BLAgent framework enhances file-level bug localization with agentic RAG

Researchers have developed BLAgent, a novel agentic retrieval-augmented generation (RAG) framework designed to improve file-level bug localization in software maintenance. BLAgent integrates code structure-aware encoding, dual-perspective query transformation, and a two-phase agentic reranking process. This approach balances localization accuracy with computational cost, achieving over 78% Top-1 accuracy on SWE-bench Lite with open-source models and over 86% with a closed-source model, while being significantly more cost-effective than existing methods. When integrated into an automated program repair framework, BLAgent has demonstrated an improvement of up to 25% in end-to-end repair success. AI

IMPACT Enhances efficiency and accuracy in software debugging and automated program repair.

RANK_REASON The cluster contains a research paper detailing a novel framework for bug localization. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• BLAgent
• Hugging Face
• Md Afif Al Mamun
• retrieval-augmented generation
• SWE-bench Lite

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →