Skip to content

README.md

Latest commit

 

History

History
52 lines (39 loc) · 1.07 KB

README.md

File metadata and controls

52 lines (39 loc) · 1.07 KB
Raw

Machine Learning Example

Sandbox configuration for ML development with PyTorch, TensorFlow, or JAX.

Setup

cd your-ml-project
cp /path/to/watermelon/docs/examples/python-ml/.watermelon.toml ./
watermelon run

Inside the sandbox

python -m venv venv
source venv/bin/activate
pip install torch transformers jupyter
jupyter notebook --ip=0.0.0.0 --port=8888
# Visit http://localhost:8888 on your host

Resource Allocation

ML workloads need more resources. Adjust based on your host capacity:

[resources]
memory = "32GB"  # For larger models
cpus = 8
disk = "100GB"   # For datasets

Model Downloads

The example allowlist includes Hugging Face Hub. Add only the other model sources your project needs:

[network]
allow = [
    "pypi.org",
    "files.pythonhosted.org",
    "huggingface.co",
    "*.huggingface.co",
    "download.pytorch.org",        # PyTorch model zoo
    "storage.googleapis.com",      # TensorFlow Hub
]

GPU Support

GPU passthrough requires additional Lima configuration. See Lima documentation for GPU setup.