evaluations

Effect of augmentations and head pretraining on EfficientNetB0 (default HEAD, LR0.02, ReduceLROnPlato X=0.2 patience=3

Effect of augmentations and head pretraining on ResNet18 (default HEAD, LR0.02, ReduceLROnPlato X=0.2 patience=3)

Effect of augmentations and head pretraining for ResNet18 and EfficientNetB0 (default HEAD, LR0.02, ReduceLROnPlato X=0.2 patience=3)

MAP evaluation of overfit model with train loss 0.85 on train set

MAP evaluation of model with val loss 2.737, train loss 1.4 on train+val+test sets