cs.CL（2026-05-14）

📊 共 26 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 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (3) 支柱六：视频提取与匹配 (Video Extraction) (1 🔗1)

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

#题目一句话要点标签🔗⭐
1 Text Knows What, Tables Know When: Clinical Timeline Reconstruction via Retrieval-Augmented Multimodal Alignment 提出检索增强的多模态对齐框架，用于临床时间线重建，提升时间戳精度。 large language model multimodal
2 AI Knows When It's Being Watched: Functional Strategic Action and Contextual Register Modulation in Large Language Models 大型语言模型在社会观察下表现出策略性行为和语境适应性，揭示其社会感知能力 large language model
3 Non-linear Interventions on Large Language Models 提出非线性干预方法，突破线性干预局限，提升大语言模型控制能力 large language model
4 From Scenes to Elements: Multi-Granularity Evidence Retrieval for Verifiable Multimodal RAG 提出GranuRAG，通过多粒度证据检索解决可验证多模态RAG中细粒度查询问题。 multimodal
5 Tokenizer Fertility and Zero-Shot Performance of Foundation Models on Ukrainian Legal Text: A Comparative Study 对比研究：乌克兰语法律文本上大模型分词器效率与零样本性能 foundation model
6 Dimension-Level Intent Fidelity Evaluation for Large Language Models: Evidence from Structured Prompt Ablation 提出维度级意图保真度评估框架，用于评估大语言模型在用户特定任务中的表现。 large language model
7 Video2GUI: Synthesizing Large-Scale Interaction Trajectories for Generalized GUI Agent Pretraining 提出Video2GUI框架，从互联网视频合成大规模GUI交互轨迹，用于GUI Agent预训练。 large language model multimodal
8 MeMo: Memory as a Model MeMo：提出一种基于记忆的模块化框架，用于增强LLM的知识更新能力。 large language model
9 Improving Multi-turn Dialogue Consistency with Self-Recall Thinking 提出自回忆思考(SRT)框架，提升多轮对话一致性并降低延迟。 large language model
10 Graphs of Research: Citation Evolution Graphs as Supervision for Research Idea Generation 提出Graphs of Research以解决研究创意生成中的引用关系问题 large language model
11 Is Grep All You Need? How Agent Harnesses Reshape Agentic Search 研究Agentic Search中Grep与向量检索的性能差异及影响因素 large language model
12 MetaBackdoor: Exploiting Positional Encoding as a Backdoor Attack Surface in LLMs MetaBackdoor：利用LLM中的位置编码作为后门攻击面 large language model
13 EndPrompt: Efficient Long-Context Extension via Terminal Anchoring EndPrompt：通过末端锚定的高效长文本扩展方法 large language model ✅
14 Mining Subscenario Refactoring Opportunities in Behaviour-Driven Software Test Suites: ML Classifiers and LLM-Judge Baselines 提出一种基于机器学习和LLM的BDD测试套件子场景重构机会挖掘方法 large language model
15 GroupMemBench: Benchmarking LLM Agent Memory in Multi-Party Conversations GroupMemBench：用于评估LLM Agent在多方对话中记忆能力的基准测试。 large language model
16 Retrieval-Augmented Large Language Models for Schema-Constrained Clinical Information Extraction 提出检索增强的LLM用于模式约束的临床信息抽取，提升护士-患者对话记录结构化效率。 large language model
17 FINESSE-Bench: A Hierarchical Benchmark Suite for Financial Domain Knowledge and Technical Analysis in Large Language Models FINESSE-Bench：用于评估大语言模型金融领域知识和技术分析能力的分层基准套件 large language model
18 Reasoning Models Don't Just Think Longer, They Move Differently 通过轨迹几何分析，揭示推理模型在不同难度问题上的行为差异 chain-of-thought
19 Neural Activation Patterns Across Language Model Architectures: A Comprehensive Analysis of Cognitive Task Performance 分析六种LLM架构在认知任务中的神经激活模式，揭示模型特性。 large language model
20 Capability Conditioned Scaffolding for Professional Human LLM Collaboration 提出能力条件支架框架，提升专业领域人机协作可靠性 large language model
21 Greedy or not, here I come: Language production under vocabulary constraints in humans and resource-rational models 研究人类在词汇约束下的语言生成，对比贪婪与全局优化模型 large language model
22 Fluency and Faithfulness in Human and Machine Literary Translation 研究表明文学翻译中流畅性与忠实性存在权衡，大型语言模型亦然。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (3 篇)

#题目一句话要点标签🔗⭐
23 Reinforcement Learning with Semantic Rewards Enables Low-Resource Language Expansion without Alignment Tax 提出基于语义奖励的强化学习方法，解决低资源语言扩展中的“对齐税”问题。 reinforcement learning large language model
24 Learning from Failures: Correction-Oriented Policy Optimization with Verifiable Rewards 提出CIPO，利用失败轨迹进行纠正学习，提升LLM推理能力 reinforcement learning large language model
25 Ideology Prediction of German Political Texts 提出基于Transformer的模型，用于预测德语政治文本的意识形态倾向。 MAE foundation model

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

#题目一句话要点标签🔗⭐
26 COTCAgent: Preventive Consultation via Probabilistic Chain-of-Thought Completion COTCAgent：通过概率性思维链补全实现预防性临床咨询 feature matching large language model chain-of-thought ✅

⬅️ 返回 cs.CL 首页 · 🏠 返回主页