- Download the dataset
wget https://huggingface.co/datasets/m-ric/huggingface_doc/resolve/main/huggingface_doc.csv
- Build the db
python build_vector_database.py
To run a basic RAG test
python rag.py
- Generate the dataset
python test_procedure_for_rag/generate_qa_pairs.py data/chroma_db_1000/ data/
- Generate the answers using the RAG
python test_procedure_for_rag/generate_answers.py data/chroma_db_1000/ data/qa_dataset_limit\=10.csv data/
- Evaluate
python test_procedure_for_rag/evaluate.py data/chroma_db_1000/ data/qa_dataset_limit\=10.csv data/qa_dataset_limit\=10_answers.csv