Pengwei Song & Yifan Qi
The pre-trained transformer based model can be downloaded from Github.
Use the T2T-VIT model:
CUDA_VISIBLE_DEVICES=1,2 python transfer_learning.py --lr 0.05 --b 32 —num-classes 100 --img-size 224 --dataset cifar100 --model t2t_vit_14 --transfer-learning True --transfer-model T2T_ViT_14.pth