Zero-shot Classification on Let It Wag!
50.2Top-1 AccuracyDynamiCS
Evaluation Results
| Method | Links | |
|---|---|---|
| DynamiCSDataset (Data Size)=DataComp-DFN (130M), Samples Seen@Resolution=1.28B@112 + 128M@224, GPU-hours=1632026.04 | 50.2 | |
| DynamiCSDataset (Data Size)=DataComp-DFN (130M), Samples Seen@Resolution=0.64B@112 + 128M@224, GPU-hours=952026.04 | 46.5 | |
| DynamiCSDataset (Data Size)=LAION-400M (298M), Samples Seen@Resolution=2.56B@112 + 128M@224, GPU-hours=2992026.04 | 45.5 | |
| DFN*Dataset (Data Size)=DataComp-DFN (130M), Samples Seen@Resolution=1.28B@112 + 128M@224, GPU-hours=1512026.04 | 42.4 | |
| DynamiCSDataset (Data Size)=LAION-400M (298M), Samples Seen@Resolution=1.28B@112 + 128M@224, GPU-hours=1632026.04 | 42.1 | |
| HQ-CLIPDataset (Data Size)=DataComp-DFN (130M), Samples Seen@Resolution=3.20B@224, GPU-hours=≈ 26752026.04 | 38.2 | |
| CLIPA*Dataset (Data Size)=LAION-400M (298M), Samples Seen@Resolution=2.56B@112 + 128M@224, GPU-hours=2692026.04 | 36.4 | |
| RECLIP*Dataset (Data Size)=LAION-400M (298M), Samples Seen@Resolution=2.56B@112 + 128M@224, GPU-hours=2802026.04 | 36 | |
| DataCompDataset (Data Size)=DataComp (1.28B), Samples Seen@Resolution=1.28B@224, GPU-hours=≈ 10702026.04 | 33.7 |