Hello,
I was trying to connect a custom NLP model to rasa which with input X returns text output Y. It is hosted on cloud.
As I am new to it, I am confused how can I achieve this result with custom actions. And where does the endpoint.yml come into the picture as it was suggested in the docs for connecting custom NLG models
You can use custom python to connect via HTTP if you’d like. If your endpoint talks HTTP this is certainly possible.
I would like to give a small warning here though. Depending on the generative model you should be aware that you cannot control what comes out of it. Even models like GPT-3 are able to generate texts that read like bigotry so I’d take care ensuring this does not happen.
Out of curiosity, what is the use-case?