Hello, can you elucidate how the posted ckpt was trained?
Is it trained with imagenet in latent space, and then linearly tuned with BAPPS?
Or maybe one of the two has been skipped?
Hi. I reproduced it based on VGG, and basically used the imagenet pretrained model
The checkpoint file is only linear tuned with BAPPS using pytorch pretrained VGG
Hello, can you elucidate how the posted ckpt was trained? Is it trained with imagenet in latent space, and then linearly tuned with BAPPS? Or maybe one of the two has been skipped?