PDF GPT query interface

Tool to search PDF documents using OpenAI and FAISS (Facebook AI Similarity Search).

• Recursively extract text from PDF documents from the given directory
• Load and split PDF document in to text chunks using PyPDFLoader langchain document loader
• Embed text chunks using OpenAI API embedding
• Store embeddings in FAISS index
• Use retrieval to find relevant text chunks for a given query using OpenAI API LLM

Install dependencies

## Create conda environment using environment.yml
conda env create -f ./environment.yml
conda activate pdf-gpt-query

Run

Use default PDF in data directory:

OPENAI_API_KEY="..." python main.py 

Use a directory with PDFs (it will recursively find all PDFs in the directory, can be slow):

OPENAI_API_KEY="..." python main.py ~/Documents