​

Reward Kit Documentation

Welcome to the Reward Kit documentation. This guide will help you create, test, and deploy reward functions for evaluating and optimizing LLM responses.

​

Getting Started

​

Developer Guide

• Getting Started with Reward Functions: Learn the basics of reward functions
• Reward Function Anatomy: Understand the structure of reward functions
• Core Data Types: Explore the data models used in reward functions
• Evaluation Workflows: Learn the complete lifecycle from development to deployment

​

Examples and Built-in Reward Functions

• Reward Functions Overview: Overview of all built-in reward functions
• Basic Reward Function: Simple example evaluating response clarity
• Advanced Reward Functions: More complex examples with multiple metrics

​

Built-in Reward Function Documentation

• Code Execution Evaluation: Evaluate code by running it locally
• Code Execution with E2B: Evaluate code using E2B cloud sandbox
• Function Calling Evaluation: Evaluate function calls made by AI models
• JSON Schema Validation: Validate JSON outputs against schemas
• Math Evaluation: Evaluate mathematical answers in responses

​

Tutorials

• Creating Your First Reward Function: Step-by-step guide to creating a reward function

​

API Reference

• Coming soon

​

Command Line Interface

• Coming soon

​

Best Practices

• Coming soon

​

Community and Support

• GitHub Issues: Report bugs and request features
• Contributing Guide: How to contribute to the Reward Kit project