Download WD tagger models using HF's snapshot_download() #1162
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The issue is that newer WD tagger models were released and they don't follow the usual TF repo structure, while their ONNX versions can be used as a drop-in replacements:
https://huggingface.co/SmilingWolf/wd-convnext-tagger-v3
https://huggingface.co/SmilingWolf/wd-vit-tagger-v3
https://huggingface.co/SmilingWolf/wd-swinv2-tagger-v3
This PR replaces old hf_hub_download code with snapshot_download() to allow downloading the entire repo instead of chosen files.
This change quickly allows downloading and running the newest WD tagger ONNX models while (hopefully) not breaking compatibility with the old code and models. Since these new models don't rely on tensorflow, and only ONNX version runs without any modifications, I would like to ask you how would you prefer to deal with it.