Hi May I know why the hugging face is saying we dont have the access but actually we have been approved?

#16
by Tianyinus - opened

Hi, thanks for this amazing work!

Previously we have applied and got approved, in the webpage it says:
Gated model You have been granted access to this model

However, in the python downloading, it says:
Cannot access gated repo for url https://huggingface.co/MahmoodLab/UNI/resolve/main/config.json.
Access to model MahmoodLab/UNI is restricted. You must have access to it and be authenticated to access it. Please log in.

But actually I have loged in with my token.

I heard my collegues also have the same issue with UNI, may I raise this issue for your further notice? Thanks!

and I find manually build the model and then use huggingface weight is ok:
api_token = os.environ["HF_TOKEN"]

Download the model weights

model_weights_path = hf_hub_download(repo_id="MahmoodLab/UNI", filename="pytorch_model.bin", use_auth_token=api_token)

Load the model using timm and the downloaded weights

Create the ViT model using the configuration from Hugging Face

model = timm.create_model(
"vit_large_patch16_224", # Model architecture
pretrained=True, # Load pretrained weights
img_size=224, # Image size is 224x224
num_classes=0, # No classification head (feature extractor mode)
init_values=1.0, # Layer scale initialization value
global_pool="token", # Use token pooling (default for ViT)
dynamic_img_size=True # Allow dynamic image size
)

model.load_state_dict(torch.load(model_weights_path))

Sign up or log in to comment