Running the Notebooks
Every notebook on this site can be read right here — all code, figures, and explanations are rendered as web pages. But if you want to run the code yourself, edit it, or experiment, you have several free options. Pick whichever fits your situation.
Quick comparison
| Platform | Account needed | GPU access | Best for |
|---|---|---|---|
| View on GitHub | None | — | Reading the raw notebook without the Quarto formatting |
| Google Colab | ✔ Free tier | Most users — especially OSU students (institutional Google accounts get enhanced resources) | |
| Binder | None | ✗ | Quick exploration without creating any account |
| Kaggle | Kaggle (free) | ✔ 30 h/week | ML practitioners who already use Kaggle |
| GitHub Codespaces | GitHub | ✗ | Full VS Code dev environment in the browser; students get free hours via GitHub Education |
Details
Google Colab (recommended for most users)
Click the Open in Colab badge on any notebook page. Colab fetches a fresh copy from this repository and opens it in your own session — you can run and edit freely without affecting the original. Save your work to your Google Drive.
OSU students: Your institutional Google account (buckeyemail.osu.edu) gives you enhanced Colab features — more RAM, longer runtimes, and priority GPU access compared to free personal accounts.
GPU: Needed for the CNN lectures and labs (Weeks 7–10). In Colab, go to Runtime → Change runtime type → T4 GPU.
Binder
Click the Launch Binder badge. No account needed — Binder builds a temporary JupyterLab session with all dependencies pre-installed. The first launch takes 30 seconds to a few minutes while it builds the environment; subsequent launches are faster.
⚠ Sessions are temporary — your changes are lost when you close the browser tab. Download any work you want to keep.
Kaggle Notebooks
Click the Open in Kaggle badge. Kaggle provides free GPU access (30 hours per week) and has a large community of ML practitioners. You’ll need a free Kaggle account.
GitHub Codespaces
Click the Open in Codespaces badge. This launches a full VS Code editor in your browser with Jupyter support, all packages pre-installed, and the complete repository available.
Students: Sign up for GitHub Education with your .edu email to get 60+ free Codespace hours per month.
Run locally
Clone the repository and install the dependencies:
git clone https://github.com/buckai-observatory/edu-ai-earth-sciences.git
cd edu-ai-earth-sciences
pip install -r requirements.txt
jupyter lab