This folder contains some trained model checkpoints (autoencoder + discriminator) made using Seraena.
./taesdxl_photos_96
contains checkpoints from the basic taesdxl training notebook trained from scratch for ~300k steps (on photos cropped to 96px resolution)../taesdxl_photos_256
contains checkpoints from the basic taesdxl training notebook finetuned for ~10k steps on photos cropped to 256px resolution (starting from the checkpoint fromtaesdxl_photos_96
).
model.pth
contains the autoencoder weightsseraena.pth
contains the discriminator weights