Multimodal Reward Modeling on UniReward In-Domain (test)
99.3Quality ScoreUniRM
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| UniRMForward=12026.01 | 99.3 | 86 | 90.3 | 83.8 | 84 | 88.68 | 6.49 | |
| SW-RM-V2-LLaMA3.1-8BForward=52026.01 | 60.75 | 43.1 | 46.77 | 45.76 | 53.58 | 49.99 | 7.15 | |
| SW-VL-RM-7BForward=52026.01 | 56.07 | 41.19 | 41.24 | 69.73 | 58.27 | 53.3 | 12.19 | |
| SW-RM-V2-Qwen0.6BForward=52026.01 | 54.91 | 44.05 | 47.24 | 48.91 | 46.17 | 48.26 | 4.12 | |
| GuardRankForward=52026.01 | 33.18 | 85.95 | 91.01 | 84.99 | 80.99 | 75.22 | 23.77 |