Software Engineering Skill Evaluation on SWE-Skills-Bench (gitlab-ci-patterns)
7Pass RateS2L
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| S2LSetting=Our method2026.06 | 7 | 41.09 | 0.49 | |
| Full Skill TextSetting=Full Skill Text prompting2026.06 | 6 | 12.53 | 0.8 | |
| Vanilla LLMSetting=Vanilla baseline2026.06 | 5 | — | — |