GAIR-NLP/auto-j

Features

• Generative Reward Models - Generative judge for alignment evaluation.
• Model Evaluation Metrics - Provides a generative judge for evaluating model alignment.