Latent-GRPO
| Method | Links | ||
|---|---|---|---|
2026.01 | 78.51 | 811.51 | |
2026.01 | 77.53 | 1,081.47 | |
2026.01 | 77.44 | 2,357.31 | |
2026.01 | 65.77 | 1,608.34 | |
2026.01 | 62.63 | 1,084.72 | |
2026.01 | 58.47 | 718.63 | |
2026.01 | 55.63 | 723.12 | |
2026.01 | 52.94 | 1,224.15 | |
2026.01 | 42.14 | 814.22 | |
2026.02 | 36.2 | 98,537 | |
2026.02 | 34.2 | 126,045 | |
2026.02 | 32.8 | - | |
2026.02 | 32.8 | 113,358 |