Quick Start¶

Training a Model¶

Training is configured via YAML files. To launch a training run with the default Samudra configuration:

uv run -m ocean_emulators.train configs/samudra_om4/train.yaml

The FOMO model supports multi-scale training across different resolutions:

uv run -m ocean_emulators.train configs/fomo_om4/train_multiscale.yaml

Data Paths¶

Training configs reference OM4 ocean model data stored in Zarr format. Update the data paths in your config to point to your data location:

# configs/data/om4.yaml
data:
  path: "s3://<your-bucket>/path/to/OM4.zarr"  # Update with your data path

See configs/data/ for example data configurations at 1°, 1/2°, and 1/4° resolutions.

Evaluation¶

Run a long autoregressive rollout against ground-truth data:

uv run -m ocean_emulators.eval configs/samudra_om4/eval.yaml

This produces metrics (RMSE, bias, anomaly correlation) and writes predicted fields to a Zarr output file.

Visualization¶

Generate maps, time series, and probability density plots from evaluation outputs:

uv run -m ocean_emulators.viz configs/samudra_om4/viz.yaml

Configuration¶

All commands accept --help for available options:

uv run -m ocean_emulators.train --help
uv run -m ocean_emulators.eval --help

You can override any config key from the command line:

uv run -m ocean_emulators.train configs/samudra_om4/train.yaml --epochs 100 --lr 1e-4

See Configuration for details on the configuration system.