Hello, I have a fine-tuned DialoGPT model and I was wondering if it could be combined with Deepy… My idea is to use Deepy for NLU and dialogue management, and DialoGPT for generating replies. Did anyone try something like this?
Hello, you can use your fine-tuned DialoGPT model to create a skill in the folder
deepy/skill, add it to
docker-compose.yml and to
deepy/agent/pipeline_conf.json. The response of the skill will be used as hypotheses for bot response.
Thanks! Do you have any tutorial to recommend for making a new skill using my fine-tuned DialoGPT model?