cs.CL（2026-09-28）

📊 共 32 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) (22) 支柱二：RL算法与架构 (RL & Architecture) (10 🔗2)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (22 篇)

#题目一句话要点标签🔗⭐
1 Reference-Grounded Data Curation for Instruction-Following Thai-English Machine Translation 提出参考基础数据整理方法以解决指令遵循机器翻译问题 instruction following
2 How to Tame a Multi-Headed Hydra? Adaptive Multi-Category Safety Steering for Large Language Models 提出CAM-Steer框架以解决多类别安全引导问题 large language model
3 Toward a Graded Measure of Belief Stability in Large Language Models 提出分级信念稳定性度量以评估大语言模型的可靠性 large language model
4 Understanding Clinical Cognitive Dialogues Using Large Language Models 提出大语言模型以理解临床认知对话结构 large language model
5 Who Gets a Token, and What Does It Carry? Unequal Name Support and Concept Access in Large Language Models 提出NameTrace框架以解决语言模型中的不平等名称支持问题 large language model
6 When Confidence Rises Too Early: Detecting Shortcut Reasoning via Premature Answer Commitment 提出ConfLens框架以解决模型早期自信导致的推理捷径问题 large language model chain-of-thought
7 FactorEngram: Factorized N-gram Memory with Basis-Level Gating for Language Models 提出FactorEngram以解决语言模型中记忆检索的局限性 large language model
8 A mechanistic study of language model introspection 研究语言模型内省机制以提升其内部状态感知能力 large language model
9 Echoes of Deeds: Moral History Can Shape and Steer LLM Behavioral Choices 提出MoralLedger框架以研究道德历史对LLM行为选择的影响 large language model
10 Semantic Uncertainty Quantification Needs Factual Equivalence 提出一种新方法以提高语义不确定性量化的准确性 large language model
11 Late Attention Layers Alone Can Copy Entity Tokens, but Not Without Attending to Their Context 提出新方法研究语言模型中的实体复制机制 large language model
12 AraDynFact: Dynamic Evaluation of Factual Knowledge in Arabic 提出AraDynFact以解决阿拉伯语知识评估不足问题 large language model
13 TRACE: Single-Pass Decoding-Trace Risk Localization for Generation Calibration 提出TRACE以解决大语言模型生成校准中的风险定位问题 large language model
14 MemoReason: Evaluating the Effect of Parametric Memory on Contextual Reasoning in LLMs 提出MemoReason以评估参数记忆对LLM上下文推理的影响 large language model
15 Epistemic Policy Divergence in Multi-Turn LLM Contamination: A Protocol-Gradient Investigation 提出五种污染协议以解决多轮对话中的知识污染问题 large language model
16 Measuring Collapse and Correction in Homogeneous-Panel LLM Debate 提出审计协议以解决多代理LLM辩论中的评估问题 large language model
17 SCBO: Semantically Coherent Batching and Ordering for LLM-Based Social Surveys 提出SCBO以解决LLM社交调查中的批处理与排序问题 large language model
18 Pass or Fail? Evaluating LLMs on Two Greek Examination Benchmarks 提出Prot-Ex和Pan-Ex以评估希腊语LLMs的能力 large language model
19 ReMCTS: Reflection-Enhanced Monte Carlo Tree Search for Code Generation 提出ReMCTS以解决代码生成中的语义隐患问题 large language model
20 Rewarding Novel Deductions: Solver-guided Process Rewards for Logical Reasoning 提出SPRING以解决小规模LLM逻辑推理问题 large language model
21 CRISP: Cultural Reward Modeling for Implicit Situated Propriety 提出CRISP-RM以解决文化适应性行为建模问题 large language model
22 LLMs are not stochastic parrots: Evidence for meaning-mediated abstraction from conlang-like tasks 通过构造语言任务验证LLM的意义中介抽象能力 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (10 篇)

#题目一句话要点标签🔗⭐
23 InfiMed2: A Generalist Medical Multimodal Foundation Model from Contextual Evidence and Stability-Aware Supervision 提出InfiMed2以解决医学多模态模型训练中的数据设计挑战 reinforcement learning foundation model multimodal
24 Unbiased Top-$k$ Estimation for On-Policy Distillation 提出TT-OPD以解决TK-OPD中的偏差问题 distillation large language model
25 Look Before You Select: Rethinking Vocabulary Sparsification in On-Policy Distillation 提出SparseOPD以解决长序列蒸馏中的内存瓶颈问题 distillation multimodal
26 MAS-OPD: On-Policy Distillation for Multi-agent Systems 提出MAS-OPD以解决多智能体系统中的角色协作问题 reinforcement learning distillation privileged information
27 MS-GLA: Multi-Scale Gated Linear Attention for Addressing Representational Bottlenecks via Multi-Temporal Resolution 提出MS-GLA以解决Gated Linear Attention的表示瓶颈问题 SSM linear attention
28 Rubric-Aware On-Policy Self-Distillation for LLM Personalization 提出GRASP框架以实现大语言模型个性化响应 reinforcement learning distillation ✅
29 ColNanoVDR: Document-Free Query Distillation for Multi-Vector Visual Document Retrieval via Optimal Transport 提出ColNanoVDR以解决多向量视觉文档检索中的查询编码瓶颈问题 distillation
30 Understanding On-Policy Distillation: A Mechanistic Interpretability Perspective via Sparse Crosscoders 提出稀疏交叉编码器以解析在政策蒸馏中的机制 distillation
31 Beyond Token Alignment: Event Completion for Cross-Tokenizer On-Policy Distillation 提出事件集完成蒸馏以解决跨标记器知识转移问题 distillation
32 When Words Fall Short: Iterative Synergy Between Verbalized Reasoning and Hidden Features for LLM Confidence Estimation 提出IPoET框架以提升LLM的置信度估计 policy learning large language model ✅

⬅️ 返回 cs.CL 首页 · 🏠 返回主页