SDPO (on-policy)
| Method | Links | ||||
|---|---|---|---|---|---|
2026.01 | 56.1 | 58.3 | - | - | |
2026.04 | 55.8 | 68.3 | - | 72.8 | |
2026.01 | 54.2 | 63.8 | - | - | |
2026.04 | 54 | 54 | - | 54 | |
2026.04 | 53.8 | 58.6 | - | 65.8 | |
2026.04 | 52.1 | 58.5 | - | 58.5 | |
2026.01 | 51.5 | 52.9 | - | - | |
2026.01 | 47.8 | 62 | - | - | |
2026.04 | 46.9 | 68.1 | - | 70.6 | |
2026.01 | 44.4 | 44.4 | - | - | |
2026.04 | 39.1 | 51.6 | - | 55.5 | |
2026.01 | 34.3 | 51.8 | - | - | |
2026.04 | 30.8 | - | - | - | |
2026.04 | 30.5 | - | - | - | |
2026.01 | 27.9 | - | - | - | |
2026.01 | 18.1 | - | - | - | |
2026.01 | - | - | 27.9 | - | |
2026.01 | - | - | 34.3 | - | |
2026.01 | - | - | 51.8 | - | |
2026.01 | - | - | 30.3 | - | |
2026.01 | - | - | 49.4 | - | |
2026.01 | - | - | 53.1 | - | |
2026.01 | - | - | 53.1 | - | |
2026.01 | - | - | 18.1 | - | |
2026.01 | - | - | 54.2 | - | |
2026.01 | - | - | 63.8 | - | |
2026.01 | - | - | 54.2 | - | |
2026.01 | - | - | 63.8 | - | |
2026.01 | - | - | 56.1 | - | |
2026.01 | - | - | 58.3 | - |