Image Classification on ImageNet-ReaL 1 (test)
91.12Top-1 AccuracyMPL
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MPLPre-training dataset=unlabelled or weakly labelled data (proprietary), Resolution=448, Evaluation protocol=Fine-tuned2021.05 | 91.12 | — | 20,480 | |
| ViT-H/14Pre-training dataset=JFT-300M (proprietary), Resolution=448, Evaluation protocol=Fine-tuned2021.05 | 90.72 | 15 | 2,300 | |
| BiT-R152x4Pre-training dataset=JFT-300M (proprietary), Resolution=448, Evaluation protocol=Fine-tuned2021.05 | 90.54 | 26 | 9,900 | |
| Mixer-H/14Pre-training dataset=JFT-300M (proprietary), Resolution=448, Evaluation protocol=Fine-tuned2021.05 | 90.18 | 40 | 1,010 | |
| ViT-L/16Pre-training dataset=ImageNet-21k (public), Resolution=448, Evaluation protocol=Fine-tuned2021.05 | 88.62 | 32 | 180 | |
| Mixer-L/16Pre-training dataset=ImageNet-21k (public), Resolution=448, Evaluation protocol=Fine-tuned2021.05 | 87.86 | 105 | 410 |