cs.CL（2025-01-28）

📊 共 24 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) (18 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (5 🔗2) 支柱六：视频提取与匹配 (Video Extraction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (18 篇)

#题目一句话要点标签🔗⭐
1 Contextual Reinforcement in Multimodal Token Compression for Large Language Models 提出基于上下文强化学习的多模态Token压缩方法，提升大语言模型效率。 large language model multimodal
2 "Ownership, Not Just Happy Talk": Co-Designing a Participatory Large Language Model for Journalism 共建新闻业专属大语言模型，解决通用模型适应性问题并保障记者权益。 large language model foundation model
3 MME-Industry: A Cross-Industry Multimodal Evaluation Benchmark MME-Industry：面向工业场景的多模态大语言模型评测基准 large language model multimodal
4 Mitigating Hallucinated Translations in Large Language Models with Hallucination-focused Preference Optimization 提出基于幻觉偏好优化的LLM翻译方法，显著降低翻译幻觉 large language model
5 Large Language Model Critics for Execution-Free Evaluation of Code Changes 提出基于大语言模型的代码变更评估方法，无需执行即可评估代码质量。 large language model ✅
6 DOCS: Quantifying Weight Similarity for Deeper Insights into Large Language Models 提出DOCS指标，量化大语言模型权重相似性，揭示模型架构特性 large language model
7 ASTRAL: Automated Safety Testing of Large Language Models ASTRAL：一种用于大规模语言模型自动化安全测试的框架 large language model
8 3D-MoE: A Mixture-of-Experts Multi-modal LLM for 3D Vision and Pose Diffusion via Rectified Flow 提出3D-MoE，通过混合专家模型和修正流扩散提升3D视觉和姿态扩散性能。 large language model instruction following
9 A Stochastic Dynamical Theory of LLM Self-Adversariality: Modeling Severity Drift as a Critical Process 提出随机动力学理论以理解LLM自我对抗性问题 large language model chain-of-thought
10 Divergent Emotional Patterns in Disinformation on Social Media? An Analysis of Tweets and TikToks about the DANA in Valencia 分析社交媒体上关于西班牙DANA风暴的虚假信息，揭示不同平台的情感模式差异。 large language model multimodal
11 Contextual Agent Security: A Policy for Every Purpose 提出Conseca框架，为通用智能体生成即时、可验证的上下文安全策略 generalist agent
12 Over-Tokenized Transformer: Vocabulary is Generally Worth Scaling 提出Over-Tokenized Transformer，通过解耦输入输出词汇表提升语言模型性能。 large language model
13 FactCG: Enhancing Fact Checkers with Graph-Based Multi-Hop Data FactCG：利用图结构多跳数据增强事实核查器，提升LLM幻觉检测能力 large language model
14 Irony Detection, Reasoning and Understanding in Zero-shot Learning 提出IDADP框架，利用LLM零样本学习提升反讽检测、推理与理解能力 multimodal
15 An LLM Benchmark for Addressee Recognition in Multi-modal Multi-party Dialogue 构建多模态多人对话语料库，并用LLM基准测试解决寻址对象识别难题 large language model
16 Tailored Truths: Optimizing LLM Persuasion with Personalization and Fabricated Statistics 利用个性化信息和虚假数据优化LLM说服力，提升信息操纵能力 large language model
17 Detecting harassment and defamation in cyberbullying with emotion-adaptive training 提出情感自适应训练框架，提升低资源场景下网络欺凌中骚扰与诽谤的检测性能。 large language model
18 Through the Prism of Culture: Evaluating LLMs' Understanding of Indian Subcultures and Traditions 评估LLM对印度亚文化理解：揭示文化偏见与提升文化敏感性 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (5 篇)

#题目一句话要点标签🔗⭐
19 CHiP: Cross-modal Hierarchical Direct Preference Optimization for Multimodal LLMs 提出CHiP：跨模态分层直接偏好优化，缓解多模态LLM幻觉问题 DPO direct preference optimization large language model ✅
20 Memorize and Rank: Elevating Large Language Models for Clinical Diagnosis Prediction MERA：利用大语言模型进行临床诊断预测，解决数据稀缺和候选疾病空间大的问题 contrastive learning large language model
21 Multimodal Magic Elevating Depression Detection with a Fusion of Text and Audio Intelligence 提出基于教师-学生架构的多模态抑郁症检测模型，提升文本与音频特征融合效果。 teacher-student multimodal
22 xJailbreak: Representation Space Guided Reinforcement Learning for Interpretable LLM Jailbreaking 提出xJailbreak，利用表征空间引导强化学习实现可解释的LLM越狱攻击 reinforcement learning large language model ✅
23 COS(M+O)S: Curiosity and RL-Enhanced MCTS for Exploring Story Space via Language Models COS(M+O)S：结合好奇心和强化学习的MCTS，用于探索语言模型的故事空间。 reinforcement learning chain-of-thought

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

#题目一句话要点标签🔗⭐
24 Why Do We Laugh? Annotation and Taxonomy Generation for Laughable Contexts in Spontaneous Text Conversation 提出一种基于LLM辅助的笑点分类方法，用于提升对话AI的自然交互能力。 HuMoR

⬅️ 返回 cs.CL 首页 · 🏠 返回主页