Intervention accuracy prediction on CELLO
94.3COIBridgeVLM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| BridgeVLMFine-tuning dataset=CELLO, Zero-shot status=false2026.06 | 94.3 | 80 | 96.6 | 90.3 | |
| PHI-4Fine-tuning dataset=CELLO, Zero-shot status=false2026.06 | 80 | 76 | 95.2 | 83.7 | |
| PHI-4Fine-tuning dataset=None, Zero-shot status=true2026.06 | 60 | 60 | 78.5 | 66.2 | |
| BridgeVLMFine-tuning dataset=CVLB, Zero-shot status=true2026.06 | 58.6 | 48.6 | 82.2 | 63.1 |