This cookbook demonstrates how to build a Retrieval-Augmented-Generation (RAG) pipeline, step-by-step.
It features an automated evaluation process and has many knobs to let you activate or deactivate some RAG tweaks and see if the performance increases.
The specific use-case demonstrated is a QA bot that answers your questions about Hugging Face packages based on their documentation. But this can be adapted, it's very simple to swith to another base of documents!
👉 Try it yourself! 🚀