Real-world Spatial Understanding on RWQA
67.84Top-1 AccuracyBase model
Evaluation results
| Method | Links | |
|---|---|---|
| Base modelTraining Approach=Training-free, VLM=Qwen 2.5 VL2026.04 | 67.84 | |
| Entropy-Gradient GroundingTraining Approach=Training-free, VLM=Qwen 2.5 VL2026.04 | 66.93 | |
| Entropy-Gradient GroundingTraining Approach=Training-free, VLM=InternVL 3.52026.04 | 64.71 | |
| Base modelTraining Approach=Training-free, VLM=InternVL 3.52026.04 | 61.83 | |
| Entropy-Gradient GroundingTraining Approach=Training-free, VLM=LLaVA 1.62026.04 | 60.39 | |
| Base modelTraining Approach=Training-free, VLM=LLaVA 1.62026.04 | 58.3 | |
| ViCropTraining Approach=Training-free, VLM=LLaVA 1.62026.04 | 56.99 | |
| Base modelTraining Approach=Training-free, VLM=LLaVA 1.52026.04 | 48.76 | |
| Entropy-Gradient GroundingTraining Approach=Training-free, VLM=LLaVA 1.52026.04 | 48.24 | |
| ViCropTraining Approach=Training-free, VLM=LLaVA 1.52026.04 | 47.97 |