# Read data of pdf or just image format as a part of promt

**URL:** <https://discuss.huggingface.co/t/read-data-of-pdf-or-just-image-format-as-a-part-of-promt/41251>\
**Category:** Intermediate\
**Created:** [May 29, 2023, 2:17am UTC](https://discuss.huggingface.co/t/read-data-of-pdf-or-just-image-format-as-a-part-of-promt/41251 "2023-05-29T02:17:05Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Viking714](https://avatars.discourse-cdn.com/v4/letter/v/65b543/32.png) [@Viking714](https://discuss.huggingface.co/u/Viking714)\
**Post date:** [May 29, 2023, 2:17am UTC](https://discuss.huggingface.co/t/read-data-of-pdf-or-just-image-format-as-a-part-of-promt/41251/1 "2023-05-29T02:17:05Z")

</div>

I have a use want to develop, but I don’t know which workflow I should take by huggingface. First I want to load text(maybe several thousand words) which is in pdf format, and according to the text material I want to ask some questions to model(just like gpt or oasst-sft-4-pythia-12b-epoch-3.5 or some other test-generation model). I know in openai there is function can extraction main meaning of pdf, I don’t want to transfer learn the pretrained model, just ask questions and pdf text together. I have two concerns:1 is how should load the pdf, 2 is if the text is a little long, is it ok just put the text as a part of promt? ifover the tokens limit? which model should I use in huggingface for this situation? Thank you very much.