Hi all.
Just wanted to let y'all know that there's been another iteration on our plugins. This time, we've added initial support for training Hugging Face transformer models directly on Prodigy data. We noticed that some of our users were interested in training a HF model directly without a spaCy wrapper. The new prodigy-hf plugin should make this much easier.
The recipe comes with two recipes right now:
hf.train.ner
which allows you to train NER recipeshf.upload
which allows you to upload your annotated datasets directly to Hugging Face Hub
If you want to learn more you can check out our updated docs. If you want to customise the training further you can also use these recipes as a place to start and customise.