I am a bit confused about the graph that is created by ‘rasa visualize’. Am I right that the machine learning algorithm basically creates a graph based on the stories and that all conversations flows will then follow the possible paths in the graph? In other words, once the model is estimated the possible dialog flows are fully predictable based on the graph? Thank you in advance for your insights!
rasa visualize is independent of model training. It basically picks up all the stories and constructs a graph of all paths in your data. This can be an important step for you to validate how your stories look and see if they cover all possibilities you want them to cover before training your model on them.
Thanks for your reply! Is there any other way to visualize the output of the model as to validate it will give the correct answer to certain questions?
Now I’m a little bit confused: what is the use of traning RASA core after Rasa has showed all posible paths?
So, what extra value will core training add?
I suppose the predictions of the paths?
So, as Martine asks, how to check the path áfter training.
More-over, if I understand the documentation on augmentation well, RASA glues stories. Is clueing being done before or after training?