Visual Reasoning on OCR-VQA Text-Rich (test)
83Union ScoreSFT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SFTTraining Strategy=Supervised Post-Training2026.05 | 83 | 78 | |
| LLaVA-OV-7BTraining Strategy=Previous Methods2026.05 | 81 | 67 | |
| GRPO+OursTraining Strategy=Reinforcement Post-Training2026.05 | 79 | 75 | |
| DAPO+OursTraining Strategy=Reinforcement Post-Training2026.05 | 78 | 75 | |
| R1-Onevision-7BTraining Strategy=Previous Methods2026.05 | 77 | 58 | |
| Chain-of-ThoughtTraining Strategy=Supervised Post-Training2026.05 | 77 | 76 | |
| GRPOTraining Strategy=Reinforcement Post-Training2026.05 | 77 | 71 | |
| DAPOTraining Strategy=Reinforcement Post-Training2026.05 | 75 | 70 | |
| Qwen2.5-VL-3BTraining Strategy=Previous Methods2026.05 | 69 | 51 | |
| VisionReasoner-7BTraining Strategy=Previous Methods2026.05 | 62 | 45 |