Skip to content

Traingen

Shared training utilities for design generator packages.

lightning

Optional Lightning helpers for train-ourselves packages.

cli

Shared LightningCLI entry point for train-ourselves packages.

lightning_cli_class 

lightning_cli_class() -> type[LightningCLI]

Return LightningCLI without importing Lightning at package import time.

Source code in lib/traingen/src/traingen/lightning/cli.py 
11
12
13
14
15
def lightning_cli_class() -> type[LightningCLI]:
    """Return LightningCLI without importing Lightning at package import time."""
    from lightning.pytorch.cli import LightningCLI

    return LightningCLI

main 

main(args: list[str] | None = None) -> None

Run a model-agnostic LightningCLI from YAML class_path entries.

Parameters:

Name Type Description Default
args list[str] | None

Optional CLI arguments for tests. When omitted, LightningCLI reads sys.argv.

 None

Raises:

Type Description
SystemExit

If LightningCLI argument parsing fails.
Source code in lib/traingen/src/traingen/lightning/cli.py 
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
def main(args: list[str] | None = None) -> None:
    """Run a model-agnostic LightningCLI from YAML ``class_path`` entries.

    Args:
        args: Optional CLI arguments for tests. When omitted, LightningCLI reads
            ``sys.argv``.

    Raises:
        SystemExit: If LightningCLI argument parsing fails.
    """
    lightning_cli_class()(
        model_class=None,
        datamodule_class=None,
        subclass_mode_model=True,
        subclass_mode_data=True,
        args=args,
    )