Camphr provides functionality for using transformers on spaCy. The purpose of this feature is similar to spacy-transformers.
You should use Camphr in the following cases:
1. If you want to extend or combine model with pytorch (spacy-transformers is thinc-based, while Camphr is pytorch-based.)
2. If you want to fine-tune easily with CLI
3. If you want to use better token alignment algorithm (Camphr uses pytokenizations (https://github.com/tamuhey/tokenizations) to align transformers' tokens and spaCy's tokens. pytokenizations is faster and more robust than spacy-transformers' alignment algorithm)
1. If you want to extend or combine model with pytorch (spacy-transformers is thinc-based, while Camphr is pytorch-based.)
2. If you want to fine-tune easily with CLI
3. If you want to use better token alignment algorithm (Camphr uses pytokenizations (https://github.com/tamuhey/tokenizations) to align transformers' tokens and spaCy's tokens. pytokenizations is faster and more robust than spacy-transformers' alignment algorithm)