OOV token not found in NLU

Hi. I have updated the config.yml by adding the following lines.

  • name: CountVectorsFeaturizer OOV_token: oov token_pattern: (?u)\b\w+\b
  • name: CountVectorsFeaturizer analyzer: char_wb min_ngram: 1 max_ngram: 4

I have added multiple entries of oov in NLU. But during training, the rasa gives me warming.

UserWarning: The out of vocabulary token ‘oov’ was configured, but could not be found in any one of the NLU training examples. All unseen words will be ignored during prediction.

What I am doing wrong here?

First step: Tell us which RASA version you’re using and provide some of the training examples with oov-tokens.

I am using Rasa 1.8.1.

Here is my training example.


  • hey
  • hello
  • hi
  • good morning
  • morning bot
  • good evening
  • hey oov
  • Hi
  • Hey
  • Hi ovv
  • Hey oov
  • Hello
  • Good morning
  • hi oov

I am still looking answer for this.