Installation

To participate in this challenge, you will need to run your images within an Anaconda environment. As part of your submission, you must export the environment configuration and upload it.

However, please note:

The submission environment will be created using Micromamba (due to performance reasons).
If you choose to install only one of the tools, it should still work fine for exporting the necessary dependencies, as they share almost all commands.

Why Anaconda and Micromamba?

Anaconda is the most comprehensive and robust tool for creating and managing environments.
Micromamba is, for the most part, fully compatible with Anaconda env files, ensuring smooth transitions.

💡 Recommendation

Build and test your environment with Anaconda first.
Before uploading, confirm that Micromamba can recreate that environment locally from your exported environment.yml.

Checking this in advance avoids dependency surprises when your submission is rebuilt and run on our servers.

Additionally, you can recreate the container where your submission will be executed using Docker. The prebuilt container is available on Docker Hub and can be pulled with the following command:

docker pull arclabmit/ai-challenges:micromamba-submission

This container includes NVIDIA drivers, Micromamba, and essential scientific libraries such as numpy and pandas.

Install Anaconda 🐍

Follow the Anaconda installation guide here. The devkit is tested for Python 3.9+ on Ubuntu and macOS.

ℹ️ Info

Anaconda ships with its own Python interpreter. The evaluation container runs Python 3.9.21, so target Python 3.9+ (and avoid older releases) to prevent version mismatches at evaluation time.

Install Micromamba 𓆙

Refer to the Micromamba installation guide here.

Install Docker 🐋

If you want to test your submission in the same environment it will be executed in, you need to install Docker.

1. Docker Prerequisites

Windows: WSL2
1. Open PowerShell as Administrator and run:
```
wsl --install
```
2. Restart your computer when prompted.
3. Install a Linux distro from Microsoft Store (e.g., Ubuntu).
4. Set WSL 2 as the default:
```
wsl --set-default-version 2
```
5. Verify WSL installation:
```
wsl --list --verbose
```
macOS: Install Homebrew
Linux: Update and upgrade system packages:
```
sudo apt update && sudo apt upgrade
```

2. Download Docker Installer

Windows: Download Docker Desktop from the official site.
macOS: Install using Homebrew:
```
brew install --cask docker
```
Linux: Install Docker with:
```
sudo apt install docker.io
```

3. Run the Installer

Windows: Double-click the .exe file and follow the instructions.
macOS: Follow the Homebrew prompts.

Linux: Start Docker service and enable it on boot:

sudo systemctl start docker && sudo systemctl enable docker

4. Verify Installation

Run the following commands to ensure Docker is installed correctly:

docker --version
docker run hello-world

[Optional] Install NVIDIA Container Toolkit

If your submission requires GPU acceleration and you want to test it locally, install the NVIDIA container toolkit.

Download the Devkit

Download the devkit using the terminal and move into the newly created folder:

cd && git clone https://github.com/ARCLab-MIT/STORM-AI-devkit-2025.git && cd STORM-AI-devkit-2025

This command will clone the repository into your home directory. While you can place it in another directory, the rest of our tutorials assume you are working from your home directory.