I would like to know how to optimize the documents that are added to a GPT so that it reads them correctly. In the case of PDFs, I think I understand that it reads plain TXT texts better. What file types does it support? Thank you.
Hey there and welcome to the community!
Do you know precisely what problem is occurring when you try to have GPT “read” a PDF? If you provide us your prompt and its response, we’ll have a better understanding of how we could help you.
In a general sense, yes, txt files are the easiest text formats for GPT to handle. GPT can analyze most file types, unless it’s specialized or niche, so I think the bigger question is what file type do you want analyzed? Don’t be afraid to give it different files to see what happens. There’s no consequences for giving it a file it can’t interpret; it’ll just say so.
In terms of optimizing files for GPT, that question is more difficult to answer without knowing precisely what you want it to do, and what has currently been unsatisfactory.
Thanks for your answer. I am a teacher and I usually use PDF, text only, but not all of them are equally readable. I am thinking of transforming the contents to TXT. A greeting from Spain.
So, since you’re a teacher, question for you: Are these PDFs you’re working with Scans of text or some other physical medium (book, paper, etc), or were they originally another e-doc (like a word doc) that was converted to PDF? Or perhaps made entirely from a computer and the doc(s) were saved as a PDF?
I learned this the hard way, but AIs have a much harder time with scanned PDF images. Not just for ChatGPT, but even some AIs that are specifically built to convert PDFs to plaintext. Now, that doesn’t mean those tools aren’t useful, but make sure to check for errors, as they do happen.
Converting them to txt files sounds like a wise idea at this point. Go ahead and try it out with a couple pages, see if GPT “reads” them correctly, and then feel free to return to ask more question if there’s still some problems with it.
Best of luck!