agedi.cli.train

Attributes

_VALID_NOISERS
_DEFAULT_NOISER

Functions

train(→ None)	Train an AGeDi diffusion model from the command line.
_parse_override_value(raw)	Coerce a CLI override string to int, float, bool, or str.

Module Contents

agedi.cli.train._VALID_NOISERS

agedi.cli.train._DEFAULT_NOISER = 'CellPositions'

agedi.cli.train.train(**params) → None

Train an AGeDi diffusion model from the command line.

INPUT can be a trajectory file or a YAML configuration file.

 Trajectory mode — build the model from CLI options:

agedi train training_data.traj –noisers ConfinedCellPositions,Types

 Config mode — load all settings from a YAML file:

agedi train my_train.yaml

In config mode, optional KEY=VALUE pairs can be appended to override individual config entries without editing the file:



agedi train my_train.yaml feature_size=128 epochs=200

A ready-to-edit YAML template is available at:



python -c “import agedi, pathlib; print(pathlib.Path(agedi.__file__).parent / ‘conf’ / ‘train.yaml’)”

agedi.cli.train._parse_override_value(raw: str)

Coerce a CLI override string to int, float, bool, or str.