Spaces:
Running
on
L40S
access request by duplicating the space
hello
can an admin allow me to access to duplicate this space ?
Cannot access gated repo for url https://huggingface.co/stabilityai/stable-virtual-camera/resolve/main/model.safetensors.
Access to model stabilityai/stable-virtual-camera is restricted and you are not in the authorized list. Visit https://huggingface.co/stabilityai/stable-virtual-camera to ask for access.
thanks
You have to set a HF_TOKEN as a secret in the space settings. Make sure you clicked agree on the model https://huggingface.co/stabilityai/stable-virtual-camera
I need an authorization - I can give you my HF key in private DM - but not working like this
impossible to to deploy - see my image
The authorization process is automated. Have you clicked agree on the model page https://huggingface.co/stabilityai/stable-virtual-camera ? It's been set to auto-approve for your account once you clicked agree.
Then, create a huggingface token with read access and set that as "HF_TOKEN" secret in the space settings. That should then allow you to download the model. Let me know if you still face any issues.
deployment worked - thanks - but...50 steps.... and stuck...
what is wrong ?
and also do you have a doc explaining the different parameters ? like cfg ? scale etc.. ?
it's running torch.compile for faster inference, so there will be some delay before it starts sampling for the first time. Subsequent inferences won't have this delay.
You can disable torch.compile by setting COMPILE=False in demo_gr.py, or modify requirements.txt to use a non-nightly stable pytorch version (2.6.0 or above would be best).
There's a doc on the gradio demo at https://github.com/Stability-AI/stable-virtual-camera/blob/main/docs/GR_USAGE.mdcfg means classifier free guidance. See section 4.5 of the paper (Guidance scale on generation uncertainty) for more details https://arxiv.org/pdf/2503.14489