Hello, Im amazed at this implementation of GPT-3.
I would love to add an extra later to theprompt by preprocessing the text and adding entities with metalabel while passing prompt to GPT-3
An example would be:
Buy [Brand] Adidas shoes from Amazon [company] India [location] under $200 [numeric]
We could achieve this using the hello haptik repository (link: https://github.com/hellohaptik/chatbot_ner)
Im not completely sure if this will help the output. What do you think?
Kudos!