Systex/faiss_index/README_faiss.md

Comparing Faiss with the original RAG. Both using the TAIDE LLM model. Both using created embeddings from the Supabase db.

Did not apply multi_query on Faiss (as it generates very similar answers from the original RAG). Applied multi_query on the original RAG only.

Evaluate on: 1. speed 2. RAGAS score

Some of the tested result is in the output.txt

Findings:

• Response speed: It's not worth using Faiss to speed up llm on small dataset. Its index method is designed for large dataset.
• Response quality: If relevant documents can be found, the original RAG which spplied the multi query method can generate more detailed answers. Faiss' answers are shorter.
  

Faiss: https://engineering.fb.com/2017/03/29/data-infrastructure/faiss-a-library-for-efficient-similarity-search/