RASA Spacy sklearn pipe line


(DIleepkumar T) #1

I have created a config file as below “language”:“en”, “pipeline”:“spacy_sklearn”

Do I need to specify any other parameters additionally? What is the difference between my config file and below one? language: “en”

pipeline:

  • name: “nlp_spacy”
  • name: “tokenizer_spacy”
  • name: “intent_entity_featurizer_regex”
  • name: “intent_featurizer_spacy”
  • name: “ner_crf”
  • name: “ner_synonyms”
  • name: “intent_classifier_sklearn”

I was not able to classify my intents as expected. I think issue is with stop words. Can I get some info on this also?


(Deepak Shetty) #2

https://rasa.com/docs/nlu/pipeline/#section-pipeline

spacy_sklearn is a preconfigured pipeline and is the same as language: “en”

pipeline:

  • name: “nlp_spacy”
  • name: “tokenizer_spacy”
  • name: “intent_entity_featurizer_regex”
  • name: “intent_featurizer_spacy”
  • name: “ner_crf”
  • name: “ner_synonyms”
  • name: “intent_classifier_sklearn”

so its the same. For the rest hard to say without any data.


(Yogesh Kothiya) #3

The one you using is just a shortcut for a full list of components which you compared with. You can get detail information about every component here.

Please share your training data then only we can comment about intent misclassification.