Trajectory Verification on SWE-bench Verified Qwen3-Coder-Flash trajectories
62RM@32SWE-RM-30A3B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SWE-RM-30A3BType=EF (execution-free)2025.12 | 62 | 0.783 | 0.051 | |
| DEEP SWEType=EB (execution-based)2025.12 | 54.6 | — | — | |
| DEEP SWEType=EF (execution-free)2025.12 | 53.2 | 0.758 | 0.124 | |
| AGENTLESSType=EB (execution-based)2025.12 | 52.6 | — | — | |
| SWE-GYMType=EF (execution-free)2025.12 | 51.2 | 0.776 | 0.223 |