Software Engineering Skill Evaluation on SWE-Skills-Bench (turborepo)
30Pass RateS2L
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| S2LSetting=Our method2026.06 | 30 | -12.72 | 0.79 | |
| Vanilla LLMSetting=Vanilla baseline2026.06 | 20 | — | — | |
| Full Skill TextSetting=Full Skill Text prompting2026.06 | 10 | 58.66 | -0.17 |