cs.CL（2026-02-02）

📊 共 45 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) (37 🔗7) 支柱二：RL算法与架构 (RL & Architecture) (6 🔗1) 支柱一：机器人控制 (Robot Control) (2)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (37 篇)

#题目一句话要点标签🔗⭐
1 S3-CoT: Self-Sampled Succinct Reasoning Enables Efficient Chain-of-Thought LLMs S3-CoT：自采样精简推理，实现高效思维链大语言模型 large language model chain-of-thought ✅
2 Abstract Activation Spaces for Content-Invariant Reasoning in Large Language Models 提出抽象激活空间框架，提升大语言模型在内容无关推理中的鲁棒性 large language model
3 Game of Thought: Robust Information Seeking with Large Language Models Using Game Theory 提出Game of Thought框架，利用博弈论提升大语言模型在信息搜寻任务中的鲁棒性。 large language model
4 Large Language Models for Mental Health: A Multilingual Evaluation 多语言心理健康领域：评估大型语言模型性能与翻译质量影响 large language model
5 Evaluating Metalinguistic Knowledge in Large Language Models across the World's Languages 构建多语言元语言知识评估基准，揭示大语言模型在不同语言上的结构理解能力差异 large language model
6 AR-MAP: Are Autoregressive Large Language Models Implicit Teachers for Diffusion Large Language Models? AR-MAP：利用自回归大语言模型作为扩散大语言模型的隐式教师，实现高效偏好对齐。 large language model ✅
7 There Is More to Refusal in Large Language Models than a Single Direction 揭示大语言模型拒绝行为的复杂性：并非单一激活方向控制 large language model
8 Orthogonal Hierarchical Decomposition for Structure-Aware Table Understanding with Large Language Models 提出正交分层分解框架，提升LLM对复杂表格的理解与推理能力 large language model
9 Data Distribution Matters: A Data-Centric Perspective on Context Compression for Large Language Model 从数据分布角度出发，研究数据分布对大语言模型上下文压缩质量的影响。 large language model
10 Mechanistic Indicators of Steering Effectiveness in Large Language Models 利用内部模型信号诊断大语言模型steering有效性，提升行为控制可靠性 large language model
11 Steering Vector Fields for Context-Aware Inference-Time Control in Large Language Models 提出Steering Vector Fields，解决大语言模型推理时控制向量的不可靠性问题 large language model
12 LEC-KG: An LLM-Embedding Collaborative Framework for Domain-Specific Knowledge Graph Construction -- A Case Study on SDGs 提出LEC-KG框架，利用LLM与知识图谱嵌入协同构建领域知识图谱，以解决可持续发展目标知识图谱构建难题。 large language model chain-of-thought
13 ROG: Retrieval-Augmented LLM Reasoning for Complex First-Order Queries over Knowledge Graphs 提出ROG：一种检索增强的LLM推理框架，用于解决知识图谱上的复杂一阶逻辑查询 large language model chain-of-thought
14 Cross-Lingual Stability of LLM Judges Under Controlled Generation: Evidence from Finno-Ugric Languages 研究表明，在芬兰-乌戈尔语系中，LLM评判器跨语言稳定性不足，尤其在语用判断上。 large language model instruction following ✅
15 CodeOCR: On the Effectiveness of Vision Language Models in Code Understanding CodeOCR：探索视觉语言模型在代码理解中的有效性，实现高效代码表示。 large language model multimodal
16 Why Steering Works: Toward a Unified View of Language Model Parameter Dynamics 提出统一视角以优化语言模型控制方法 large language model ✅
17 Reward-free Alignment for Conflicting Objectives 提出RACO框架，通过冲突规避梯度下降解决多目标LLM对齐问题 large language model
18 MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents MemSkill：学习和进化记忆技能，赋能自进化Agent large language model
19 OpenSeal: Good, Fast, and Cheap Construction of an Open-Source Southeast Asian LLM via Parallel Data OpenSeal：通过并行数据高效构建开源东南亚语言大模型 large language model
20 Towards AI Evaluation in Domain-Specific RAG Systems: The AgriHubi Case Study AgriHubi：面向芬兰语农业决策支持的领域RAG系统及评估方法 large language model
21 Automated Multiple Mini Interview (MMI) Scoring 提出多智能体提示框架，提升大型语言模型在自动化多重迷你面试评分中的表现 large language model
22 Language Steering for Multilingual In-Context Learning 提出语言向量引导方法，提升多语言上下文学习中非英语语言性能。 large language model
23 The Shape of Beliefs: Geometry, Dynamics, and Interventions along Representation Manifolds of Language Models' Posteriors 研究LLM信念的几何结构，提出场感知线性引导方法以提升干预效果 large language model
24 Am I More Pointwise or Pairwise? Revealing Position Bias in Rubric-Based LLM-as-a-Judge 揭示基于评分细则的LLM评判中的位置偏差，并提出平衡排列策略进行缓解。 large language model
25 Focus-dLLM: Accelerating Long-Context Diffusion LLM Inference via Confidence-Guided Context Focusing 提出Focus-dLLM，通过置信度引导的上下文聚焦加速长文本扩散LLM推理。 large language model ✅
26 Out of the Memory Barrier: A Highly Memory Efficient Training System for LLMs with Million-Token Contexts OOMB：一种高内存效率的LLM训练系统，支持百万token上下文 large language model ✅
27 Dicta-LM 3.0: Advancing The Frontier of Hebrew Sovereign LLMs 发布Dicta-LM 3.0：推进希伯来语主权LLM前沿，提供多种尺寸和工具调用支持。 large language model
28 Beyond Local Edits: Embedding-Virtualized Knowledge for Broader Evaluation and Preservation of Model Editing 提出嵌入虚拟知识以解决模型编辑评估不足问题 large language model
29 From Code-Centric to Concept-Centric: Teaching NLP with LLM-Assisted "Vibe Coding" 提出LLM辅助的“Vibe Coding”教学法，提升NLP概念理解。 large language model
30 AXE: Low-Cost Cross-Domain Web Structured Information Extraction AXE：提出一种低成本的跨领域网页结构化信息抽取方法，利用小型LLM实现高效抽取。 large language model
31 WorldCup Sampling for Multi-bit LLM Watermarking 提出WorldCup，通过层级竞争采样实现多比特LLM水印嵌入，提升容量与鲁棒性。 large language model
32 ARTIS: Agentic Risk-Aware Test-Time Scaling via Iterative Simulation 提出ARTIS，通过迭代模拟实现Agent在测试时风险感知的计算资源分配，提升Agent可靠性。 large language model
33 The Art of Socratic Inquiry: A Framework for Proactive Template-Guided Therapeutic Conversation Generation 提出苏格拉底探究框架，提升LLM在认知行为治疗中的主动引导能力 large language model
34 Provable Defense Framework for LLM Jailbreaks via Noise-Augumented Alignment 提出噪声增强对齐的认证防御框架，提升LLM抵抗恶意越狱攻击的鲁棒性。 large language model
35 LLM-based Embeddings: Attention Values Encode Sentence Semantics Better Than Hidden States 提出基于LLM注意力值的句子嵌入方法，优于基于隐藏状态的方法 large language model
36 FS-Researcher: Test-Time Scaling for Long-Horizon Research Tasks with File-System-Based Agents FS-Researcher：面向长程研究任务，基于文件系统的LLM智能体测试时扩展框架 large language model ✅
37 Argument Rarity-based Originality Assessment for AI-Assisted Writing 提出基于论证稀有度的原创性评估框架AROA，用于评估AI辅助写作中学生文章的原创性。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (6 篇)

