Visual Reasoning on nuScenes Multi-View (test)
92UnionDAPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DAPOTraining Strategy=Reinforcement Post-Training2026.05 | 92 | 74 | |
| R1-Onevision-7BTraining Strategy=Previous Methods2026.05 | 90 | 81 | |
| SFTTraining Strategy=Supervised Post-Training2026.05 | 90 | 78 | |
| Chain-of-ThoughtTraining Strategy=Supervised Post-Training2026.05 | 90 | 75 | |
| GRPO+OursTraining Strategy=Reinforcement Post-Training2026.05 | 90 | 75 | |
| LLaVA-OV-7BTraining Strategy=Previous Methods2026.05 | 88 | 44 | |
| GRPOTraining Strategy=Reinforcement Post-Training2026.05 | 88 | 72 | |
| DAPO+OursTraining Strategy=Reinforcement Post-Training2026.05 | 88 | 81 | |
| VisionReasoner-7BTraining Strategy=Previous Methods2026.05 | 85 | 70 | |
| Qwen2.5-VL-3BTraining Strategy=Previous Methods2026.05 | 82 | 74 |