Linux/MacOS Installation

Installation

Estimated time: 8 minutes

Prerequisites

  1. Docker version >18.09 (Docker installation instructions)
  2. Python >=3.6
  3. >5GB of free machine storage
  4. The atlas_installer.py file.

Steps:

  1. Create a new, empty directory where you will install Atlas.

  2. Copy the atlas_installer.py file into this directory.

  3. Create and activate a Python >=3.6 virtual environment using Conda or venv to minimize dependency issues.

  4. Run the install script with python atlas_installer.py and follow the instructions.

Tip

Running python atlas_installer.py --help will give you troubleshooting advice if the script isn't working as expected.

Tip

The longest part of the script is pulling the Atlas docker images, if the script fails at this point, you can re-run it using python atlas_installer.py -dp to skip over the download and unpacking and go directly to the image pull.


Start Up

After completing the installation section, you can do the following to start Atlas:

  1. Validate that you are in the same Python environment that was used to run the installation script.
  2. Run atlas-server start.

GPU Mode

If you installed Atlas with GPU support, you can start the Atlas server in GPU mode by running atlas-server start -g. This will allow Atlas to use all CUDA-enabled GPUs on your system.

Success

Validate that the GUI is running by going to the GUI. This is your centralized location to track all of your experiments.


Hello Atlas

After completing the start-up section, follow the next few steps to launch your first Atlas job:

  1. Navigate to where you'd like to create your Atlas project directory.
  2. Ensure that you are in the environment that was used during installation.
  3. Run foundations init hello-atlas to create an example project.
  4. Navigate into the newly created hello-atlas directory.
  5. Run the sample code provided by running python main.py.
  6. Head to the GUI to see your experiment!

Note

When you run a job for the first time, it will download the appropriate worker image needed.
This will take roughly 1.5 minutes.