#题目一句话要点标签🔗⭐
38 PretrainRL: Alleviating Factuality Hallucination of Large Language Models at the Beginning PretrainRL：通过强化学习预训练缓解大语言模型的事实性幻觉问题 reinforcement learning large language model
39 Training LLMs for Divide-and-Conquer Reasoning Elevates Test-Time Scalability 提出基于强化学习的DAC训练框架，提升LLM在复杂推理任务上的可扩展性 reinforcement learning large language model chain-of-thought
40 SafePred: A Predictive Guardrail for Computer-Using Agents via World Models SafePred：基于世界模型的计算机代理预测性安全防护 world model
41 Proof-RM: A Scalable and Generalizable Reward Model for Math Proof 提出Proof-RM，一种可扩展且泛化的数学证明奖励模型。 reinforcement learning large language model
42 Kimi K2.5: Visual Agentic Intelligence Kimi K2.5：面向通用智能体的视觉代理智能模型与Agent Swarm框架 reinforcement learning multimodal
43 D-CORE: Incentivizing Task Decomposition in Large Reasoning Models for Complex Tool Use D-CORE：通过激励任务分解提升大型推理模型在复杂工具使用中的性能 reinforcement learning distillation ✅

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
44 Indications of Belief-Guided Agency and Meta-Cognitive Monitoring in Large Language Models 评估大语言模型中的信念引导代理与元认知监控 manipulation large language model
45 Quantifying the Gap between Understanding and Generation within Unified Multimodal Models 提出GapEval基准，量化统一多模态模型理解与生成能力之间的差距 manipulation multimodal

⬅️ 返回 cs.CL 首页 · 🏠 返回主页