How to train pdf document

  • Resolved waqasnu1

    (@waqasnu1)


    1 year, 11 months ago

    I have a pdf document having more than 3000 pages. How I can train it using gpt? Where should I upload or provide the file? Also whenever chatgpt widget is used, I wish to return the results from my trained pdf data only nothing from web or gpt own ai.

Viewing 3 replies - 1 through 3 (of 3 total)
  • Plugin Author senols

    (@senols)

    Hi @waqasnu1

    I am currently researching the most effective method to embed PDF documents. The process will likely involve converting the PDF to text and then splitting it into smaller segments, as directly embedding the entire document may not be feasible. I’m working on finding the best solution and hope to have more information soon.

    Thread Starter waqasnu1

    (@waqasnu1)

    That would be awesome as pdf document is widely used and we have most of the data in pdf’s.

    I just noticed a strange issue, whenever I enable embeddings only from the chatgpt settings, and try any query it starts returning the result from embeddings as it is without any artificial intelligence. So let’s say there is some URL in embeddings, querying anything related to that URL starts returning the URL and the same is for other content types.

    The reason why I am pointing this out is that I tried to create some embeddings from my pdf document content by copy pasting page text and creating a new embedding entry but this process is returning the whole embedded entry as it is when queried on chatgpt widget.

    Thanks and your plugin is awesome

    Plugin Author senols

    (@senols)

    Thank you ??

    I am closing this topic for now as this is a feature request and will take some time to implement ??

Viewing 3 replies - 1 through 3 (of 3 total)
  • The topic ‘How to train pdf document’ is closed to new replies.