📚 Welcome to RM Library

Explore our comprehensive collection of 40+ ready-to-use reward models designed for various evaluation scenarios. Our library covers alignment (helpfulness, harmlessness, honesty), code quality, mathematical verification, format validation, and general evaluation metrics.

🎯 What you'll find:

• Alignment Models: 21 models for HHH (Helpfulness, Harmlessness, Honesty) evaluation based on RewardBench2 and RMB Bench
• Code Quality: 4 models for syntax checking, style validation, patch similarity, and execution testing
• Math Evaluation: Mathematical expression verification with LaTeX support
• Format & Style: 5 models for format validation, length control, repetition detection, and privacy protection
• General Metrics: 4 models including accuracy, F1 score, ROUGE, and numerical accuracy

💡 Use the search bar below to find specific models, or browse by category to explore our full collection.