CodeLLMPaper 记录代码大语言模型相关论文观点看法 论文解读(持续更新ing) 主题 题目 推荐 api幻觉 Towards Mitigating API Hallucination in Code Generated by LLMs with Hierarchical Dependency Aware ★ 代码embedding-rerank模型 CORNSTACK: HIGH-QUALITY CONTRASTIVE DATA FOR BETTER CODE RETRIEVAL AND RERANKING ★★ plan agent做代码生成 Planning In Natural Language Improves LLM Search For Code Generation ★ 用伪代码来代替cot UNLOCKING REASONING POTENTIAL IN LARGE LANGUAGE MODELS BY SCALING CODE-FORM PLANNING ★ 代码embedding-rerank模型 SWERANK: Software Issue Localization with Code Ranking ★★ SWE agent规划操控接口 SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering ★★ 代码仓库图关系rag OrcaLoca: An LLM Agent Framework for Software Issue Localization ★★★ 代码仓库图关系rag LocAgent: Graph-Guided LLM Agents for Code Localization ★★★★ 模块思想数据构建 MoTCoder: Elevating Large Language Models with Module-of-Thought ★ 图关系代码大模型 CODE GRAPH MODEL (CGM):A GRAPH-INTEGRATED LARGE LANGUAGE MODEL FOR REPOSITORY-LEVEL SOFTWARE ENGINEERING TASKS ★ swe新评测数据集 SWE-Dev: Evaluating and Training Autonomous Feature-Driven Software Development ★★ swe新评测数据集 SWE-rebench: An Automated Pipeline for Task Collection and Decontaminated Evaluation of Software Engineering Agents ★★ 代码agent搜索优化 SWE-SEARCH: 通过蒙特卡洛树搜索和迭代优化来增强软件智能体 ★★ AI智能体工作能力评测 TheAgentCompany：一个"模拟公司"来评测AI智能体的真实工作能力 ★★ 代码库探索与理解 RepoMaster: 让AI学会"抄"代码，高效解决复杂任务 ★★ 深度研究智能体 Code Researcher: Deep Research Agent for Large Systems Code and Commit History ★★★ 自适应规划多智能体框架 AdaCoder: 一个更聪明、更高效的AI代码生成框架 ★ 多智能体协作学习框架 LessonL: 三个臭皮匠，赛过诸葛亮——揭秘让AI程序员组团学习的"武林秘籍" ★★ 代码效率优化/强化学习 面向正确性与效率的代码生成：两阶段强化学习框架 ★★ 无数据代码生成/元记忆 元记忆驱动的数据无依赖代码生成（M2WF） ★ 推理记忆/智能体自进化 ReasoningBank：面向推理记忆的智能体自进化框架 ★★ 仓库级代码问答评测 SWE-QA：仓库级代码问答基准与Agent式答题框架 ★★ 多语言仓库级补全评测 M2RC-EVAL：多语言仓库级代码补全评测与指令数据集 ★★ 贡献者 Star History