cs.CL（2025-09-11）

📊 共 19 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (14 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (5)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (14 篇)

#题目一句话要点标签🔗⭐
1 Pragmatic Frames Evoked by Gestures: A FrameNet Brasil Approach to Multimodality in Turn Organization 提出基于FrameNet Brasil的多模态对话轮次组织建模框架 multimodal
2 GrACE: A Generative Approach to Better Confidence Elicitation in Large Language Models GrACE：一种生成式方法，提升大语言模型置信度评估的可靠性与可扩展性 large language model
3 From scratch to silver: Creating trustworthy training data for patent-SDG classification using Large Language Models 利用大型语言模型，为专利-SDG分类创建可信的弱监督训练数据。 large language model
4 Reading Between the Lines: Classifying Resume Seniority with Large Language Models 利用大型语言模型进行简历资历分类，解决经验夸大和自我包装问题。 large language model
5 LLM-JEPA: Large Language Models Meet Joint Embedding Predictive Architectures 提出LLM-JEPA，将联合嵌入预测架构应用于LLM预训练和微调，显著提升性能。 large language model ✅
6 LLM Architecture, Scaling Laws, and Economics: A Quick Summary 总结LLM架构、扩展法则与经济性，为快速理解提供参考 large language model
7 Steering MoE LLMs via Expert (De)Activation SteerMoE：通过专家激活调控MoE LLM的安全性与可靠性 large language model
8 DeMeVa at LeWiDi-2025: Modeling Perspectives with In-Context Learning and Label Distribution Learning DeMeVa团队提出结合上下文学习与标签分布学习的视角建模方法，用于解决LeWiDi-2025任务。 large language model
9 Emulating Public Opinion: A Proof-of-Concept of AI-Generated Synthetic Survey Responses for the Chilean Case 利用大型语言模型生成合成调查回复，模拟智利公众意见，验证其可行性。 large language model
10 Hallucination Detection with the Internal Layers of LLMs 提出一种基于LLM内部层动态加权的幻觉检测方法，提升检测性能。 large language model
11 All for One: LLMs Solve Mental Math at the Last Token With Information Transferred From Other Tokens LLM通过末位Token计算解决心算问题，信息由其他Token传递 large language model
12 Bridging the Capability Gap: Joint Alignment Tuning for Harmonizing LLM-based Multi-Agent Systems 提出MOAT框架，通过联合对齐调整优化LLM多智能体系统协作能力 large language model
13 MetaRAG: Metamorphic Testing for Hallucination Detection in RAG Systems 提出MetaRAG框架以解决RAG系统中的幻觉检测问题 large language model
14 TigerCoder: A Novel Suite of LLMs for Code Generation in Bangla TigerCoder：面向孟加拉语代码生成的新型LLM套件 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🔗⭐
15 CDE: Curiosity-Driven Exploration for Efficient Reinforcement Learning in Large Language Models 提出好奇心驱动探索(CDE)框架，提升大型语言模型在强化学习中的探索效率。 reinforcement learning PPO large language model
16 Target-oriented Multimodal Sentiment Classification with Counterfactual-enhanced Debiasing 提出一种反事实增强去偏框架，用于解决目标导向的多模态情感分类中存在的偏见问题。 contrastive learning multimodal
17 Topic-Guided Reinforcement Learning with LLMs for Enhancing Multi-Document Summarization 提出主题引导的强化学习方法，利用LLM提升多文档摘要生成效果 reinforcement learning large language model
18 MR-UIE: Multi-Perspective Reasoning with Reinforcement Learning for Universal Information Extraction 提出MR-UIE，结合强化学习与多视角推理，提升通用信息抽取在复杂结构化输出场景下的性能。 reinforcement learning large language model
19 Compass-v3: Scaling Domain-Specific LLMs for Multilingual E-Commerce in Southeast Asia Compass-v3：面向东南亚电商的多语言领域专用LLM，性能超越GPT-4 direct preference optimization large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页