langflow_ragas

Ragas Evaluation on Langflow

The repository implements the code for RAGAS metrics faithfulness, answer_relevancy, context_recall, and context_precision (https://docs.ragas.io/en/stable/index.html) on a RAG pipeline.

This is done by creating the custom component ragas_custom_component.json.

Tutorial

Clone the repository

git clone https://github.com/paulomuraroferreira/langflow_ragas.git

Install langflow and ragas:

!pip install langflow==1.0.11
!pip install ragas==0.1.10

On the terminal, execute

langflow run

Upload the json RAG pipeline RAGAS metrics.json

Copy the pdf documents to the pdf_documents folder, or change the path in the Document Loader component:

Enter your OpenAI API key on both Embeddings components,

on the OpenAI models component,

and on the Ragas custom component:

Run the chunking pipeline by executing the ChromaDB component:

Enter the playground and ask questions: