Software Engineering Skill Evaluation on SWE-Skills-Bench (Aggregate)
65Pass RateS2L
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| S2LSetting=Our method2026.06 | 65 | -4.89 | 0.58 | |
| Vanilla LLMSetting=Vanilla baseline2026.06 | 59 | — | — | |
| Full Skill TextSetting=Full Skill Text prompting2026.06 | 54 | 13.39 | -0.18 |