In my data/nlu.md file, I have defined the word ê to be greeting. However, then I run rasa shell nlu, and type the word ê, the NLU model cannot detect it. But when I type other words such as ê, m tên gì or hello, then the model can detect
hi @TQuy - thanks for your question! I guess you are using the supervised embeddings pipeline ? If so you’ll have to set the minimum word length of the CountVectorFeaturizer to 1 (default is 2)
See Components