Tool-integrated Reinforcement Learning for Repo Deep Search

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ma, Zexiong, Peng, Chao, Zeng, Qunhong, Gao, Pengfei, Zou, Yanzhen, Xie, Bing
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916883004915712
author Ma, Zexiong
Peng, Chao
Zeng, Qunhong
Gao, Pengfei
Zou, Yanzhen
Xie, Bing
author_facet Ma, Zexiong
Peng, Chao
Zeng, Qunhong
Gao, Pengfei
Zou, Yanzhen
Xie, Bing
contents Issue localization, the process of identifying code locations that need modification to resolve software issues, is a critical yet challenging task in software development. The semantic gap between natural language issue descriptions and faulty code requires complex multi-hop reasoning through code dependencies. Existing LLM-based agents attempt to address this by integrating repository retrieval tools. However, this transforms issue localization into a demanding task we call Repo Deep Search, which requires the LLM to effectively utilize various repository retrieval tools throughout a multi-step reasoning and navigation process. To tackle this challenge, we present ToolTrain, a two-stage tool-integrated training framework combining rejection-sampled supervised fine-tuning and tool-integrated reinforcement learning to enhance LLMs' ability to use retrieval tools for issue localization. Experimental results show that ToolTrain-trained models achieve state-of-the-art performance, with our 32B model even surpassing Claude-3.7 on function-level localization. The results also show that improved localization performance translates to better end-to-end issue resolution performance. This further demonstrates that training for issue localization is a viable and effective strategy for improving automated software development.
format Preprint
id arxiv_https___arxiv_org_abs_2508_03012
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Tool-integrated Reinforcement Learning for Repo Deep Search
Ma, Zexiong
Peng, Chao
Zeng, Qunhong
Gao, Pengfei
Zou, Yanzhen
Xie, Bing
Software Engineering
Artificial Intelligence
Issue localization, the process of identifying code locations that need modification to resolve software issues, is a critical yet challenging task in software development. The semantic gap between natural language issue descriptions and faulty code requires complex multi-hop reasoning through code dependencies. Existing LLM-based agents attempt to address this by integrating repository retrieval tools. However, this transforms issue localization into a demanding task we call Repo Deep Search, which requires the LLM to effectively utilize various repository retrieval tools throughout a multi-step reasoning and navigation process. To tackle this challenge, we present ToolTrain, a two-stage tool-integrated training framework combining rejection-sampled supervised fine-tuning and tool-integrated reinforcement learning to enhance LLMs' ability to use retrieval tools for issue localization. Experimental results show that ToolTrain-trained models achieve state-of-the-art performance, with our 32B model even surpassing Claude-3.7 on function-level localization. The results also show that improved localization performance translates to better end-to-end issue resolution performance. This further demonstrates that training for issue localization is a viable and effective strategy for improving automated software development.
title Tool-integrated Reinforcement Learning for Repo Deep Search
topic Software Engineering
Artificial Intelligence
url https://arxiv.org/abs/2508.03012