Hugging Face is a popular library for building and training NLP models. It also hosts 1000's of pre-trained models contributed by the community.
As a developer, I (and my customers) tend to go-to HF as the primary library for NLP.
Since NeMo has the ecosystem for all 3 components of conversational AI, it'd be great if we can add HF models to NeMo NLP collections.
We've done that for RoBERTa and Albert (DistillBERT is on the way), I'm wondering if we can create a generic guide to wrap any HF model in a neural module to use inside NeMo.
Thank you planning this!