Metasurface inverse design on Out-of-Distribution (test)
92SGStarter-skill baseline
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Starter-skill baselineCondition=Starter initial skill2026.04 | 92 | 100 | 96 | -4.626 | 1.6 | |
| Post-trainCondition=Evolved skills, Agent=Claude Sonnet 4.62026.04 | 90 | 100 | 95 | -2.092 | 1.68 |