cs.CL（2025-04-13）

📊 共 22 篇论文 | 🔗 7 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (18 🔗6) 支柱二：RL算法与架构 (RL & Architecture) (4 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (18 篇)

#题目一句话要点标签🔗⭐
1 HM-RAG: Hierarchical Multi-Agent Multimodal Retrieval Augmented Generation 提出HM-RAG，一种层级多智能体多模态检索增强生成框架，用于复杂查询下的知识合成。 large language model multimodal ✅
2 Improving Multilingual Capabilities with Cultural and Local Knowledge in Large Language Models While Enhancing Native Performance Mantra-14B：通过文化和本地知识增强LLM多语言能力并提升原生性能 large language model
3 ClinicalGPT-R1: Pushing reasoning capability of generalist disease diagnosis with large language model ClinicalGPT-R1：利用大型语言模型提升通用疾病诊断的推理能力 large language model ✅
4 Can the capability of Large Language Models be described by human ability? A Meta Study 通过对比LLM与人类能力，探究LLM能力是否能用人类能力指标描述 large language model
5 Read Before You Think: Mitigating LLM Comprehension Failures with Step-by-Step Reading 提出Step-by-Step Reading (SSR)系列提示方法，提升LLM在复杂推理任务中的问题理解能力。 large language model chain-of-thought
6 Short-Path Prompting in LLMs: Analyzing Reasoning Instability and Solutions for Robust Performance 研究短路径提示对LLM推理能力的影响，并提出指令引导和微调方法提升鲁棒性 large language model chain-of-thought
7 Syzygy of Thoughts: Improving LLM CoT with the Minimal Free Resolution 提出Syzygy of Thoughts (SoT)框架，利用极小自由分解提升LLM的CoT推理能力。 large language model chain-of-thought ✅
8 Myanmar XNLI: Building a Dataset and Exploring Low-resource Approaches to Natural Language Inference with Myanmar 构建缅甸语XNLI数据集并探索低资源自然语言推理方法 large language model
9 How new data permeates LLM knowledge and how to dilute it 研究LLM学习新知识的泛化与幻觉现象，并提出数据增强和更新剪枝方法以提升知识特异性。 large language model ✅
10 Meta-Evaluating Local LLMs: Rethinking Performance Metrics for Serious Games 针对严肃游戏，提出元评估方法以评估本地LLM的性能指标 large language model
11 HalluShift: Measuring Distribution Shifts towards Hallucination Detection in LLMs HalluShift：通过测量LLM中的分布偏移来检测幻觉现象 large language model ✅
12 Span-level Emotion-Cause-Category Triplet Extraction with Instruction Tuning LLMs and Data Augmentation 提出基于指令调优LLM和数据增强的Span级情感-原因-类别三元组抽取方法 large language model ✅
13 AgentA/B: Automated and Scalable Web A/BTesting with Interactive LLM Agents AgentA/B：利用交互式LLM Agent实现自动化和可扩展的Web A/B测试 large language model
14 Leveraging Reasoning Model Answers to Enhance Non-Reasoning Model Capability 利用推理模型答案提升非推理模型能力 large language model
15 AdaSteer: Your Aligned LLM is Inherently an Adaptive Jailbreak Defender AdaSteer：提出自适应激活Steering方法，增强LLM的越狱防御能力。 large language model
16 UXAgent: A System for Simulating Usability Testing of Web Design with LLM Agents UXAgent：利用LLM Agent模拟Web设计可用性测试，辅助UX研究。 large language model
17 Measuring LLM Novelty As The Frontier Of Original And High-Quality Output 提出一种新颖性指标，通过平衡原创性和质量来评估LLM的创造性输出能力。 large language model
18 On Language Models' Sensitivity to Suspicious Coincidences 研究发现语言模型在零样本学习中对可疑巧合不敏感，但可通过提示工程增强 chain-of-thought

🔬 支柱二：RL算法与架构 (RL & Architecture) (4 篇)

#题目一句话要点标签🔗⭐
19 Kongzi: A Historical Large Language Model with Fact Enhancement 提出孔子（Kongzi）模型，通过事实增强解决历史领域大语言模型的事实性问题。 reinforcement learning large language model
20 SaRO: Enhancing LLM Safety through Reasoning-based Alignment 提出SaRO框架，通过推理对齐增强LLM的安全性，解决越狱攻击和过度对齐问题。 DPO direct preference optimization large language model
21 GRPO-LEAD: A Difficulty-Aware Reinforcement Learning Approach for Concise Mathematical Reasoning in Language Models 提出GRPO-LEAD以解决数学推理中的奖励稀疏与难度感知问题 reinforcement learning ✅
22 MADLLM: Multivariate Anomaly Detection via Pre-trained LLMs 提出MADLLM，通过预训练LLM进行多元时间序列异常检测 contrastive learning large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页