Getting CLS from DIET intent classifier at inference


Let’s say I trained a bot with rasa using a predefined pipeline that uses DIET classifier. Now I want to load the CLS token as described in the architecture that’s being used for intent classification.

I would load my model as follows

from rasa.nlu.model import Interpreter
model = Interpreter.load("my_model/nlu")
output = model.parse('text', only_output_properties=False)

Then, how can I get the CLS that was being used for that data point (in my example text).