Image Generation Assessment on MMRB2 (test)
69.2AccuracyUnifiedReward-Flex
Evaluation Results
| Method | Links | |
|---|---|---|
| UnifiedReward-Flex2026.02 | 69.2 | |
| UnifiedReward-FlexDPO=False, Protocol=Both correct2026.02 | 68.4 | |
| UnifiedReward-FlexDPO=False2026.02 | 67.5 | |
| UnifiedReward-Think2026.02 | 66 | |
| UnifiedReward2026.02 | 60 | |
| HPSv32026.02 | 58.5 | |
| PickScore2026.02 | 57.6 | |
| HPSv22026.02 | 55 |