ChatWithPDF

ChatWithPDF is a ChatGPT plugin that allows users to load and query PDF documents using ChatGPT. Users must provide a temporary PDF URL for processing. Once the PDF is loaded, users can query, analyze, or ask questions based on the PDF name.

Installation

ChatWithPDF is a ChatGPT plugin. To install ChatWithPDF, simply add it in the "Plugin Store" once you select Model: GPT-4 --> Plugins --> --> Plugin Store in the ChatGPT UI (once you have Plugin access)

Usage

First load up the PDF (either enter "load " or you can load it when asking the question)

Second ask a question about the PDF. Here are a few examples:

• "Search for the definition of machine learning in the PDF"
• "Find the section about data preprocessing in the document"
• "List the advantages of deep learning from the PDF"
• "What are the main points in the executive summary?"
• "Locate the results of the experiment in the paper"

Overview

This ChatGPT plugin offers a convenient way to process and semantically search PDF documents based on your queries. By providing a temporary PDF URL, the plugin fetches relevant information from the PDF file and returns the most suitable matches according to your search input.

Features

• Load Google Drive documents that are publically accessible
• Load and process PDF documents from a temporary URL
• Semantic search within PDF documents
• Extracts relevant information from the document based on user queries
• No installation required. Just add chatwithpdf.sdan.io in the Plugin store

Privacy

ChatWithPDF does not intentionally store any data permanently. All PDFs are embedded and immediately wiped. Embeddings are stored with ChromaDB on the same depoyment server and are wiped with each new deployment. In the future all embeddings will be deleted 1hr after the user requests them.

How it works

• Users provide a temporary PDF URL to be loaded and processed
• The plugin downloads and processes the PDF document, extracting relevant information
• User queries are matched with the processed information from the PDF
• The most relevant matches are returned and displayed to the user