Visual Question Answering on SugarCrepe
82.14Simple AccuracyDINOv2
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DINOv2temperature=0.02, number of seeds=52026.02 | 82.14 | — | — | |
| mFRESAtemperature=0.02, number of seeds=52026.02 | 77.18 | — | — | |
| DINOSAURv2temperature=0.02, number of seeds=52026.02 | 76.2 | — | — | |
| StableLSDtemperature=0.02, number of seeds=52026.02 | 72.89 | — | — | |
| SPOTtemperature=0.02, number of seeds=52026.02 | 71.65 | — | — | |
| FT-DINOSAURtemperature=0.02, number of seeds=52026.02 | 71.25 | — | — | |
| Slot Diffusiontemperature=0.02, number of seeds=52026.02 | 70.23 | — | — | |
| LLaVA-1.5 13BIntervention Strategy=Q4 system redistribution (prop)2026.01 | 68.96 | 38.46 | 76.54 | |
| LLaVA-1.5 13BIntervention Strategy=No-intervention baseline2026.01 | 68.35 | 37.25 | 77.36 |