cs.CL（2026-05-20）

📊 共 29 篇论文

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (25) 支柱二：RL算法与架构 (RL & Architecture) (2) 支柱六：视频提取与匹配 (Video Extraction) (1) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (25 篇)

#题目一句话要点标签🔗⭐
1 Automated ICD Classification of Psychiatric Diagnoses: From Classical NLP to Large Language Models 利用大型语言模型自动进行精神疾病ICD分类 large language model
2 Tracing the ongoing emergence of human-like reasoning in Large Language Models 评估大型语言模型在条件推理中类人推理能力的涌现 large language model
3 Towards Context-Invariant Safety Alignment for Large Language Models 提出锚定不变性正则化(AIR)以提升大语言模型在对抗性语境下的安全性。 large language model
4 LASH: Adaptive Semantic Hybridization for Black-Box Jailbreaking of Large Language Models 提出LASH以解决大型语言模型的黑箱越狱问题 large language model
5 Do No Harm? Hallucination and Actor-Level Abuse in Web-Deployed Medical Large Language Models 评估医疗大语言模型幻觉与滥用风险，揭示Web部署模型的安全隐患 large language model
6 Manga109-v2026: Revisiting Manga109 Annotations for Modern Manga Understanding Manga109-v2026：修订漫画109数据集，提升现代漫画理解能力 multimodal
7 TextReg: Mitigating Prompt Distributional Overfitting via Regularized Text-Space Optimization TextReg：通过正则化文本空间优化缓解提示分布过拟合 large language model
8 GraphRAG on Consumer Hardware: Benchmarking Local LLMs for Healthcare EHR Schema Retrieval 在消费级硬件上实现GraphRAG，评估本地LLM在医疗EHR模式检索中的性能。 large language model
9 The Illusion of Intervention: Your LLM-Simulated Experiment is an Observational Study 揭示LLM模拟实验中的干预幻觉：实为观测研究，关注用户漂移带来的偏差。 large language model
10 DIVE: Embedding Compression via Self-Limiting Gradient Updates DIVE：通过自限制梯度更新实现嵌入压缩，提升小样本检索性能。 large language model
11 Beyond Text-to-SQL: An Agentic LLM System for Governed Enterprise Analytics APIs 提出 Analytic Agent，利用LLM驱动的企业分析API智能体系统，解决传统Text-to-SQL在企业级应用中的局限性。 large language model
12 Strategy-Induct: Task-Level Strategy Induction for Instruction Generation Strategy-Induct：一种无需答案的任务级策略诱导指令生成方法 large language model
13 Terminal-World: Scaling Terminal-Agent Environments via Agent Skills Terminal-World：通过Agent技能扩展终端Agent环境，提升任务执行能力 large language model
14 Distributional Alignment as a Criterion for Designing Task Vectors in In-Context Learning 提出线性任务向量（LTV）方法，通过分布对齐提升上下文学习的任务向量性能。 large language model
15 Leveraging LLMs for Grammar Adaptation: A Study on Metamodel-Grammar Co-Evolution 提出基于LLM的语法适配方法，解决元模型演化后语法维护的难题。 large language model
16 Quantifying the cross-linguistic effects of syncretism on agreement attraction 利用大型语言模型量化形态同音异义对一致性吸引的跨语言影响 large language model
17 "I didn't Make the Micro Decisions": Measuring, Inducing, and Exposing Goal-Level AI Contributions in Collaboration 提出CoTrace框架，用于衡量人机协作中AI在目标塑造上的贡献 large language model
18 Do LLMs Know What Luxembourgish Borrows? Probing Lexical Neology in Low-Resource Multilingual Models 提出LexNeo-Bench基准，研究LLM在低资源语言中词汇借用和创新能力，并提出知识图谱增强的prompt方法。 large language model
19 WCXB: A Multi-Type Web Content Extraction Benchmark 提出WCXB多类型网页内容提取基准，揭示现有方法在结构化页面上的盲点。 large language model
20 LoCar: Localization-Aware Evaluation of In-Vehicle Assistants through Fine-Grained Sociolinguistic Control LoCar：通过细粒度社会语言控制，实现车载助手本地化感知评估 large language model
21 GradeLegal: Automated Grading for German Legal Cases GradeLegal：利用大型语言模型实现德国法律案例解答的自动评分 large language model
22 Cross-lingual robustness of LLM-brain alignment and its computational roots 研究表明LLM与大脑活动在跨语言场景下具有稳健的空间对齐性，但其计算根源尚不明确。 large language model
23 JobArabi: An Arabic Corpus and Analysis of Job Announcements from Social Media JobArabi：构建并分析来自社交媒体的阿拉伯语招聘信息语料库 TAMP
24 Assessing socio-economic climate impacts from text data 提出文本数据分析框架，提升气候灾害社会经济影响评估的准确性和可比性 large language model
25 HRM-Text: Efficient Pretraining Beyond Scaling 提出HRM-Text，通过层级循环模型和任务驱动预训练，显著降低大语言模型的预训练成本。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (2 篇)

#题目一句话要点标签🔗⭐
26 Mem-$π$: Adaptive Memory through Learning When and What to Generate 提出Mem-$π$，通过学习何时以及生成什么来构建LLM Agent的自适应记忆。 reinforcement learning large language model
27 Auto-Dreamer: Learning Offline Memory Consolidation for Language Agents Auto-Dreamer：学习离线记忆整合，提升语言智能体的跨任务知识复用能力 dreamer

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

#题目一句话要点标签🔗⭐
28 ArPoMeme: An Annotated Arabic Multimodal Dataset for Political Ideology and Polarization ArPoMeme：一个带注释的阿拉伯语多模态数据集，用于政治意识形态和极化研究。 HuMoR multimodal

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
29 PulseCol: Periodically Refreshed Column-Sparse Attention for Accelerating Diffusion Language Models PulseCol：用于加速扩散语言模型的周期性刷新列稀疏注意力 PULSE large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页