Repository navigation
Moving Testing Data to HuggingFace #8510
Description
I have copied out testing data from this release into a HuggingFace dataset here. I propose that we change our URLs in the tests to use these locations instead, and in the future we can more easily add items to this dataset as well as avoid abusing the release mechanism Github provides. Second to that is to move data for tutorials to another new dataset and clean up the use of multiple storage locations such as gdrive accounts. Last is to possibly do the same for model zoo items and weights unless we migrate entirely to HuggingFace.
Metadata
Metadata
Assignees
Labels
No labels
Activity
e-mny commentedon Oct 25, 2025
Hi Eric, I would like to contribute to this issue if that is okay with you. It will be my first open-source project contribution.
I had a brief look through at the tests folder, but am not able to find the URLs that you mentioned. Could you kindly point me in the right direction?
ericspod commentedon Dec 4, 2025
Hi @e-mny sorry for the delay, I've been away. Help would be greatly appreciated with this one! I have put all the files in the Huggingface dataset for testing, some are duplicate of what's stored here in the repo but that's fine. The others which are downloaded from the release in
MONAI-extra-test-datashould be there now, so I think all that needs changing is data_config.json to point to the correct new files in Huggingface.You can get the URL to a file in Huggingface to work with download_url so you don't need to use
huggingface_hub. For example I can download one of the files through the URL "https://huggingface.co/datasets/MONAI/testing_data/resolve/main/CMU-1.tiff" usingdownload_urlor evencurlon the command line.If you want to help with this issue I would suggest going through the
tests/testing_data/data_config.jsonfile, making sure that the files it lists are present in the Huggingface dataset with the correct hash, and change the URLs to point to the new locations. Once that's done run all tests to ensure they function. If you want to propose a draft PR before you've gotten everything working we can have a look if you need the help. Thanks!e-mny commentedon Jan 15, 2026
Hi Eric, thanks for the informative reply! I have edited the data_config.json file.
The problems I encountered were as follows:
No hash_val found on HF
Not sure how to find the hash_val on HF if it isn't found on the link itself, so I left the file unchanged in the data_config.json file.
Other than that, I have submitted a PR, and my PR is at #8699
Let me know if you want me to change anything.
Refactor download_url test into TestDownloadUrl class (#8829)
Refactor download_url test into TestDownloadUrl class (Project-MONAI#…