Hello, Is there any way to visualize the trained model of the rasa chatbot through jupyter note book by generating confusion matrix and F1- score? I know there is a tool called rasa lit but i don’t think they offer generate graph for both mention above . Have anyone tried before willing to share ?
You can generate confusion matrices and confidence histograms using rasa test (see this).
You can also use Tensorboard to visualize train and test accuracies and F1 for DIET, ResponseSelector, and TED (see this).
1 Like
I hope this repo will help you if you want to run inside Jupyter Notebook: EmergencyChatbot/Rasa_Emergency_Chatbot_Colab.ipynb at master · claudio1975/EmergencyChatbot · GitHub
But, I believe, rasa already has an inbuild function rasa train which will visualize your model as suggested by Chris.
@BELIEVEIT Do check this : Making Apps for the Rasa Research Team & Open Source Community @BELIEVEIT Do check this: GitHub - RasaHQ/rasalit: Visualizations and helpers to improve and debug machine learning models for R
This is for your idea, and you need to implement the code for yourself. Good Luck!
2 Likes