ragas

ragas provides a framework for evaluating Retrieval Augmented Generation (RAG) pipelines. It offers a set of metrics based on the latest research to assess the performance of RAG-generated text and integrates with CI/CD for continuous performance monitoring.

• Easy Installation: Install ragas with a simple pip command or clone the repository from GitHub
• Diverse Metrics: Evaluate generated text using various metrics including Levenshtein distance, ROUGE-N, ROUGE-L, BLEU, Bert Score, and Entailment Score
• Dataset Evaluation: Evaluate RAG pipelines on datasets by comparing the generated text with ground truth data
• Performance Insights: Gain insights about the performance of RAG pipelines and assess the factual consistency and similarity between generated text and ground truth