Evaluating RAG pipelines with Ragas + LangSmith

The article discusses the evaluation of QA (Question Answering) pipelines in LLM (Large Language Model) applications using Ragas framework and LangSmith platform. It highlights the importance of having a robust eval strategy when taking an application from a cool demo to a production-ready product, especially for LLM applications due to their stochastic nature. The article explains how Ragas helps in evaluating QA pipelines by providing metrics like context_relevancy, context_recall, faithfulness, and answer_relevancy. It also demonstrates the integration of Ragas with LangChain for running evaluations and visualizing results. Finally, it emphasizes that using Ragas and LangSmith can ensure robustness and reliability in QA systems, making them ready for real-world applications.