Quick Start

This guide helps you get started with LightRFT quickly, from basic concepts, user guide (algorithm and configuration), to running your first training experiment.

Core Concepts

• LightRFT Project Overview
  • High-Level Architecture
  • Training Flowchart
  • Core Modules
  • System Workflow
  • Extension Points

User Guide

• Supported Algorithms
  • Purpose of This Guide
  • Algorithm Overview with Implementation
  • Algorithm Architecture
  • Policy Optimization Algorithms
  • Advantage Estimation Methods
  • Reward Processing
  • Sampling Strategies
  • Implementation Notes
  • References
• Configuration Parameters
  • Table of Contents
  • Model Parameters
  • Training Parameters
  • Batch Size Configuration
  • Algorithm Parameters
  • Distributed Training
  • Inference Engine
  • Memory Optimization
  • Logging and Monitoring
  • Checkpoint Management
  • Reward Processing
  • Multimodal (VLM) Parameters
  • Complete Example Configuration
  • Parameter Validation
  • Environment Variables
  • See Also