I have a use case where I need to train chatGPT to answer questions based on a set of say 100 PDFs.
For example take a model of a car. There may be 100 non public PDFs describing its repairs, systems, regulatory approvals, pricing etc.
What’s the best way to train chatGPT to respond to questions and answer based on this information corpus.
Eg " What color and location is the fuse for the windshield wipers for Model foobar" is a question that may not be answerable using chatGPTs innate intelligence but could be specifically and easily answered by a knowledge of the pdf corpus.
I am open to solutions based on ChatGPT Plus ( preferred) but if it required me to use the API or a custom model that’s fine too.
Any suggestions?