cs.CL（2026-01-16）

📊 共 22 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (19 🔗4) 支柱二：RL算法与架构 (RL & Architecture) (2) 支柱四：生成式动作 (Generative Motion) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (19 篇)

#	题目	一句话要点	标签	🔗	⭐
1	Neural Chain-of-Thought Search: Searching the Optimal Reasoning Path to Enhance Large Language Models	提出神经链式思考搜索(NCoTS)，优化大语言模型的推理路径，提升准确率并减少计算成本。	large language model chain-of-thought	✅
2	Hierarchical Orthogonal Residual Spread for Precise Massive Editing in Large Language Models	提出HORSE方法，通过分层正交残差扩展实现大语言模型中的精确批量编辑。	large language model	✅
3	FactCorrector: A Graph-Inspired Approach to Long-Form Factuality Correction of Large Language Models	FactCorrector：一种图结构驱动的大语言模型长文本事实性纠正方法	large language model
4	Language of Thought Shapes Output Diversity in Large Language Models	提出多语言思维方法，显著提升大型语言模型输出的多样性	large language model	✅
5	ZPD Detector: Data Selection via Capability-Difficulty Alignment for Large Language Models	提出ZPD Detector，通过能力-难度对齐进行大语言模型数据选择	large language model
6	MultiCaption: Detecting disinformation using multilingual visual claims	提出MultiCaption数据集，用于检测多语言视觉声明中的虚假信息。	large language model multimodal
7	How DDAIR you? Disambiguated Data Augmentation for Intent Recognition	提出DDAIR，通过消除歧义的数据增强提升低资源意图识别。	large language model
8	CoG: Controllable Graph Reasoning via Relational Blueprints and Failure-Aware Refinement over Knowledge Graphs	CoG：通过关系蓝图和失败感知精炼实现知识图谱上的可控图推理	large language model
9	Relational Linearity is a Predictor of Hallucinations	提出关系线性性预测模型幻觉现象	large language model
10	Evaluating LLM Behavior in Hiring: Implicit Weights, Fairness Across Groups, and Alignment with Human Preferences	评估LLM在招聘中的行为：隐式权重、群体公平性及与人类偏好的一致性	large language model
11	Spectral Characterization and Mitigation of Sequential Knowledge Editing Collapse	提出REVIVE框架，通过谱分析缓解大语言模型序列知识编辑中的灾难性崩溃问题	large language model
12	Finding the Translation Switch: Discovering and Exploiting the Task-Initiation Features in LLMs	利用稀疏自编码器发现LLM翻译启动特征，提升翻译效率与鲁棒性	large language model	✅
13	The unreasonable effectiveness of pattern matching	大型语言模型展现惊人的模式匹配能力，可理解乱语并恢复语义	large language model
14	Idea First, Code Later: Disentangling Problem Solving from Code Generation in Evaluating LLMs for Competitive Programming	提出“先思路后代码”的评估框架，解耦LLM在编程竞赛中的问题解决与代码生成能力。	large language model
15	F-Actor: Controllable Conversational Behaviour in Full-Duplex Models	提出F-Actor：一个可控的全双工会话语音模型，提升人机交互自然度。	instruction following
16	Membership Inference on LLMs in the Wild	提出SimMIA框架以解决LLMs的成员推断攻击问题	large language model
17	One LLM to Train Them All: Multi-Task Learning Framework for Fact-Checking	提出基于多任务学习的自动事实核查框架，提升小模型性能。	large language model
18	Budget-Aware Anytime Reasoning with LLM-Synthesized Preference Data	提出预算感知的随时推理框架，利用LLM合成偏好数据提升推理效率	large language model
19	NAACL: Noise-AwAre Verbal Confidence Calibration for LLMs in RAG Systems	提出NAACL框架，解决RAG系统中LLM因噪声上下文导致的置信度校准问题	large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (2 篇)

#	题目	一句话要点	标签	🔗	⭐
20	Do explanations generalize across large reasoning models?	研究大型推理模型解释的泛化性，评估其在不同模型间的一致性	reinforcement learning chain-of-thought
21	Isotropy-Optimized Contrastive Learning for Semantic Course Recommendation	提出基于各向同性优化的对比学习方法，用于语义课程推荐	contrastive learning

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#	题目	一句话要点	标签	🔗	⭐
22	T$^\star$: Progressive Block Scaling for MDM Through Trajectory Aware RL	提出T$^ullet$以实现MDM的渐进块规模扩展	MDM

⬅️ 返回 cs.CL 首页 · 🏠 返回主页