Currently the models are hardcoded, for example, in Grievance Recognition and Conversation Terminator.
Task
- Enable dynamic loading of models upon request, since models are constantly being improved, if not deployed at the same endpoint we would have to change the file every time a model is updated.