Visual Question Answering on 3DRad 1.0 (test)
79.28AccuracyMedPruner
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MedPrunerBase Model=Hulu-Med-7B2026.03 | 79.28 | 26.247 | 25.982 | 22.865 | 7.123 | 51.88 | |
| VisionZipBase Model=Hulu-Med-7B2026.03 | 79.232 | 25.618 | 25.336 | 22.391 | 6.949 | 69.74 | |
| HiPruneBase Model=Hulu-Med-7B2026.03 | 79.061 | 25.721 | 25.559 | 21.339 | 6.759 | 22.3 | |
| Original (Hulu-Med-7B)Base Model=Hulu-Med-7B2026.03 | 78.767 | 23.28 | 22.917 | 18.76 | 5.695 | 100 | |
| Hulu-L1Base Model=Hulu-Med-7B2026.03 | 78.082 | 23.765 | 23.487 | 20.389 | 6.28 | 45.06 | |
| MedPrunerBase Model=MedGemma1.5-4B2026.03 | 60.843 | 5.168 | 2.057 | 1.317 | 0.277 | 4.62 | |
| Original (MedGemma1.5-4B)Base Model=MedGemma1.5-4B2026.03 | 59.155 | 4.865 | 4.094 | 2.629 | 0.516 | 100 | |
| Hulu-L1Base Model=MedGemma1.5-4B2026.03 | 57.868 | 5.009 | 1.596 | 1.034 | 0.217 | 44.89 | |
| VisionZipBase Model=MedGemma1.5-4B2026.03 | 56.537 | 5.091 | 1.155 | 0.754 | 0.165 | 69.82 | |
| HiPruneBase Model=MedGemma1.5-4B2026.03 | 55.959 | 6.109 | 2.045 | 1.365 | 0.304 | 22.3 |