extras provides a way to define a captioning model, but a very popular, fast and detailed captioning model does not yet seem supported: https://huggingface.co/SmilingWolf/wd-v1-4-vit-tagger-v2/tree/main I assume it's because it uses onnx, would this be feasible to add or is there WD14 versions that are compatible with whatever loading happens in extras? thanks
extras provides a way to define a captioning model, but a very popular, fast and detailed captioning model does not yet seem supported: https://huggingface.co/SmilingWolf/wd-v1-4-vit-tagger-v2/tree/main I assume it's because it uses onnx, would this be feasible to add or is there WD14 versions that are compatible with whatever loading happens in extras? thanks