I downloaded the original sd-v1.4 from Huggingface and did some finetuning on my dataset, and later on modify the checkpoints accordingly using script/modify_checkpoints.py. I then do the inpainting training using main.py. But the images are washed out. I read something about VAE not included but that shouldn't be the case since we are loading the entire SD-v1.4 right?
I downloaded the original sd-v1.4 from Huggingface and did some finetuning on my dataset, and later on modify the checkpoints accordingly using
script/modify_checkpoints.py
. I then do the inpainting training usingmain.py
. But the images are washed out. I read something about VAE not included but that shouldn't be the case since we are loading the entire SD-v1.4 right?