cs.CL（2025-07-16）

📊 共 28 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) 支柱二：RL算法与架构 (RL & Architecture) (3) 支柱一：机器人控制 (Robot Control) (2)

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

#题目一句话要点标签🔗⭐
1 Marco-Bench-MIF: On Multilingual Instruction-Following Capability of Large Language Models 提出 Marco-Bench-MIF，用于评估大语言模型的多语言指令遵循能力。 large language model instruction following ✅
2 Improving Drug Identification in Overdose Death Surveillance using Large Language Models 利用大型语言模型改进药物过量死亡监测中的药物识别 large language model
3 Improving Contextual ASR via Multi-grained Fusion with Large Language Models 提出一种多粒度融合的上下文ASR方法，利用大型语言模型提升关键词识别。 large language model
4 A Comparative Approach to Assessing Linguistic Creativity of Large Language Models and Humans 提出一种评估大型语言模型和人类语言创造力的通用测试方法 large language model
5 Value-Based Large Language Model Agent Simulation for Mutual Evaluation of Trust and Interpersonal Closeness 基于价值的大语言模型智能体模拟，用于互评估信任和人际亲密度 large language model
6 Graph Representations for Reading Comprehension Analysis using Large Language Model and Eye-Tracking Biomarker 利用大语言模型和眼动追踪生物标记，提出基于图表示的阅读理解分析方法。 large language model
7 Tracing Facts or just Copies? A critical investigation of the Competitions of Mechanisms in Large Language Models 探究大语言模型中机制竞争：事实追踪还是简单复制？ large language model
8 Advancing Retrieval-Augmented Generation for Structured Enterprise and Internal Data 提出一种高级RAG框架，用于处理结构化企业内部数据，提升问答性能。 large language model multimodal ✅
9 DyG-RAG: Dynamic Graph Retrieval-Augmented Generation with Event-Centric Reasoning DyG-RAG：提出事件中心动态图检索增强生成框架，解决时序推理难题。 large language model chain-of-thought ✅
10 Findings of MEGA: Maths Explanation with LLMs using the Socratic Method for Active Learning MEGA：结合苏格拉底教学法和LLM的数学解释方法，提升学生学习效果 large language model chain-of-thought
11 PARAM-1 BharatGen 2.9B Model PARAM-1：一个以印度语言多样性为核心的29亿参数语言模型 large language model foundation model
12 A Survey of Deep Learning for Geometry Problem Solving 深度学习赋能几何问题求解：综述与前瞻 large language model multimodal ✅
13 Can We Predict Alignment Before Models Finish Thinking? Towards Monitoring Misaligned Reasoning Models 提出基于CoT激活的线性探针，用于提前预测推理模型对齐状态 large language model
14 Probing for Arithmetic Errors in Language Models 利用语言模型内部激活探测算术错误并指导模型自纠错 chain-of-thought
15 Developing Visual Augmented Q&A System using Scalable Vision Embedding Retrieval & Late Interaction Re-ranker 提出一种可扩展的视觉增强问答系统，利用可扩展的视觉嵌入检索和后期交互重排序器。 multimodal
16 Beyond Single Models: Enhancing LLM Detection of Ambiguity in Requests through Debate 提出多代理辩论框架以增强LLM对请求歧义的检测能力 large language model
17 Chain-of-Descriptions: Improving Code LLMs for VHDL Code Generation and Summarization 提出Chain-of-Descriptions方法，提升代码大模型在VHDL代码生成与摘要任务上的性能 large language model
18 Text-ADBench: Text Anomaly Detection Benchmark based on LLMs Embedding Text-ADBench：基于LLM嵌入的文本异常检测基准，揭示嵌入质量是关键。 large language model ✅
19 Identifying Algorithmic and Domain-Specific Bias in Parliamentary Debate Summarisation 提出多阶段总结框架，评估LLM在议会辩论总结中的算法和领域偏差。 large language model
20 Iterative Augmentation with Summarization Refinement (IASR) Evaluation for Unstructured Survey data Modeling and Analysis 提出IASR框架，用于评估和优化LLM在非结构化调查数据建模中的增广效果。 large language model
21 TopicImpact: Improving Customer Feedback Analysis with Opinion Units for Topic Modeling and Star-Rating Prediction TopicImpact：利用观点单元改进客户反馈分析，提升主题建模和星级预测 large language model
22 Toxicity-Aware Few-Shot Prompting for Low-Resource Singlish Translation 提出一种毒性感知的少样本提示框架，用于低资源Singlish翻译，提升毒性内容翻译质量。 large language model
23 BlockBPE: Parallel BPE Tokenization 提出BlockBPE以解决GPU批量推理中的BPE瓶颈问题 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🔗⭐
24 Modeling Open-World Cognition as On-Demand Synthesis of Probabilistic Models 提出模型合成架构MSA，模拟开放世界认知中概率模型的按需合成。 world model chain-of-thought
25 Simplifications are Absolutists: How Simplified Language Reduces Word Sense Awareness in LLM-Generated Definitions 简化语言降低LLM生成定义中词义辨析能力，DPO微调可显著改善 direct preference optimization large language model
26 DualReward: A Dynamic Reinforcement Learning Framework for Cloze Tests Distractor Generation DualReward：一种用于完形填空题干扰项生成的动态强化学习框架 reinforcement learning

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
27 Evaluating the Ability of Large Language Models to Reason about Cardinal Directions, Revisited 评估大语言模型在基数方向推理能力，发现现有模型仍存在不足 locomotion large language model
28 PoTPTQ: A Two-step Power-of-Two Post-training for LLMs PoTPTQ：一种用于LLM的二步幂次量化后训练方法 manipulation large language model

⬅️ 返回 cs.CL 首页 · 🏠 返回主页