cs.CL（2026-04-24）

📊 共 17 篇论文 | 🔗 2 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (13 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (3) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (13 篇)

#	题目	一句话要点	标签	🔗	⭐
1	Large Language Models Decide Early and Explain Later	通过提前停止策略，减少大语言模型CoT推理冗余，降低计算成本。	large language model chain-of-thought
2	CNSL-bench: Benchmarking the Sign Language Understanding Capabilities of MLLMs on Chinese National Sign Language	提出CNSL-bench，用于评估MLLM在中文国家通用手语理解上的能力。	large language model multimodal instruction following
3	When AI Speaks, Whose Values Does It Express? A Cross-Cultural Audit of Individualism-Collectivism Bias in Large Language Models	揭示大型语言模型中的文化价值观偏见：以个人主义-集体主义为例	large language model
4	Preference Heads in Large Language Models: A Mechanistic Framework for Interpretable Personalization	提出差分偏好引导（DPS）框架，实现大语言模型中可控且可解释的个性化	large language model
5	How Large Language Models Balance Internal Knowledge with User and Document Assertions	提出三源交互框架，评估大语言模型在用户、文档和内部知识间的平衡能力。	large language model	✅
6	Voice Under Revision: Large Language Models and the Normalization of Personal Narrative	大型语言模型重写个人叙事文本导致风格标准化，影响文本分析。	large language model
7	Controllable Spoken Dialogue Generation: An LLM-Driven Grading System for K-12 Non-Native English Learners	提出DDPO算法，构建LLM驱动的K-12非母语英语学习者可控对话生成系统	large language model
8	ReLeVAnT: Relevance Lexical Vectors for Accurate Legal Text Classification	提出ReLeVAnT框架，利用相关性词汇向量实现精准法律文本分类	multimodal
9	Can QPP Choose the Right Query Variant? Evaluating Query Variant Selection for RAG Pipelines	提出QPP机制以优化RAG管道中的查询变体选择	large language model
10	Dharma, Data and Deception: An LLM-Powered Rhetorical Analysis of Cow-Urine Health Claims on YouTube	利用大语言模型对YouTube上牛尿疗效言论进行修辞分析，揭示信息传播策略。	large language model
11	Bridging the Long-Tail Gap: Robust Retrieval-Augmented Relation Completion via Multi-Stage Paraphrase Infusion	提出RC-RAG，通过多阶段释义注入增强检索增强关系补全，尤其提升长尾数据性能。	large language model
12	Recognition Without Authorization: LLMs and the Moral Order of Online Advice	分析LLM在在线建议中的道德判断差异，揭示其“认知而不授权”的模式	large language model
13	SHAPE: Unifying Safety, Helpfulness and Pedagogy for Educational LLMs	提出SHAPE基准与图增强教学管线，提升教育LLM的安全性、有效性和教学性	large language model	✅

🔬 支柱二：RL算法与架构 (RL & Architecture) (3 篇)

#	题目	一句话要点	标签	🔗	⭐
14	Thinking Without Words: Efficient Latent Reasoning with Abstract Chain-of-Thought	提出抽象思维链（Abstract-CoT），通过隐空间推理提升语言模型效率，同时保持性能。	reinforcement learning distillation instruction following
15	Selective Contrastive Learning For Gloss Free Sign Language Translation	提出选择性对比学习以解决无注释手语翻译问题	contrastive learning
16	Where Should LoRA Go? Component-Type Placement in Hybrid Language Models	混合语言模型中组件类型LoRA放置策略研究，显著提升微调效率	Mamba SSM

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#	题目	一句话要点	标签	🔗	⭐
17	Fine-Grained Analysis of Shared Syntactic Mechanisms in Language Models	通过因果可解释性分析，揭示语言模型中共享句法机制的细粒度特征	manipulation

⬅️ 返回 cs.CL 首页 · 🏠 返回主页