Training a test model

Test models are used for testing your chatbot's NLP question responses. If you're only testing your chatbot's scripts, actions, keyword questions, variables, or conversation flow, you don't need to train a test model.

TestBot uses training models to ensure you're testing the most up-to-date version of your NLP questions. Training a new model can take up to ten minutes. You won't be able to test your chatbot's NLP questions until the model has been trained, but you can still test the rest of your chatbot content while you wait.

Trained models are discarded:

  • 30 minutes after your last test.
  • If you or someone on your team starts training a new model.

The TestBot model is shared across all members of your team for that chatbot. You should coordinate when you train a model with your team members so as not to disrupt each other's tests.

Previewing on Facebook Messenger always trains a new model when you press the Send to Messenger button.

Do I need to train a model?

You only need a trained model if you want to test your chatbot's NLP question responses. If you are just testing your chatbot to check a script or conversation flow, or checking variables have been set correctly, you can test without a trained model.

If you're testing NLP question responses, check the Testbot message above the green start button to see if a model is available. If you see the message Requires training, you'll need to wait for a model to be trained before you can test your NLP question responses.

If you have made changes to NLP training phrases, you will need to train a new test model to see those changes in TestBot. You don't need to train a new model if you have made changes to passages, scripts, broadcasts or keyword questions: that content will update automatically in TestBot without retraining.

Was this article helpful?
0 out of 0 found this helpful



Article is closed for comments.