A blog post with details of how to get DiffusionGemma running on RunPod can be found here
Create a pod in the RunPod UI with an SSH public key and exposed SSH port. Once the pod is ready, copy its public IP address and TCP port, then run:
./deploy_runpod.py <IP> <PORT>The command:
- adds or updates
Host runpod-serverin~/.ssh/config(saving the prior config as~/.ssh/config.bak), - checks the SSH connection,
- uploads the project into
/workspace/diffusion, excluding.git,uv.lock, the local.venv, and generated Python cache files, and
Dependency installation and Jupyter kernel configuration are left for you to perform manually. Re-running the command updates the same remote folder without deleting remote-only files.
Install the latest version of uv
Create a venv with this: uv venv --python /usr/bin/python --system-site-packages
uv run python -m ipykernel install --user --name diffusion --display-name "Python (diffusion)"
Then you can access the jupyter notebook link in runpod and select the correct kernel.