utils

dasherize_class_name(instance)

Dasherizes class name of passed instance. E.g., 'LightningBoost' -> 'lightning-boost'

Parameters:

Name	Type	Description	Default
instance	Any	Instance, whose class name is to be dasherized.	required

Returns:

Name	Type	Description
str	str	Dasherized class name.

Source code in lightning_boost/systems/utils.py

def dasherize_class_name(instance: Any) -> str:
    """
    Dasherizes class name of passed instance.
    E.g., 'LightningBoost' -> 'lightning-boost'

    Args:
        instance (Any): Instance, whose class name is to be dasherized.

    Returns:
        str: Dasherized class name.
    """

    return inflection.dasherize(inflection.underscore(type(instance).__name__))

get_loss_dict(loss)

Returns dictionary of (task, (name, loss)) pairs for a potential list of loss functions.

Parameters:

Name	Type	Description	Default
loss	List[TaskLoss] | TaskLoss	Loss function(s).	required

Returns:

Type	Description
ModuleDict	ModuleDict[str, ModuleDict[str, Module]]: Task, name, loss function.

Source code in lightning_boost/systems/utils.py

def get_loss_dict(loss: List[TaskLoss] | TaskLoss) -> ModuleDict:
    """
    Returns dictionary of (task, (name, loss)) pairs for a potential list of loss functions.

    Args:
        loss (List[TaskLoss] | TaskLoss): Loss function(s).

    Returns:
        ModuleDict[str, ModuleDict[str, Module]]: Task, name, loss function.
    """

    if isinstance(loss, TaskLoss):
        loss = [loss]

    loss_dict = ModuleDict()
    for task_loss in loss:
        if task_loss.task not in loss_dict:
            loss_dict[task_loss.task] = ModuleDict()
        loss_dict[task_loss.task][dasherize_class_name(task_loss.instance)] = task_loss  # type: ignore

    return loss_dict

get_metrics_dict(metrics)

Returns dictionary of (task, (name, metric)) pairs for a list of metrics.

Parameters:

Name	Type	Description	Default
metrics	List[TaskMetric] | TaskMetric | None	Metric(s).	required

Returns:

Type	Description
ModuleDict	ModuleDict[str, ModuleDict[str, Metric]]: Task, name, metric.

Source code in lightning_boost/systems/utils.py

def get_metrics_dict(
    metrics: List[TaskMetric] | TaskMetric | None,
) -> ModuleDict:
    """
    Returns dictionary of (task, (name, metric)) pairs for a list of metrics.

    Args:
        metrics (List[TaskMetric] | TaskMetric | None): Metric(s).

    Returns:
        ModuleDict[str, ModuleDict[str, Metric]]: Task, name, metric.
    """

    if metrics is None:
        return ModuleDict({'base-task': ModuleDict()})
    else:
        if isinstance(metrics, TaskMetric):
            metrics = [metrics]
        metrics_dict = ModuleDict()
        for task_metric in metrics:
            if task_metric.task not in metrics_dict:
                metrics_dict[task_metric.task] = ModuleDict()
            metrics_dict[task_metric.task][dasherize_class_name(task_metric.instance)] = task_metric  # type: ignore

        return metrics_dict

get_models_dict(models)

Returns dictionary of (name, model) pairs for a list of DL models.

Parameters:

Name	Type	Description	Default
models	List[BaseModel] | BaseModel	DL model(s).	required

Returns:

Type	Description
ModuleDict	ModuleDict[str, Metric]: Task, DL model.

Source code in lightning_boost/systems/utils.py

def get_models_dict(
    models: List[BaseModel] | BaseModel,
) -> ModuleDict:
    """
    Returns dictionary of (name, model) pairs for a list of DL models.

    Args:
        models (List[BaseModel] | BaseModel): DL model(s).

    Returns:
        ModuleDict[str, Metric]: Task, DL model.
    """

    if isinstance(models, BaseModel):
        models = [models]

    return ModuleDict(
        {
            dasherize_class_name(model) if model.name is None else model.name: model
            for model in models
        }
    )