Let's say, as an example, that I want to create a GPT using a book from 1921 that is now out of copyright. I upload the book, but it doesn't really seem like GPT is reading through it, digesting it, and performing quick data usage. Any function using data from the uploaded book seems to take a long time, and it doesn't seem as if GPT natively understand it all. Is anyone having success with this, and if so, how? There are many custom GPTs that I have ideas to create, but they all require incorporation of the data into the fast recall responses.
#Are people having success with uploading large volumes of document data? I am having minimal success
hollow laurel
obsidian wasp
It's probably too long for the available context length so it's searching the file for given parameters, instead of reading it all
hollow laurel
what is the context length that would theoretically be allowed in order for the ingestion of the data to function with some sort of rapidity and cognitive understanding?