cs.CL（2025-03-24）

📊 共 26 篇论文 | 🔗 5 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (23 🔗5) 支柱一：机器人控制 (Robot Control) (2) 支柱二：RL算法与架构 (RL & Architecture) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (23 篇)

#题目一句话要点标签🔗⭐
1 MIRAGE: Multimodal Immersive Reasoning and Guided Exploration for Red-Team Jailbreak Attacks 提出MIRAGE框架，利用多模态沉浸式推理和引导探索实现对MLLM的红队越狱攻击。 large language model multimodal
2 A Survey of Large Language Model Agents for Question Answering 综述：基于大型语言模型Agent的问答系统研究 large language model
3 I Have Covered All the Bases Here: Interpreting Reasoning Features in Large Language Models via Sparse Autoencoders 利用稀疏自编码器解析大型语言模型中的推理特征，揭示其内部推理机制。 large language model ✅
4 Commander-GPT: Fully Unleashing the Sarcasm Detection Capability of Multi-Modal Large Language Models 提出Commander-GPT框架，无需微调即可显著提升多模态大语言模型在讽刺检测任务上的性能。 large language model
5 Self-Reported Confidence of Large Language Models in Gastroenterology: Analysis of Commercial, Open-Source, and Quantized Models 评估大型语言模型在胃肠病学问题上的自信度与准确性 large language model
6 MAGIC-VQA: Multimodal And Grounded Inference with Commonsense Knowledge for Visual Question Answering MAGIC-VQA：融合常识知识的多模态视觉问答框架 multimodal
7 J&H: Evaluating the Robustness of Large Language Models Under Knowledge-Injection Attacks in Legal Domain 提出J&H框架，评估大语言模型在法律领域知识注入攻击下的鲁棒性 large language model
8 Surgical Action Planning with Large Language Models 提出LLM-SAP框架，利用大语言模型进行机器人辅助手术中的动作规划 large language model
9 TIB-STC: A Large-Scale Structured Tibetan Benchmark for Low-Resource Language Modeling 构建大规模结构化藏语基准数据集TIB-STC，促进低资源语言建模 large language model instruction following ✅
10 Overtrained Language Models Are Harder to Fine-Tune 揭示大语言模型过度训练导致微调困难的“灾难性过度训练”现象 large language model
11 Evaluating Bias in LLMs for Job-Resume Matching: Gender, Race, and Education 评估LLM在职位-简历匹配中的偏见：性别、种族与教育 large language model
12 Language Model Uncertainty Quantification with Attention Chain 提出UQAC方法，通过注意力链高效量化LLM在复杂推理中的不确定性 large language model
13 Understanding and Improving Information Preservation in Prompt Compression for LLMs 提出Prompt压缩评估框架并改进软提示方法，提升LLM信息保持能力与下游任务性能。 large language model ✅
14 Masks and Mimicry: Strategic Obfuscation and Impersonation Attacks on Authorship Verification 利用LLM的混淆和模仿攻击评估作者身份验证模型的对抗鲁棒性 large language model
15 LLM-Based Insight Extraction for Contact Center Analytics and Cost-Efficient Deployment 提出基于LLM的联络中心洞察提取系统，实现低成本呼叫分析与智能应用。 large language model
16 LookAhead Tuning: Safer Language Models via Partial Answer Previews LookAhead Tuning：通过预览部分答案前缀，提升微调后语言模型的安全性 large language model
17 Exploring Training and Inference Scaling Laws in Generative Retrieval 探索生成式检索中的训练和推理扩展定律，揭示模型规模、数据规模和计算资源对性能的影响。 large language model
18 xKV: Cross-Layer SVD for KV-Cache Compression xKV：通过跨层奇异值分解压缩KV缓存，提升长文本LLM推理效率。 large language model ✅
19 AgentDropout: Dynamic Agent Elimination for Token-Efficient and High-Performance LLM-Based Multi-Agent Collaboration 提出AgentDropout以解决多智能体系统中的低通信效率问题 large language model ✅
20 LANGALIGN: Enhancing Non-English Language Models via Cross-Lingual Embedding Alignment LANGALIGN：通过跨语言嵌入对齐增强非英语语言模型 large language model
21 Autoregressive Language Models for Knowledge Base Population: A case study in the space mission domain 提出基于自回归语言模型的知识库填充方法，应用于航天任务领域。 large language model
22 Fact-checking AI-generated news reports: Can LLMs catch their own lies? 评估大语言模型对自身生成新闻报告的真伪辨别能力，揭示其局限性。 large language model
23 Enhancing Multi-Label Emotion Analysis and Corresponding Intensities for Ethiopian Languages 增强埃塞俄比亚语多标签情感分析，引入情感强度建模。 large language model

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
24 Predicting the Road Ahead: A Knowledge Graph based Foundation Model for Scene Understanding in Autonomous Driving 提出基于知识图谱的场景理解基础模型FM4SU，用于自动驾驶 motion planning scene understanding foundation model
25 AlphaSpace: Enabling Robotic Actions through Semantic Tokenization and Symbolic Reasoning AlphaSpace：通过语义Token化和符号推理实现机器人动作 manipulation

🔬 支柱二：RL算法与架构 (RL & Architecture) (1 篇)

#题目一句话要点标签🔗⭐
26 Teaching LLMs for Step-Level Automatic Math Correction via Reinforcement Learning 提出StepAMC，利用强化学习提升LLM在步骤级自动数学纠错中的性能 reinforcement learning large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页