[ Vision Transformer ViT, TimeSformer] PyTorch Image Models. Evolution of Vision and Video. TIMM,CNN