fire
numpy
retry
tqdm
torch
torchvision
pillow
adamp
contrastive_learner>=0.1.0
linear_attention_transformer
vector-quantize-pytorch
