Hidden directed graph traversal on Hidden directed graph (val)
76.5Success RatePyligent
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Pyligent2026.07 | 76.5 | 10.96 | 2.14 | 47.7 | |
| Gold-only SFT2026.07 | 3.8 | 24.21 | 0.47 | 23.2 | |
| Base model + thinking2026.07 | 0.2 | 14.69 | 0 | — | |
| Base model2026.07 | 0 | 29.85 | 0 | — |