Change the Model that File Q&A Uses


I’m looking for a way to change the model used in File Q&A. I notice it uses text-embedding-ada-002 to index and then text-davinci-003 to return your search results. How would I change the code to allow a model other than text-davinci-003 for search results? I tried changing a few fields related to the model, but kept getting errors. :sweat_smile:

Any help would be appreciated!


Are you trying to use one of the newer chat models? If so, you’re probably forgetting to format the prompt using ChatML.

Yeah I had a feeling after watching this: Getting started with the NEW OpenAI ChatGPT API (gpt-3.5-turbo) - YouTube

The code is different so I’m not sure where to make the change though.