Evaluation

Measuring model quality for tasks and safety.

Topics

• Task metrics (accuracy, F1, BLEU, ROUGE)
• Human eval vs automated eval
• Prompt robustness and adversarial testing
• Safety, bias, and fairness checks