Model evaluation

Rasa Open Source
1

Can anyone please explain, how to evaluate the model on test data. I was working with rasa workshop notebook.

from rasa_nlu.evaluate import run_evaluation

run_evaluation(“nlu.md”, model_directory)

In this implementation, how can we pass an unseen test data. From my understanding, I think it need a json file. Can any one please explain model evaluation.

Thanks Rohit

2

You can run the evaluation script in crossvalidation mode, or create a seperate test set and provide this as a file

3

And what’s the format for the test json file, please?

1 Like