Video Reward Assessment on VideoGen-Reward Bench
59.68VQ Accuracy (w/ Ties)VideoReward*
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| VideoReward*SFT=true, RM=true, note=trained on internal dataset2026.03 | 59.68 | 75.66 | 66.03 | 74.7 | 53.8 | 72.2 | 61.26 | 73.59 | |
| VideoReward*SFT=true, RM=false, note=trained on internal dataset2026.03 | 54.23 | 73.61 | 61.16 | 65.56 | 52.6 | 70.95 | 58.39 | 70.16 | |
| CriticAgentSFT=true, RM=true2026.03 | 49.79 | 63.94 | 59.98 | 62.42 | 47.98 | 66.52 | 57.31 | 68.86 | |
| LiFT-CriticSFT=true, RM=false2026.03 | 47.53 | 55.97 | 59.04 | 54.91 | 33.79 | 55.43 | 39.08 | 57.26 | |
| VisionRewardSFT=true, RM=true2026.03 | 47.43 | 59.03 | 59.03 | 60.98 | 46.56 | 61.15 | 56.77 | 67.59 | |
| VideoScoreSFT=true, RM=false2026.03 | 47.41 | 47.72 | 59.05 | 51.09 | 37.24 | 50.34 | 41.8 | 50.22 | |
| CriticAgentSFT=true, RM=false2026.03 | 47.16 | 59.27 | 59.08 | 61.2 | 44.71 | 60.39 | 53.2 | 63.16 | |
| UnifiedRewardSFT=true, RM=true2026.03 | 41.27 | 39.42 | — | — | 40.11 | 36.58 | 53.31 | 58.83 | |
| VideoScore2SFT=true, RM=false2026.03 | 34.67 | 65.87 | — | — | 48.7 | 65.92 | 54.53 | 65.59 |