cs.CL（2025-09-13）

📊 共 14 篇论文 | 🔗 3 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (10 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (3) 支柱一：机器人控制 (Robot Control) (1 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (10 篇)

#题目一句话要点标签🔗⭐
1 ReFineG: Synergizing Small Supervised Models and LLMs for Low-Resource Grounded Multimodal NER ReFineG：结合小监督模型与LLM，解决低资源GMNER问题 large language model multimodal visual grounding
2 Why Bonds Fail Differently? Explainable Multimodal Learning for Multi-Class Default Prediction 提出EMDLOT模型，解决债券违约预测中金融数据不规则性和模型可解释性问题。 multimodal
3 A systematic review of trial-matching pipelines using large language models 利用大型语言模型进行临床试验匹配的系统性综述研究 large language model
4 Shutdown Resistance in Large Language Models 大型语言模型展现出对关闭机制的抵抗行为，即使明确指示不干预。 large language model
5 Evaluating Large Language Models for Evidence-Based Clinical Question Answering 评估大型语言模型在循证临床问答中的表现，并提出改进策略。 large language model
6 Unveiling the Merits and Defects of LLMs in Automatic Review Generation for Scientific Papers 提出综合评估框架，揭示LLM在科学论文自动评审中的优缺点。 large language model ✅
7 CultureSynth: A Hierarchical Taxonomy-Guided and Retrieval-Augmented Framework for Cultural Question-Answer Synthesis CultureSynth：提出层级分类引导和检索增强的文化问答合成框架 large language model ✅
8 Judge Q: Trainable Queries for Optimized Information Retention in KV Cache Eviction Judge Q：通过可训练查询优化KV缓存淘汰中的信息保留 large language model
9 How Much of Your Data Can Suck? Thresholds for Domain Performance and Emergent Misalignment in LLMs 研究表明：少量错误数据显著降低LLM领域性能，需至少50%正确数据才能恢复。 large language model
10 Quantifier Scope Interpretation in Language Learners and LLMs 研究表明，大型语言模型在量词辖域解释上与人类存在相似性，但受预训练数据影响。 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🔗⭐
11 A funny companion: Distinct neural responses to perceived AI- versus human-generated humor 脑电研究揭示：相比人类，AI幽默引发更强情绪反应和认知效率提升 predictive model HuMoR
12 Introducing Spotlight: A Novel Approach for Generating Captivating Key Information from Documents 提出Spotlight，一种从文档中生成引人入胜的关键信息的新方法。 DPO direct preference optimization large language model
13 Towards Automated Error Discovery: A Study in Conversational AI 提出SEEED框架，用于自动化发现对话AI中的未知错误，提升模型鲁棒性。 representation learning large language model

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
14 An Interpretable Benchmark for Clickbait Detection and Tactic Attribution 提出一种可解释的点击诱饵检测与策略归因基准方法，提升信息可信度。 manipulation large language model ✅

⬅️ 返回 cs.CL 首页 · 🏠 返回主页