RAGatouille
AIOpen SourceFreePython library that simplifies using ColBERT retrieval methods in RAG pipelines for scalable, accurate BERT-based search.
RAGatouille
Python library that simplifies using ColBERT retrieval methods in RAG pipelines for scalable, accurate BERT-based search.
About RAGatouille
RAGatouille is an open-source Python library designed to make ColBERT-style late-interaction retrieval methods easy to use and train within Retrieval-Augmented Generation (RAG) systems. It provides high-level APIs, modular components for indexing/encoding/scoring, and utilities to train and evaluate retrieval models on custom corpora. The project emphasizes bridging research and production by exposing practical workflows, LangChain integration points, and tools to deploy scalable BERT-based search over large text collections with improved accuracy over basic vector search.
Screenshots
