QuickStart

Get started with MinT in minutes. This section covers installation, environment setup, and hands-on tutorials.

Getting Started

• Installation — Install the MinT SDK and configure your API key
• Navigating the Docs — How to find what you need

Tutorials

Step-by-step examples that walk you through training with MinT:

• QuickStart Tutorial — Train a model to solve multiplication using SFT, then refine with RL
• Custom Reward — Define and use custom reward functions
• Mini RL Trip — A minimal end-to-end RL training example
• Multi-Turn Training — Train on multi-turn conversation data
• RL Best Practices — Tips for stable and efficient RL training
• Math RL — Train models on mathematical reasoning
• Code RL — Train models on code generation
• Chat RL — Train chat models with RL
• Sampling Log — Inspect and analyze sampling outputs

Source Code

All tutorial code is available in the mint-quickstart repository.