Model Performance Optimization Guide

This guide provides multiple methods that can be implemented in order to achieve the best performance when running training and inference on PyTorch models with the Intel® Gaudi® AI accelerator. Start with the Optimization Checklist to review the three main steps for model optimization.

• Model Optimization Checklist

• General Model Optimizations

• Inference Optimizations

• Handling Dynamic Shapes

• Fused Optimizers and Custom Ops for Intel Gaudi

• HPU Graphs for Training

• Optimizing Training Platform

• Parallel Compilation