pdfGPT icon indicating copy to clipboard operation
pdfGPT copied to clipboard

Upload PDF Display Error

Open wunuo1 opened this issue 2 years ago • 8 comments

Seems to be due to the issue of the PDF file being too large?What is the supported file content size?

wunuo1 avatar Apr 28 '23 08:04 wunuo1

Running on the demo URL

wunuo1 avatar Apr 28 '23 08:04 wunuo1

Same issues here

mokby avatar May 01 '23 14:05 mokby

I am working on this

bhaskatripathi avatar May 02 '23 02:05 bhaskatripathi

Same over here. Any update on the "error" issue? grafik

Kunstbanause avatar May 07 '23 20:05 Kunstbanause

I think I might know why , I upload a paper with sevealr formulas and charts,I got an erroe,but when I upload a file just words ,then it would work correctly.However, I use it is for reading API books which include many formulas and charts,I pretty need this function.Again ,thanks for the developer of this project , you are really awsome.

mokby avatar May 08 '23 00:05 mokby

I tried running the source code and found that an excessive number of tokens can lead to errors. The model's tokens are limited, perhaps due to the same reason

wunuo1 avatar May 08 '23 02:05 wunuo1

You can try reducing max_ token or use gpt-3.5 turbo

completions = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        temperature=0,
        messages=[
                {"role": "user", "content": prompt}
            ]
    )
    message = completions['choices'][0]['message']['content']

wunuo1 avatar May 08 '23 02:05 wunuo1

You can try reducing max_ token or use gpt-3.5 turbo

completions = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        temperature=0,
        messages=[
                {"role": "user", "content": prompt}
            ]
    )
    message = completions['choices'][0]['message']['content']

Thanks,that helps a lot , I`ll try it.

mokby avatar May 08 '23 02:05 mokby