Intel Labs announced RAG Foundry, an open-source Python framework aimed at developing and experimenting with retrieval-augmented generation (RAG) systems. This framework is designed to address challenges in training and evaluating RAG models, offering a unified workflow for data creation, training, inference, and evaluation at no cost.

The RAG Foundry framework integrates key components such as data creation, training, inference, and evaluation into a single workflow. Its modular structure, controlled via configuration files, allows isolated experimentation and customization. 

Key benefits of the RAG Foundry include enhanced flexibility for researchers to experiment with various RAG aspects, such as data selection, retrieval accuracy, and prompt design, improving model performance across different contexts.