Rasa Knowledge Base bot with spacy pipeline

Hello, I created a working knowledge base english bot based on the example on github and would like an assist how to configure my pipeline with spacy instead of supervised_embeddings. I don’t know if is possible. Anyone can help me with this?

My current config is below:

language: en pipeline: supervised_embeddings policies:

  • name: MemoizationPolicy
  • name: KerasPolicy
  • name: MappingPolicy

I tried with pipeline: pretrained_embeddings_spacy but my bot seems to not understand the user. I also get this error: \lib\site-packages\rasa\nlu\extractors\crf_entity_extractor.py:590: UserWarning: Number of features (1) for attribute ‘text_dense_features’ does not match number of tokens (6). Set ‘return_sequence’ to true in the corresponding featurizer in order to make use of the features in ‘CRFEntityExtractor’. f"Number of features ({len(features)}) for attribute "

1 Like

any help with this?

@sterpapi did you create a custom policy? because in rasa docs there is 3 policies based on ML (TEDPolicy, Memoization Policy, Augmented Memoization Policy) and one rule based policy. so what is KerasPolicy and MappingPolicy?