A simple Gradio interface for semantic search across multiple PDF documents using a combination of BM25 and vector embeddings to find relevant documents. The script builds a FAISS index on corpus of the uploaded documents, and first uses BM25 to find the top relevant results, then reranks them using cosine similarity to the search query.

Link to venv docs

python3 -m venv venv

UNIX/MacOS:

source venv/bin/activate

Windows:

venv/Scripts/activate

If this is your first time running this or the package dependencies have changed, run this command to install all dependencies.

pip install -r requirements.txt

Run the app in reload mode with this command. This will let the app reload automatically when changes are made to the python script.

python main.py