First Incorrect Step Identification on MMMU
16.4FISI F1 ScoreVisualPRM-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| VisualPRM-8BModel=VisualPRM-8B2025.11 | 16.4 | |
| TIM-PRM-8BModel=TIM-PRM-8B2025.11 | 13.4 | |
| TIM-PRM-2BModel=TIM-PRM-2B2025.11 | 12.7 | |
| Qwen3-VL-8BModel=Qwen3-VL-8B2025.11 | 10.6 | |
| MM-PRM-8BModel=MM-PRM-8B2025.11 | 6.4 | |
| Qwen3-VL-2BModel=Qwen3-VL-2B2025.11 | 4.6 |