GatedRepoError: 403 Client Error
I'm sure everyone else on Earth knows how to resolve this but sadly I do not.
I've tried every manner of HF_TOKEN, secret key and use_auth_token=hf_auth I could find and it still won't let me download.
and my huggingface page says
Gated model: You have been granted access to this model
Please, any help would be welcome!
I'm working in google colab, building from source and this is where the code fails:
!huggingface-cli login
from diffusers import StableDiffusion3Pipeline
pipe = StableDiffusion3Pipeline.from_pretrained("stabilityai/stable-diffusion-3-medium-diffusers",
torch_dtype=torch.float16, use_auth_token=hf_auth)
hey did you add the repo permissions?
Settings -> Access Tokens -> Manage -> Edit permissions -> Search for repos
Add "stabilityai/stable-diffusion-3-medium-diffusers" there.
hey did you add the repo permissions?
Settings -> Access Tokens -> Manage -> Edit permissions -> Search for repos
Add "stabilityai/stable-diffusion-3-medium-diffusers" there.
I get to settings->access tokens->manage-> then it gives me 2 options : [Invalidate and Refresh] or [Delete]
Aha!
when creating with fine-grained access
fingers crossed, I don't have time to test it now, maybe after lunch.
Aha! Ahahaha!
Got it loading anyways, was the secret of :
token=hf_auth
in the pipe constructor
uhg! it was loading wonderfully then ... cuda OOM
I saw someone else fix that vram thing here somewhere...
torch.cuda.empty_cache()
to the rescue!