Skip to content

PyTorch Native Training

This tutorial guides you through building native training pipelines on Trainium using PyTorch XLA (torch-neuronx).

Overview

PyTorch XLA enables running PyTorch code on Neuron devices with minimal modifications.

Prerequisites

  • trn1.2xlarge or larger instance
  • Neuron SDK 2.18 or higher
  • PyTorch 2.1 or higher

Step 1: Basic Training Loop

Coming Soon

Detailed step-by-step guide will be added soon.

References