Image Captioners Are Scalable Vision Learners Too

Benchmark Model Rank Results
image-classification-on-flowers-102CapPa L/14 + MAP#20Accuracy: 98.9
image-classification-on-imagenet-v2CapPa L/14 + MAP#26Top 1 Accuracy: 76.8
image-classification-on-imagenet-realCapPa L/14 + MAP#30Accuracy: 89.3%
image-classification-on-objectnetCapPa L/14 + MLP#31Top-1 Accuracy: 56.8
image-classification-on-imagenetCapPa L/14 (336px)#83Top 1 Accuracy: 87.7%