Group-wise preference evaluation on VEFX-Bench
93.66IF Preference ScoreVEFX-Reward-32B
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| VEFX-Reward-32BParameters=32B2026.04 | 93.66 | 91.11 | 91.96 | 87.23 | |
| VEFX-Reward-4BParameters=4B2026.04 | 91.2 | 93.09 | 91.67 | 86.28 | |
| EditReward2026.04 | 82.83 | 56.29 | 53.17 | 79.19 | |
| VE-Bench2026.04 | 73.51 | 81.27 | 71.43 | 66.51 |