Hello everyone. I wanted to know if Rasa once installed can work completely offline? can I also train it offline? thank you
Yep! Everything happens offline, unless you need the internet in your custom actions (e.g. making API calls to hosted services)
Rasa works offline unless you deploy it into a REST service. In that case, you would have to ping that REST service for an output. However, Rasa has a nice library in python that lets you load a trained NLU model and predict on phrases completely offline. This is achieved using from rasa.nlu.model import Interpreter
and you load the model using my_interpreter = Interpreter.load("path/to/model")
. Finally you can receive the response using my_interpreter.parse("some phrase")
. I currently use this option to perform high volume testing on my model to measure the accuracy of its intent and entity detection.
Sure, but you can also call the REST service via localhost
which should work offline