研究人员开发了一种新颖的方法，使LLM智能体能够定位用于代码变更的文件，从线性探索转向领域范围的并行策略。该方法在SWE Bench Pro基准测试中使用Ansible进行了测试，显示出性能的提高，其中一个Haiku类模型在其同类模型中取得了最高的微F1分数，并优于其他基线。研究还发现，文档演变仍然是一个挑战，并且朴素的文件系统访问会负面影响定位准确性。 AI
影响 这项研究可能带来更高效的AI驱动的软件开发工具，从而改进代码定位和问题解决。
排序理由 该集群包含一篇详细介绍LLM智能体新方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2606.11976v1 Announce Type: cross Abstract: Software engineering tools increasingly rely on LLM based agents to localize files to change to resolve a software issue. Most AI agents explore repositories linearly, that is, visiting one directory or file per step. We postulate…
Software engineering tools increasingly rely on LLM based agents to localize files to change to resolve a software issue. Most AI agents explore repositories linearly, that is, visiting one directory or file per step. We postulate that this is a structural mismatch for changes th…
Software engineering tools increasingly rely on LLM based agents to localize files to change to resolve a software issue. Most AI agents explore repositories linearly, that is, visiting one directory or file per step. We postulate that this is a structural mismatch for changes th…