Once you have created the stories, you can train the dialogue management model, an LSTM based Recurrent Neural Network.
Rasa Core uses the outputs of the NLU model, that is, intents and entities, and then trains a neural network model on the stories. Rasa Core uses Tensorflow, a library for training neural networks, at the backend. More specifically, it uses LSTM neural networks implemented in Keras. You’ll learn more about Tensorflow, Keras and LSTMs in the next course on DL.
You can enable Rasa core to also provide support to test the chatbot by typing rasa shell through the command line prompt.
You can try out some variations in your replies and see how your chatboat responds.
Note: While you are testing the chatbot, you need to make sure to start your action server if you want to run your custom actions.
In the next segment, you’ll learn is how to create more stories, i.e., how you can add variance in the data by providing more training examples using Rasa Core’s ‘interactive learning’ feature.