Model Evaluation
Automatic evaluation in Amazon Bedrock allows using standard metrics (such as accuracy, robustness, and toxicity) to evaluate models by comparing their outputs against a pre-labeled reference dataset, without human intervention.
check_circleTherefore, the correct answer is: Automatic evaluation.
cancelThe option that says: Human evaluation is incorrect because it involves human reviewers, which is not automatic.
cancelThe options that say: Performance/Cost evaluation are incorrect because 'Model Evaluation' focuses on response quality (accuracy, toxicity), not latency or price.