General MLLM Evaluation on Average V* HR-Bench POPE
84.26AccuracySpecEyes
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SpecEyesBase Model=DeepEyes, Aggregation Strategy=min2026.03 | 84.26 | 1.73 | |
| SpecEyesBase Model=Thyme, Aggregation Strategy=min2026.03 | 83.99 | 1.42 | |
| SpecReasonBase Model=Thyme2026.03 | 83.78 | 0.48 | |
| SpecEyesBase Model=DeepEyes, Aggregation Strategy=bottom2026.03 | 82.34 | 1.82 | |
| SpecEyesBase Model=DeepEyes, Aggregation Strategy=log2026.03 | 82.31 | 1.8 | |
| ThymeBase Model=Thyme2026.03 | 82.29 | 1 | |
| SpecEyesBase Model=Thyme, Aggregation Strategy=log2026.03 | 82.09 | 1.49 | |
| SpecEyesBase Model=Thyme, Aggregation Strategy=bottom2026.03 | 81.95 | 1.63 | |
| DeepEyesBase Model=DeepEyes2026.03 | 81.39 | 1 | |
| SpecEyesBase Model=DeepEyes, Aggregation Strategy=mean2026.03 | 80.53 | 2.31 | |
| SpecEyesBase Model=Thyme, Aggregation Strategy=mean2026.03 | 80.53 | 1.7 | |
| Qwen3-VL-2BMode=draft only2026.03 | 78.93 | 4.13 | |
| SpecReasonBase Model=DeepEyes2026.03 | 66.85 | 0.43 |