Spatial Reasoning on BLINK (full)
56AccuracyArgos
Evaluation Results
| Method | Links | |
|---|---|---|
| Argos2025.12 | 56 | |
| Qwen2.5VL 7B2025.12 | 54.4 | |
| Qwen2.5VL 7BPrompting Strategy=CoT2025.12 | 53.5 | |
| Video-R1Training Protocol=SFT2025.12 | 52.7 | |
| Video-R1Training Protocol=RL2025.12 | 49 |