Story not detected?

Hi there,

I’ve this story:

- story: interactive_story_1
  - intent: password_recover
    - password: password
    - secret_question: i know the secret question
  - slot_was_set:
    - secret_question: i know the secret question
  - action: utter_password_recover_knows_answer

I’ve already tried lots of times rasa interactive / shell, and when the user asks the question “I lost my password, but i know the secret question. How do I recover it?” (which was the question that “created” the story above) it always gets “action_default_fallback”.

Any idea why is the story not detected? I’ve at least 25 examples in the intent “password_recover”.

This is my config file:

recipe: default.v1
language: pt

# # No configuration for the NLU pipeline was provided. The following default pipeline was used to train your model.
# # If you'd like to customize it, uncomment and adjust the pipeline.
# # See for more information.
  - name: WhitespaceTokenizer
  - name: RegexFeaturizer
  - name: LexicalSyntacticFeaturizer
  - name: CountVectorsFeaturizer
  - name: CountVectorsFeaturizer
    analyzer: char_wb
    min_ngram: 1
    max_ngram: 4
  - name: CRFEntityExtractor
  - name: DIETClassifier
    epochs: 100
    learning_rate: 0.005
    entity_recognition: False
    #batch_size: [8, 32]
  - name: EntitySynonymMapper
  - name: ResponseSelector
    epochs: 100
  - name: FallbackClassifier
    threshold: 0.3
    ambiguity_threshold: 0.1

  - name: MemoizationPolicy
  - name: TEDPolicy
    max_history: 5
    epochs: 30
  - name: RulePolicy


@nik202 can you give me a hand, please? Thanks a lot!

@nonola Sure, can you tell why you are using CRFEntityExtractor ? when you have DIET? any specific reason for that?

I was testing the trainning time diference between DIET with / without entities detection, and CRFEntityExtractor (this one is faster extracting entities). Should I forget it?

@nonola try without please and check its working or not?

I tried and this is what I get:

? Your input -> perdi a minha senha e não sei a resposta de segurança                                                                                        
? Is the intent 'senha_recuperar' correct for 'perdi a minha [senha](senha) e [não sei a resposta de segurança]{"entity": "pergunta_secreta", "value": "n\u00
e3o sei pergunta secreta"}' and are all entities labeled correctly? Yes                                                                                      
Chat History

 #    Bot                                                                                                                                    You        
 1    action_listen                                                                                                                                     
 2                       perdi a minha [senha](senha) e [não sei a resposta de segurança]{"entity": "pergunta_secreta", "value": "n\u00e3o sei pergunta 
                                                                                                                           intent: senha_recuperar 1.00 
 3    slot{"pergunt                                                                                                                                     
      "não sei                                                                                                                                          

Current slots: 
        pergunta_secreta: não sei pergunta secreta, tipo_problema: None, senha: senha, session_started_metadata: None

? The bot wants to run 'action_default_fallback', correct? (Y/n)                                                 

If you want I can translate it to English, but “pergunta_secreta = secret_question”.