Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/home/bart/git/streetscapes/.venv/lib/python3.12/site-packages/transformers/models/auto/image_processing_auto.py", line 514, in from_pretrained raise initial_exception File "/home/bart/git/streetscapes/.venv/lib/python3.12/site-packages/transformers/models/auto/image_processing_auto.py", line 501, in from_pretrained config_dict, _ = ImageProcessingMixin.get_image_processor_dict( ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/bart/git/streetscapes/.venv/lib/python3.12/site-packages/transformers/image_processing_base.py", line 282, in get_image_processor_dict resolved_processor_file = cached_file( ^^^^^^^^^^^^ File "/home/bart/git/streetscapes/.venv/lib/python3.12/site-packages/transformers/utils/hub.py", line 277, in cached_file file = cached_files(path_or_repo_id=path_or_repo_id, filenames=[filename], **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/bart/git/streetscapes/.venv/lib/python3.12/site-packages/transformers/utils/hub.py", line 469, in cached_files raise OSError(f"{e}") from e OSError: Repo id must be in the form 'repo_name' or 'namespace/repo_name': 'https://huggingface.co/jinfengxie/BFMS_1014/raw/main/config.json'. Use repo_type argument if needed.

Fix Plan

The issue arises from the change in how transformers handles URLs for loading models and configurations. To fix this, you need to specify the repository ID and the filename separately.

Step-by-Step Solution:

1. Specify Repository ID and Filename: Instead of passing the full URL to from_pretrained, you should pass the repository ID and specify the filename using the subfolder argument if necessary.
2. Use repo_id and revision: If your model or config is not in the main branch, ensure you specify the correct revision.

Example Code:

from transformers import AutoImageProcessor

# Define the repository ID and filename
repo_id = "jinfengxie/BFMS_1014"
filename = "config.json"

# Load the image processor
image_processor = AutoImageProcessor.from_pretrained(
    repo_id,
    subfolder="raw/main",  # Specify the subfolder if the file is not in the root
    filename=filename,
    revision="main"  # Specify the revision if not using the default branch
)

Verification

After applying the fix, verify that the AutoImageProcessor can be instantiated successfully without raising any errors. You can do this by running the modified code snippet above.

Extra Tips

• Always ensure you're using the correct repository ID and filename when loading models or configurations from the Hugging Face hub.
• If you're working with specific revisions or branches, make sure to specify the revision argument accordingly.
• Keep your transformers library up to date to leverage the latest features and fixes.