Video Preference Modeling on VideoGen-Reward
71.8Tau (%)VR-THINKER
Evaluation results
| Method | Links | ||
|---|---|---|---|
| VR-THINKERSize=7B, Protocol=Reasoning-based Reward Models2025.10 | 71.8 | 80.5 | |
| UnifiedReward-ThinkSize=7B, Protocol=Reasoning-based Reward Models2025.10 | 69.7 | 79.1 | |
| UnifiedRewardSize=7B, Protocol=Generative-based Reward Models2025.10 | 67.1 | 78.6 | |
| VideoRewardSize=2B, Protocol=Classifier-based Reward Models2025.10 | 60.8 | 73.8 | |
| VisionRewardSize=13B, Protocol=Classifier-based Reward Models2025.10 | 57.9 | 68.4 | |
| VideoScoreSize=7B, Protocol=Classifier-based Reward Models2025.10 | 41.9 | 50.2 | |
| LiFTSize=13B, Protocol=Generative-based Reward Models2025.10 | 40.1 | 57.9 |