Slapo Documentation

Slapo is a schedule language for progressive optimization of large model training. It decouples model execution from definition, enabling users to use a set of schedule primitives to convert a PyTorch model for common model training optimizations without directly changing the model itself, and thus addresses the tension between training efficiency and usability.

Getting Started

• Installation
• Quick Start

Tutorials

• Optimize Attention Module on A Single Device
• Optimize MLP Module on Multi-Device
• Debugging with Print

Reference

• Python API
• Index