After seeing much desire for a GPT chatbot that uses your own PDFs, I decided to code this project. I hope you enjoy it!
To set up this project, first you need to add your OPENAI_API key as an env variable, or add it on line 13 of the .py file and deleting line 12. Next, run pip install -r /path/to/requirements.txt. Lastly, put all of your PDFs you wish to chat with in a central repository, and add the path to PATH_TO_PDFs in the .py file.
Use the command
python -m streamlit run gpt_cust_kb.py.
This will start the app in your browser. Once you have uploaded your PDFs, refresh the browser, select a manual, and ask away!
WORK IN PROGRESS
Currently the .py file is all you need to produce the functionality shown in the video. Manual/PDF selection is not working as of yet, but due to the questions about how to make this work I wanted to provide some source code for the time being. Follow for updates as I change this project into something that allows you to run locally and upload your PDFs, then select them on the sidebar as you wish. Thanks!
TODO: Write credits
TODO: Write license