Documentation
Getting Started

Getting Started

Quick Start / moai Setup

🗿 moai is a PyTorch-based AI Model Development Kit (MDK) that aims to improve data-driven model workflows, design and understanding.

Steps

Prerequisites

The following python packages are required for using the supported features and the underlying model development kit:

  •  Python 3.10
  •  PyTorch 2.2.0 (cuda version)
  •  Cuda 11.4

Install moai

Clone the master branch from moai repository and install it by opening a command line on the source directory and running:

pip install -e .

Validate moai installation

To validate that the moai mdk has been successfully installed and there is no dependency missing, run the following MNIST example from the (outer) moai directory:

Train a simple multi-layer perceptron (MLP) using the MNIST data:

# train
python -m moai run fit moai/conf/examples/MNIST/main.yaml +DATA_ROOT=path/to/download/and/save/MNIST

Test the trained by loading its checkpoint:

# train
python -m moai run test moai/conf/examples/MNIST/main.yaml +DATA_ROOT=path/to/download/and/save/MNIST +CKPT=path/to/trained/model/checkpoint

Next

Note that moai operates as the backbone of the developed tools and plugins, which can be explored in the following sections:

GANimator Rerun Animation
Last updated on
DocumentationGANimator