An Open-source Tool for Training
Deep Learning Models in the Cloud

Spotty makes training of deep learning models on AWS Spot Instances and on GCP Preemtible VMs as simple as training on your local machine.

Get started now View it on GitHub


Getting Started

Installation

Requirements:

Use pip to install or upgrade Spotty:

pip install -U spotty

Prepare a configuration file

Prepare a spotty.yaml file and put it to the root directory of your project:

  • See the file specification here.
  • Read this article for a real-world example.

Start an instance

Use the following command to launch an instance with the Docker container:

spotty start

It will start a Spot instance, restore snapshots if any, synchronize the project with the running instance and start the Docker container with the project environment.

Train your models or run notebooks

To connect to the running container via SSH, use the following command:

spotty ssh

It runs a tmux session, so you can always detach this session using Ctrl + b, then d combination of keys. To be attached to that session later, just use the spotty ssh command again.

Also, you can run custom scripts inside the Docker container using the spotty run <SCRIPT_NAME> command. Read more about custom scripts in the documentation: Configuration File: “scripts” section.