Rankify

Why we included this project

Rankify is a Python library for teams building search and question-answering systems. It wraps the whole retrieve-rerank-generate loop in one importable toolkit, so instead of gluing together separate libraries you can call a one-line pipeline that picks a retriever (BM25, or dense models like BGE or ColBERT), a reranker, and a generator, then swap components as your needs change. It bundles 40 pre-retrieved benchmark datasets and a CLI for indexing your own corpora, which makes it handy for reproducing published results and for checking which retrieval or reranking model actually performs best on your data. Optional extras let you install only the parts you need, so a small experiment doesn't drag in the whole stack, while heavier GPU-backed deployments can pull in everything. If you want a consistent, well-documented way to compare modern retrieval and reranking approaches before committing to a production stack, this is a good starting point.