daveshap / literaturereviewbot

Experiment to use GPT-3 to help write grant proposals.

Literature Review Bot

Setup

Docker and Qdrant

1. Download and setup Docker and/or Docker Desktop
2. Pull the Qdrant container and run it: docker run -p 6333:6333 -v ./qdrant_storage:/qdrant/storage qdrant/qdrant

Download arXiv metadata from Kaggle

1. https://www.kaggle.com/datasets/Cornell-University/arxiv

Or download my processed data directly (but it is out of date)

1. https://www.kaggle.com/datasets/ltcmdrdata/arxiv-embeddings (I will not be maintaining this)

Process data

1. Run generate_embeddings.py to fill up embeddings folder (you may need to create this folder first)
2. Fire up Qdrant if its not already running
3. Run index_arxiv_metadata.py to upload embeddings to Qdrant
4. Run search_server.py and go to http://127.0.0.1 to search for your articles
5. Download the PDFs you want from arXiv into th PDFs folder
6. Run generate_literature_review.py to create your final literature review