IMAGECODE Task on IMAGECODE (val)
38.5Accuracy (All)ContextBLIP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ContextBLIPParams=240M, Protocol=Fine-tune2024.05 | 38.5 | 28.7 | 81.2 | |
| BLIPParams=223M, Protocol=Fine-tune2024.05 | 36.2 | 26.3 | 79.3 | |
| CLIPParams=473M, Protocol=Fine-tune2024.05 | 30.6 | 22.3 | 67 | |
| ContextBLIPParams=240M, Protocol=Zero-shot2024.05 | 30.6 | 19.9 | 77 | |
| BLIPParams=223M, Protocol=Zero-shot2024.05 | 28.4 | 17 | 77.9 | |
| ALBEFProtocol=Zero-shot2024.05 | 28.2 | 17 | 77 | |
| OFAProtocol=Fine-tune2024.05 | 27.2 | 21 | 52.1 | |
| UNITERProtocol=Fine-tune2024.05 | 26 | 19.9 | 52.8 | |
| ViLBERTProtocol=Fine-tune2024.05 | 25.1 | 19.4 | 49.5 | |
| CLIPParams=473M, Protocol=Zero-shot2024.05 | 21.8 | 14.9 | 51.6 | |
| UNITERProtocol=Zero-shot2024.05 | 19.8 | 13.6 | 42.9 | |
| ViLBERTProtocol=Zero-shot2024.05 | 18.5 | 14 | 37.9 |