Lasagna
Swap one RAG pipeline stage, hold the rest fixed, and measure whether the gain survives composition. The framework behind "Retrieval Augmented Generation Compos…
Lasagna is a framework for retrieval-augmented generation. It allows writing one pipeline stage and sweeping it against built-in components with every other stage held fixed, producing a significance-tested verdict. One decorator registers a technique as a component of its stage, interchangeable with every built-in component, for rerankers, query enhancers, chunkers, embedders, retrievers, document processors and generators.
Each comparison in a sweep gets a paired sign-flip test with Holm correction. Lasagna is open source under the Apache 2.0 license. The framework can be installed via pip with the command pip install lasagna-rag.
Summary drafted from the project's own website. Every sentence is backed by text on that page and was reviewed before publishing.