系。 large language model
18 Neutralizing the Narrative: AI-Powered Debiasing of Online News Articles 提出基于大型语言模型的AI驱动框架，用于在线新闻文章的偏差检测与消除。 large language model
19 Locations of Characters in Narratives: Andersen and Persuasion Datasets 构建Andersen和Persuasion数据集，用于评估LLM在叙事文本中理解人物与地点关系的能力 large language model
20 BabyLM's First Words: Word Segmentation as a Phonological Probing Task 利用词语分割作为语音探测任务，研究BabyLM中的语音表征 large language model
21 Inherent and emergent liability issues in LLM-based agentic systems: a principal-agent perspective 基于委托代理视角分析LLM智能体系统中固有的和新出现的责任问题 large language model
22 Efficient Dynamic Clustering-Based Document Compression for Retrieval-Augmented-Generation 提出EDC2-RAG框架，通过动态聚类压缩文档，提升RAG在知识问答和幻觉检测任务中的性能。 large language model ✅
23 Beyond the Next Token: Towards Prompt-Robust Zero-Shot Classification via Efficient Multi-Token Prediction 提出P3方法，通过高效多Token预测提升零样本分类的Prompt鲁棒性 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (9 篇)

#题目一句话要点标签🔗⭐
24 Align to Structure: Aligning Large Language Models with Structural Information 提出结构对齐方法，提升大型语言模型在长文本生成中的连贯性和结构性。 reinforcement learning RLHF large language model ✅
25 Algorithmic Prompt Generation for Diverse Human-like Teaming and Communication with Large Language Models 提出基于质量多样性优化LLM提示的算法，用于生成多样化类人团队协作行为 reinforcement learning large language model
26 Online Difficulty Filtering for Reasoning Oriented Reinforcement Learning 提出平衡在线难度过滤方法，提升面向推理的强化学习训练效率与性能 reinforcement learning curriculum learning large language model
27 Learning Natural Language Constraints for Safe Reinforcement Learning of Language Agents 提出基于自然语言约束的安全强化学习框架，提升语言Agent在真实场景中的安全性。 reinforcement learning RLHF large language model
28 Nemotron-H: A Family of Accurate and Efficient Hybrid Mamba-Transformer Models 提出Nemotron-H混合Mamba-Transformer模型，旨在提升推理效率并保持精度。 Mamba distillation
29 Enhancing Personalized Multi-Turn Dialogue with Curiosity Reward 提出基于好奇心奖励的个性化多轮对话方法，提升LLM用户建模能力。 reinforcement learning RLHF large language model
30 Distillation and Refinement of Reasoning in Small Language Models for Document Re-ranking 提出结合知识蒸馏与强化学习的小模型训练方法，用于推理型文档重排序。 reinforcement learning distillation
31 AIR: A Systematic Analysis of Annotations, Instructions, and Response Pairs in Preference Dataset AIR框架：通过解耦偏好数据集的注释、指令和响应对，实现高效对齐。 preference learning large language model
32 Sample, Don't Search: Rethinking Test-Time Alignment for Language Models 提出QAlign，通过采样而非搜索优化语言模型在测试时的对齐问题。 DPO direct preference optimization

🔬 支柱六：视频提取与匹配 (Video Extraction) (1 篇)

#题目一句话要点标签🔗⭐
33 CliME: Evaluating Multimodal Climate Discourse on Social Media and the Climate Alignment Quotient (CAQ) 提出CliME多模态气候数据集与CAQ评估指标，用于评估LLM在气候讨论中的表现。 HuMoR large language model multimodal

⬅️ 返回 cs.CL 首页 · 🏠 返回主页