cs.CL（2025-10-05）

📊 共 22 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) (16 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (5 🔗1) 支柱六：视频提取与匹配 (Video Extraction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (16 篇)

#题目一句话要点标签🔗⭐
1 Evaluation of Clinical Trials Reporting Quality using Large Language Models 利用大型语言模型评估临床试验报告质量 large language model chain-of-thought
2 Scaling Code-Assisted Chain-of-Thoughts and Instructions for Model Reasoning 提出Caco框架，通过代码驱动自动生成高质量、可验证、多样化的指令-CoT推理数据，提升模型推理能力。 large language model chain-of-thought
3 Systematic Diagnosis of Brittle Reasoning in Large Language Models 提出数学推理诊断框架，揭示大语言模型在组合推理上的脆弱性 large language model
4 Large Language Models Hallucination: A Comprehensive Survey 全面综述：剖析大语言模型幻觉现象的成因、检测与缓解策略 large language model
5 Equipping Retrieval-Augmented Large Language Models with Document Structure Awareness RDR2：提出文档结构感知的检索增强大语言模型框架，提升复杂场景知识利用能力。 large language model
6 Epistemic Diversity and Knowledge Collapse in Large Language Models 提出衡量LLM知识多样性的方法，揭示其知识塌缩风险及影响因素。 large language model
7 Self Speculative Decoding for Diffusion Large Language Models 提出自推测解码(SSD)加速扩散大语言模型推理，无需额外模块。 large language model
8 Benchmarking Open-Source Large Language Models for Persian in Zero-Shot and Few-Shot Learning 评估开源大语言模型在波斯语零样本和少样本学习中的性能表现 large language model
9 Thinking on the Fly: Test-Time Reasoning Enhancement via Latent Thought Policy Optimization 提出LTPO，通过优化隐空间向量提升LLM在测试时的推理能力 large language model chain-of-thought
10 Read the Scene, Not the Script: Outcome-Aware Safety for LLMs 提出CS-Chain-4k数据集，解决LLM中结果盲视的安全对齐问题 large language model
11 Probing Geometry of Next Token Prediction Using Cumulant Expansion of the Softmax Entropy 提出基于累积量展开的框架，用于探究LLM预测下一个token时的几何结构。 large language model
12 Fine Tuning Methods for Low-resource Languages 针对低资源语言，提出一种通用数据集构建与Gemma 2模型微调方法。 large language model
13 Unveiling LLMs' Metaphorical Understanding: Exploring Conceptual Irrelevance, Context Leveraging and Syntactic Influence 揭示大语言模型隐喻理解能力：概念不相关性、语境利用与句法影响分析 large language model
14 Does Using Counterfactual Help LLMs Explain Textual Importance in Classification? 研究反事实推理对LLM文本分类重要性解释能力的影响，并提出决策变化率评估框架。 large language model
15 LLM Microscope: What Model Internals Reveal About Answer Correctness and Context Utilization LLM Microscope：利用模型内部激活预测答案正确性与上下文利用率 large language model ✅
16 Simulating and Understanding Deceptive Behaviors in Long-Horizon Interactions 提出长时交互欺骗行为模拟框架，揭示LLM在动态压力下的欺骗风险。 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🔗⭐
17 Exploring Chain-of-Thought Reasoning for Steerable Pluralistic Alignment 探索思维链推理以实现可控的多元化对齐 reinforcement learning large language model chain-of-thought
18 Pushing on Multilingual Reasoning Models with Language-Mixed Chain-of-Thought 提出Language-Mixed CoT，提升多语言推理模型在韩语等场景下的性能。 distillation chain-of-thought ✅
19 Teaching LLM to be Persuasive: Reward-Enhanced Policy Optimization for Alignment frm Heterogeneous Rewards 提出REPO框架，通过异构奖励优化LLM，提升在线旅游议价场景的说服力。 reinforcement learning PPO DPO
20 PoLi-RL: A Point-to-List Reinforcement Learning Framework for Conditional Semantic Textual Similarity 提出PoLi-RL框架，解决条件语义文本相似度任务中强化学习训练难题。 reinforcement learning large language model
21 AgriGPT-VL: Agricultural Vision-Language Understanding Suite AgriGPT-VL：农业视觉-语言理解套件，解决领域模型稀缺问题 reinforcement learning large language model multimodal

🔬 支柱六：视频提取与匹配 (Video Extraction) (1 篇)

#题目一句话要点标签🔗⭐
22 Visual Lifelog Retrieval through Captioning-Enhanced Interpretation 提出CIVIL系统，通过图像描述增强的视觉生活日志检索，解决第一人称视角下的记忆检索问题。 first-person view

⬅️ 返回 cs.CL 首页 · 🏠 返回主页