Suggestions for using rasa

I want to use only these pipeline as my requirement is entity extraction from lookup files and regex matching for a given input. pipeline:

  • name: SpacyNLP case_sensitive: true
  • name: SpacyTokenizer
  • name: RegexFeaturizer
  • name: SpacyFeaturizer
  • name: CRFEntityExtractor
  • name: “regex.RegexEntityExtractor”
  • name: EntitySynonymMapper
  • name: SklearnIntentClassifier

I don’t have any requirement of building a chat bot. Do I need to install entire rasa tool? Since my requirement is not so big, I feel like installing entire rasa tool creates a huge footprint. Is there any other way? I am new to nlp and rasa. Any suggestions are welcome.

Hi @SaiSatwik, I think the only way would be to install from source and remove the extra pyproject.toml dependencies. But NLU is the vast majority of the dependency size, I don’t think removing core dependencies would do that much for you. you can pip install "rasa[spacy]" to get the extra spacy requirements without the mitie/convert etc other dependencies.

1 Like