Create a checkpoints
Folder in the Repository if not exist.
We provide checkpoints for the following models:
- DDIM: FFHQ128 (72M, 130M), Bedroom128, Horse128
- DiffAE (autoencoding only): FFHQ256, FFHQ128 (72M, 130M), Bedroom128, Horse128
- DiffAE (with latent DPM, can sample): FFHQ256, FFHQ128, Bedroom128, Horse128
- DiffAE's classifiers (for manipulation): FFHQ256's latent on CelebAHQ, FFHQ128's latent on CelebAHQ
Checkpoints ought to be put into a separate directory checkpoints
.
Download the checkpoints and put them into checkpoints
directory. It should look like this:
You have to make sure that your google drive folder is public , this command will install gdown and download the checkpoints to your repo :
sh download-models.sh "https://drive.google.com/drive/folders/1rNCIHv48y2l0w_ERhkjhahIohkb0S8BE?usp=sharing"
this is the path for the checkpoints in google drive :
kk-digital-dataset/diffae-checkpoints/pretrained/
checkpoints folder should look like this :
checkpoints/
- bedroom128_autoenc
- last.ckpt # diffae checkpoint
- latent.ckpt # predicted z_sem on the dataset
- bedroom128_autoenc_latent
- last.ckpt # diffae + latent DPM checkpoint
- bedroom128_ddpm
- ...
First, install MegaCMD:
!pip install megatools -y
Then, login to your Mega account:
!megareg --username "your_email" --password "your_password"
Zip the Folder Before Uploading
!zip -r checkpoints.zip /content/*.pth
Then upload the ZIP to Mega:
!megaput checkpoints.zip --path /Mega/Checkpoints/
Once uploaded, generate a share link:
!megals -e /Mega/Checkpoints/checkpoints.zip