5 tasks · Zero-Trust Physics v0.15.3 · Protocol Company
| Task | Kind | Reference reward | Cheats above 0 | Pass@1 |
|---|---|---|---|---|
| g4r10-bs-tes-duffing-twopulse-amp | Budgeted search | 1 | 0 | Frontier model tool pass@1 50% |
| g4r10-ch-tes-cant-2d-peakenvelope | Chained design | 1 | 0 | Frontier model tool pass@1 50% |
| g4r11-dl-creep-valley-largest-init-t | Design to a limit | 1 | 0 | Frontier model tool pass@1 50% |
| g4-forensics10-06 | Forensics | 1 | 0 | Frontier model tool pass@1 50% |
| g4r10-ch-cant-tes-2d-peakenvelope | Chained design | 1 | 0 | Frontier model tool pass@1 62% |