General Understanding on MMMU
56.8Accuracy (MMMU)TemplateRL
Evaluation Results
| Method | Links | |
|---|---|---|
| TemplateRLBackbone=Qwen2.5-VL-3B-Instruct, Batch size=128, Samples per question=16, Template guidance quantity=22025.05 | 56.8 | |
| GRPOBackbone=Qwen2.5-VL-3B-Instruct, Batch size=128, Samples per question=162025.05 | 52.7 | |
| CoTBackbone=Qwen2.5-VL-3B-Instruct, Batch size=128, Samples per question=162025.05 | 51.2 |