Scaffolding a New Project

The noether-init command scaffolds a minimal, ready-to-train Noether project. It creates all required Python modules, Hydra configuration files, schemas, trainers, and callbacks, giving you a working starting point that you can adapt to your own use case.

Example Usage

uv run noether-init my_project

This creates a my_project/ directory. After completion, noether-init prints a summary of the configuration and the noether-train command to start training.

Arguments

Option	Values	Default
project_name (required)	Positional argument. Must be a valid Python identifier (no hyphens).
--tracker, -t	wandb, trackio, tensorboard, disabled	disabled
--hardware	gpu, mps, cpu	gpu
--project-dir, -d	Parent directory for the project folder	current directory
--wandb-entity	W&B entity name (only with --tracker wandb)	your W&B username

Generated Project Structure

The generated project contains:

my_project/
├── pyproject.toml            # Project config with emmiai-noether dependency
└── my_project/               # Python package
    ├── __init__.py
    ├── callbacks/             # Training callbacks
    ├── configs/
    │   ├── tracker/           # Tracker configs (wandb, disabled, etc.)
    │   └── base_experiment.yaml  # Main training config
    ├── datasets/              # Dataset implementation
    ├── models/                # Model implementation
    ├── pipelines/             # Data pipeline
    ├── schemas/               # Configuration dataclasses
    └── trainer/               # Training loop implementation

Running Training

After scaffolding, start training with:

cd my_project
uv run noether-train --hp my_project/configs/base_experiment.yaml