Video Generation Assessment on MJBench (test)
72AccuracyUnifiedReward-Flex
Evaluation Results
| Method | Links | |
|---|---|---|
| UnifiedReward-Flex2026.02 | 72 | |
| UnifiedReward-Think2026.02 | 70.9 | |
| UnifiedReward-FlexDPO=False, Protocol=Both correct2026.02 | 70.3 | |
| UnifiedReward-FlexDPO=False2026.02 | 69.1 | |
| UnifiedReward2026.02 | 68.8 | |
| VideoReward2026.02 | 63.4 | |
| VideoScore2026.02 | 62.8 | |
| LiFT2026.02 | 51 |