FuseSearch: Learning Adaptive Parallel Execution for Efficient Code Localization

ACL ARR 2026 January Submission7795 Authors

Keywords: Code Localization, LLM Agents, Reinforcement Learning, Parallel Tool Execution, Quality-Efficiency Trade-off, Software Engineering

Abstract: Code localization is a primary bottleneck in automated software development. While parallel tool execution can accelerate discovery, existing agents suffer from a 34.9\% redundant tool invocation rate, negating the benefits of parallelism. We introduce FuseSearch, which reframes parallel code localization as a quality--efficiency co-optimization problem. By defining tool efficiency---the ratio of novel information gain to total invocations---we employ a two-stage SFT and RL pipeline to train models in adaptive parallel strategies. Unlike fixed-breadth methods, FuseSearch dynamically adjusts search breadth based on task context, transitioning from exploration to refinement. On SWE-bench Verified, FuseSearch-4B matches SOTA performance (84.7\% file-level and 56.4\% function-level $F_1$ scores) while being 93.6\% faster, using 67.7\% fewer turns and 68.9\% fewer tokens. Our findings demonstrate that efficiency-aware training inherently boosts quality by eliminating noisy, redundant signals, enabling high-performance, low-cost localization agents. Code: \url{https://anonymous.4open.science/r/FuseSearch-2BDD}.

Paper Type: Long

Research Area: Code Models

Research Area Keywords: AI / LLM Agents; Code Models

Contribution Types: NLP engineering experiment, Approaches low compute settings-efficiency

Languages Studied: English

Submission Number: 7795