Prodigy-HF plugin enables direct Hugging Face training and dataset upload from Prodigy

AI Impact Summary

Prodigy introduces the Prodigy-HF plugin, enabling direct training and reuse of Hugging Face models on Prodigy-annotated data via hf.train.ner and hf.train.textcat, with models serialized to disk and the option to upload to HF Hub using hf.upload. The integration leverages AutoTokenizer/AutoModel from Hugging Face to support any transformers model, reducing data export steps and enabling end-to-end annotation-to-model cycles. Expect quicker iteration for NER and text classification workflows and easier sharing of datasets across teams.