Cross-view Temporal Reasoning on EgoExo-Con G-ExoEgo
18.7AccuracyQwen2.5-VL-7B + View-GRPO
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen2.5-VL-7B + View-GRPOBackbone=Qwen2.5-VL-7B, Training Protocol=View-GRPO2025.10 | 18.7 | |
| Qwen2.5-VL-3B + View-GRPOBackbone=Qwen2.5-VL-3B, Training Protocol=View-GRPO2025.10 | 14.9 | |
| Qwen2.5-VL-7B + SFTBackbone=Qwen2.5-VL-7B, Training Protocol=SFT2025.10 | 14.9 | |
| Qwen2.5-VL-7B + GRPOBackbone=Qwen2.5-VL-7B, Training Protocol=GRPO2025.10 | 14.3 | |
| Qwen2.5-VL-3B + SFTBackbone=Qwen2.5-VL-3B, Training Protocol=SFT2025.10 | 12.9 | |
| InternVL3.5-8B + View-GRPOBackbone=InternVL3.5-8B, Training Protocol=View-GRPO2025.10 | 10.6 | |
| Qwen2.5-VL-3B + GRPOBackbone=Qwen2.5-VL-3B, Training Protocol=GRPO2025.10 | 10.3 | |
| Qwen2.5-VL-3BBackbone=Qwen2.5-VL-3B, Training Protocol=Base2025.10 | 7.9 | |
| Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B, Training Protocol=Base2025.10 | 6.9 | |
| InternVL3.5-8BBackbone=InternVL3.5-8B, Training Protocol=Base2025.10 | 3 |