List your company

Software Engineering Issue Resolution on Multi-SWE-Bench Mini

20Resolution RateHybrid
9.612.31517.7Dec 26, 2025

Evaluation results

MethodLinks
HybridFeedback=Hybrid (Execution-free + Execution-based)2025.1220
Execution-free onlyFeedback=Execution-free (SWE-RM)2025.1218.8
Execution-based onlyFeedback=Execution-based2025.1218.5
Poor Calibrated RMFeedback=Ablated RM (Poorly Calibrated)2025.1210