New quality evaluations are now available in public preview, accessible through the Azure AI SDK today with UI support coming in October 2024. ROUGE, BLEU, METEOR, and GLEU are popular math-based metrics that can help AI developers assess text-based outputs for qualities such as similarity to expected outputs, precision, recall, and grammatical correctness. AI developers can evaluate each metric using a dedicated evaluator and combine multiple evaluators into a holistic evaluation run. After reviewing evaluation results, users may decide to compare different models, adjust their grounding data, or make other changes through prompt engineering before rerunning the evaluation to see the impact of their changes. Overall, these evaluations can help AI developers improve the quality, accuracy and trustworthiness of their AI application.
Learn more.