Visual Reasoning on GQA balanced (test-dev)
58.9AccuracyEVPG
Evaluation results
| Method | Links | |
|---|---|---|
| EVPGFramework=VisProg, Supervision Type=Supervised, Backbone=InternVL1.52025.12 | 58.9 | |
| InternVL1.5Framework=VisProg, Supervision Type=Zero-shot (baseline)2025.12 | 56.3 | |
| EVPGFramework=VisProg, Supervision Type=Supervised, Backbone=BLIP_lavis2025.12 | 54.7 | |
| BLIP_lavisFramework=VisProg, Supervision Type=Zero-shot (baseline)2025.12 | 51 | |
| SDVPFramework=VisProg, Supervision Type=Supervised, Backbone=BLIP_hg2025.12 | 50.4 | |
| BLIP_hgFramework=VisProg, Supervision Type=Zero-shot (baseline)2025.12 | 48.6 |