Multimodal Understanding and Reasoning on MMMU
59AccuracyPTA-GRPO
Evaluation results
| Method | Links | |
|---|---|---|
| PTA-GRPOBase Model=Qwen2.5-7B-VL2025.10 | 59 | |
| SRPOBase Model=Qwen2.5-7B-VL2025.10 | 57.1 | |
| GPT-4V2024.06 | 56.8 | |
| BaseBase Model=Qwen2.5-7B-VL2025.10 | 54.3 | |
| LLaVA-1.5-LLaMA3-8BBase Model=LLaMA3, Model size=8B2024.06 | 45.2 | |
| LLaVA-1.5-13BModel size=13B2024.06 | 36.4 | |
| LLaVA-1.5-7BModel size=7B2024.06 | 33.6 |