Visual Grounding on V* Relative Position 52
89.47AccuracySpecEyes
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SpecEyesBase Model=DeepEyes, Aggregation Strategy=min2026.03 | 89.47 | 1.9 | |
| SpecEyesBase Model=DeepEyes, Aggregation Strategy=log2026.03 | 88.16 | 2.05 | |
| SpecEyesBase Model=DeepEyes, Aggregation Strategy=mean2026.03 | 84.21 | 3.35 | |
| SpecEyesBase Model=DeepEyes, Aggregation Strategy=bottom2026.03 | 84.21 | 2.12 | |
| Qwen3-VL-2BMode=draft only2026.03 | 82.89 | 5.31 | |
| DeepEyesBase Model=DeepEyes2026.03 | 82.89 | 1 | |
| ThymeBase Model=Thyme2026.03 | 82.89 | 1 | |
| SpecEyesBase Model=Thyme, Aggregation Strategy=log2026.03 | 82.89 | 1.45 | |
| SpecEyesBase Model=Thyme, Aggregation Strategy=min2026.03 | 82.89 | 1.42 | |
| SpecEyesBase Model=Thyme, Aggregation Strategy=mean2026.03 | 80.26 | 1.83 | |
| SpecEyesBase Model=Thyme, Aggregation Strategy=bottom2026.03 | 80.26 | 1.84 | |
| SpecReasonBase Model=Thyme2026.03 | 75 | 0.53 | |
| SpecReasonBase Model=DeepEyes2026.03 | 73.91 | 0.38 |