Hello everyone, whenever i type some random queries, i am getting intent as “None” even if I write query with simple mistake.
I am using tensorflow pipeline.
Please help me out.
Hey @raghibomar. Could you share your pipeline configuration, some training data and the command you are using to parse the inputs? It would be easier to find out where the problem is if I could reproduce it 
language: “en”
pipeline: “tensorflow_embedding”
response = requests.get(“http://localhost:5005/conversations/"+hashed_sender_id+"/respond",params={"query”:user_message})
here “user message” is message that user typed
How many training examples do you have? Tensorflow embedding might need a bit more training examples to learn which entites to extract.
Right now we have approximately 1800 sentences in nlu_data.md
Hmm, that looks like a lot. The thing is that tf embedding model can predict None it doesn’t find the likely intents, also, since the sentences are tokenized by words, typos increase the likelihood of that the model will be confused if typos are present. Are you able to get the right predictions if you send queries without the typos and that should potentially correspond to the intents you have defined in your training data?