Home » Get Started » Get started with Intel® Gaudi® AI accelerators

Get started with Intel® Gaudi® AI accelerators

Here are the resources to help you begin running models on Intel® Gaudi®  processors. Assuming your environment is configured and ready for model training or inference, explore the following materials. For any additional system configuration requirements, refer to the Setup and Install page to get started.

For those looking to move straight to the Quick Start Guide for simple steps or if you’re using Intel Gaudi for the first time, select the Quick Start and Customer Onboarding option below.​

This is the fastest way to start running on Intel® Gaudi® accelerators:

  1. Get Access to an Intel Gaudi accelerator instance​
  2. Load a PyTorch Docker Image that contains Intel Gaudi software and PyTorch Framework​
  3. Load a Set of models from GitHub; you can select the options below: ​

For first time users, follow the Quick Start Guide and Customer Onboarding below:

This is the fastest way to start running on Intel Gaudi: ​

  1. Get Access to an Intel Gaudi accelerator instance​
    ​
  2. Load a PyTorch Docker Image that contains Intel Gaudi software and PyTorch Framework​
    ​
  3. Load a Set of models from GitHub; you can select the options on the right: ​

For first time users, follow the Quick Start Guide and Customer Onboarding below

These are the options for running models on Intel Gaudi processors​:

  1. Hugging Face – we have partnered with Hugging Face and created the Optimum Habana Library; the interface between Hugging Face Transformers and Diffusers libraries and Intel Gaudi. This includes many task examples (language-modeling,  text-generation, etc..) and fully optimized Generative AI and LLM models for use with full documentation.​
    ​
  2. GPU Migration Toolkit: For those that want to bring their own custom models from GPU or from the public domain can use the GPU Migration Toolkit to convert a GPU based model to run on Intel Gaudi processors.​
    ​
  3. Intel Gaudi Model References: This is a set of fully optimized and documented models and simple examples that allow a user to quickly run NLP, Generative AI and Computer vision models;​
    ​
  4. Intel Gaudi Tutorials: Jupyter Notebooks providing easy to use examples for both training and Inference

Quick Start Guide and Customer Onboarding

If you’re a first time user, start here and we’ll walk you thru all the steps to get started using Intel Gaudi with dedicated tutorials and videos.

Learn more

Using the GPU Migration Toolkit

Migrate your GPU based models to run on Intel Gaudi with minimal changes.

Learn more

Using Hugging Face On Intel Gaudi

How to run Hugging Face transformer and Diffusion models on Intel Gaudi.

Learn more

Using Model References and Examples

Use Intel Gaudi’s model references  and examples to get started.

Learn more

Generative AI & Large Language Models

Learn more about training and inference of large scale and foundation models on Intel Gaudi.

Learn more

Tutorials

Step-by-step tutorials that walk you through creating and training your models.

Learn more
Sign up for the latest Habana developer news, events, training, and updates.
Sign up

Docs

Installation Guide

Detailed instructions for installation of the Habana Gaudi driver and SynapseAI software stack if you need to setup your environment.

PyTorch User Guide

Detailed description of how to run your model on PyTorch.

Distributed Training with PyTorch

Learn how to scale your PyTorch models training with Distributed Data Parallel APIs support and Gaudi NIC or Host NIC for system scaling.

Tutorials

PyTorch Tutorials

PyTorch Lightning Tutorials

GitHub

Habana Models

View repository of all Habana reference models and instructions on how to run and train your models, including TensorFlow and PyTorch.