waifu-diffusion
waifu-diffusion copied to clipboard
stable diffusion finetuned on weeb stuff
Added options to: - Disable Inference (it consumes about 2gb of VRAM even when not active) - Disable wandb and: - if no hftoken is provided it just fills it...
Hi there, when typing "pip install -r requirements.txt" it throws an error and won't install. ``` C:\Users\distr\Desktop\ai stuff\waifu-diffusion>pip install -r requirements.txt Collecting diffusers>=0.5.1 (from -r requirements.txt (line 1)) Using cached...
i got this problem when using "diffusers_trainer.py": Exception caught on rank 0 at step 10, saving checkpoint... int() argument must be a string, a bytes-like object or a number, not...
On this line, grad clipping occurs: https://github.com/harubaru/waifu-diffusion/blob/27d301c5b96834536166cc2f12e7a9bb4079fb96/trainer/diffusers_trainer.py#L931 However, if fp16 is enabled then the clipping would be applied to the _scaled_ gradients, due to `GradScaler`. According to PyTorch documentation (https://pytorch.org/docs/master/notes/amp_examples.html#gradient-clipping),...
What's the minimum GPU ram required to run this (diffusers_trainer.py)? Thanks in advance.
It's awfully slow. I am willing to do it if someone gives me directions