cs.CL（2026-09-17）

📊 共 24 篇论文 | 🔗 8 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (18 🔗7) 支柱二：RL算法与架构 (RL & Architecture) (5 🔗1) 支柱四：生成式动作 (Generative Motion) (1)

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

#	题目	一句话要点	标签	🔗	⭐
1	Chain-of-Thought Entropy as a Reliability Signal: A Preregistered Reproduction	通过链式思维熵信号提升语言模型答案可靠性	large language model chain-of-thought
2	Summarization Bias: The Directional Collapse of Objective Projection into Told-Mode Labels in Large Language Models --- A Conceptual Framework and Registered Test Protocol	提出总结偏差以解决大语言模型叙事理解问题	large language model
3	From Parameters to Behaviors: A Survey of Model Fusion for Large Language Models	提出模型融合方法以整合大语言模型的能力	large language model	✅
4	SAFARI: An Industrial Benchmark for LLM-Assisted Hazard Analysis and Risk Assessment	提出SAFARI基准以解决LLM辅助的汽车风险评估问题	large language model chain-of-thought	✅
5	Learn Before You Judge: Progressive Knowledge-to-Decision Alignment for Explainable Hateful Meme Detection	提出ProKDA以解决仇恨表情包检测中的解释性与准确性问题	large language model multimodal	✅
6	Chronicle: Cut-Point Replay for Regression Testing of LLM Agents	提出Chronicle以解决LLM代理回归测试中的非确定性问题	large language model	✅
7	Xeno-Interpretability: Investigating the Alien Minds of LLMs	提出外部可解释性以探索大型语言模型的内部结构	large language model
8	KoNeoBench: A Curated Evaluation Dataset for LLM Understanding of Korean Neologisms	提出KoNeoBench以解决韩语新词理解评估问题	large language model	✅
9	Form Over Content In Gradient-Based Data Attribution Methods	提出基于梯度相似性的格式优先于内容的数据归因方法	large language model
10	Steering the Compass: Aligning Dynamic Psychological Counseling Conversations with Cognitive Behavioral Therapy Strategies	提出StratCBT数据集以提升心理咨询对话的灵活性与效果	large language model	✅
11	An Analysis of Training-Free Self-Reported Confidence in Language Models	分析语言模型自报告信心的训练无关信号	large language model
12	To Copy or Not to Copy: Controlling Speculative Decoding via Intrinsic Model Signals	提出SwitchSD以解决现有复制策略的效率问题	large language model
13	Think Thrice Before Reranking: Multi-perspective Evidence and Reasoning Integration for Text Reranking	提出MERIT-Rank以解决文本重排序中的单一推理轨迹问题	large language model
14	DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression	提出DeepSeek-V4.1-Flash以解决KV缓存压缩问题	multimodal	✅
15	Before the Arrest: Benchmarking LLMs on Criminal Profiling from Incomplete Evidence	提出PIJ基准以解决犯罪侦查中的推理挑战	large language model
16	F$^{2}$DR: A Fine-Grained Full-Pipeline Reward Framework for DeepSearch Workflows	提出F2DR框架以解决DeepSearch工作流评估问题	large language model
17	CoLearn: An Agentic Tutor that Learns its Learner in a Human--AI Co-Learning Loop	提出CoLearn以解决个性化辅导不足的问题	large language model
18	$μ^2$-Bench: A Multilingual Machine Unlearning Benchmark	提出$μ^2$-Bench以评估多语言机器遗忘能力	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	JEPA-Anything: Learning Predictive Models across Different Worlds	提出JEPA-Anything以解决跨领域预测模型的局限性问题	world model world models JEPA	✅
20	RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning	提出RetireOPD以解决多回合强化学习中的教师监督不可靠问题	reinforcement learning distillation privileged information
21	What Does Privileged Information Add to On-Policy Self-Distillation?	提出基于特权信息的自蒸馏方法以提升语言模型性能	distillation privileged information
22	What Does Privileged Information Add to On-Policy Self-Distillation?	提出OPSD方法以提升语言模型的自我蒸馏效果	distillation privileged information
23	Geometry of Values: Task Vector Composition for Ethical Preference Alignment in Language Models	提出任务向量组合方法以解决语言模型的伦理偏好对齐问题	direct preference optimization large language model instruction following

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#	题目	一句话要点	标签	🔗	⭐
24	Zarya: A Hybrid Autoregressive--Masked Diffusion Language Model with Flexible Training and Dual-Mode Inference	提出Zarya以解决自回归与掩蔽扩散模型的局限性	MDM

⬅️ 返回 cs.CL 首页 · 🏠 返回主页