I have a codebase I want to train GPT on so that I can ask it questions. Is there any way to accomplish this with either GPT or any other LLM?

My current challenge is the tuneable davinci model from openAi is not as good as text-davinci and gpt turbo. But also the finetuning is only based on simple labelled data. I want it to be able to interpret my codebase on its own and train up a version of an LLM which understands and can come up with ideas

Is this a long shot? I've noticed Bing can sometimes search up pages of documentation and gives decent instructions