OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Developer Tools & SDKs

Ai8X Training

Ai8X Training is an open-source codebase for developing and training deep-learning models intended for Analog Devices MAX78000 and MAX78002 edge AI devices. It works alongside a separate synthesis repository that converts trained models into C code.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Ai8X Training — OpenViST project profile in AI Developer Tools & SDKs.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Ai8X Training when the target is a MAX78000 or MAX78002 and you need the model-development and training stage. Consult the separate synthesis and MSDK documentation for subsequent conversion and embedded development stages.

Who it is for

  • Developers building deep-learning models for MAX78000 or MAX78002 devices.
  • Embedded AI teams planning a workflow from model training through model synthesis.

Documented uses

  • Develop and train models for the MAX78000 and MAX78002.
  • Prepare a trained model for conversion into C code through the separate ai8x-synthesis repository.
  • Prototype models for the supported edge-device family.
Before you choose

Requirements & limits

The repository is public and licensed under Apache-2.0.

  • The current branch requires PyTorch 2; the documentation identifies PyTorch 2.3 with Python 3.11.x as the officially supported versions.
  • Full training support is documented for amd64/x86_64 PCs running Ubuntu 20.04 or 22.04. Nvidia CUDA 12.1 or later is optional but highly recommended for acceleration.
  • Windows users can run Ubuntu through WSL2; native Windows is possible but discouraged due to compatibility and performance concerns. macOS supports MPS acceleration on Apple Silicon, while Intel Macs lack hardware accele
  • The installation instructions recommend at least 64 GB of free disk space, with datasets stored separately.
  • TensorFlow/Keras support is deprecated; the MSDK and ai8x-synthesis are separate components with distinct roles.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install ai8x-training and ai8x-synthesis on supported Ubuntu

Set up the training and synthesis repositories with Python 3.11.8 and the documented Ubuntu system dependencies.

Before you start

  • 64-bit amd64/x86_64 PC running Ubuntu Linux 20.04 or 22.04; WSL2 with either Ubuntu version is also described.
  • Python 3.11.8 or a later 3.11.x version; the develop branch supports PyTorch 2.3.
  • At least 64 GB of free disk space is recommended; datasets may require additional space.
  • Administrator privileges are needed to install system packages.

Steps

  1. Install the listed Ubuntu system packages using the README's apt-get install command.

  2. Install Python 3.11.8 with pyenv, following the README's Python setup instructions.

  3. From the project root, clone both repositories recursively: the command below and the command below.

    Documented command
    git clone --recursive https://github.com/analogdevicesinc/ai8x-training.git
    Documented command
    git clone --recursive https://github.com/analogdevicesinc/ai8x-synthesis.git

  4. Follow the README's virtual-environment instructions in ai8x-training to create the environment and install basic wheels; activate the appropriate project environment when switching between training and synthesis.

Expected outcome

The training and synthesis repositories are checked out and their documented environment setup is complete.

Guide limitations

  • The current branch requires PyTorch 2; the README identifies PyTorch 2.3 on Python 3.11.x as officially supported.
  • Official model-training platforms are Ubuntu 20.04 or 22.04 on amd64/x86_64. Other combinations may have limited support.
  • Nvidia CUDA 12.1 or later is optional and highly recommended for hardware acceleration; training without acceleration may not work well for more than simple models.
  • TensorFlow/Keras support is deprecated.
  • The README's installation guidance recommends 64 GB free disk space, but datasets can be many times that size and should be stored separately.
Documentation 1 ↗Documentation 2 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
12 October 2026
Original website title
analogdevicesinc/ai8x-training · Official homepage title · retained in its source language
How we review sourcesSuggest a correction