miniml.nn.compose

Identity

Bases: MiniMLModel

A MiniML model that returns its input unchanged or scaled.

__init__(scale=1.0)

Initialize the Identity model.

Parameters:

Name	Type	Description	Default
scale	float	The scale factor to apply to the input. Defaults to 1.0.	1.0

Take

Bases: MiniMLModel

A MiniML model that takes specific indices from the input array along a given axis.

__init__(indices, axis=-1)

Initialize the Take model.

Parameters:

Name	Type	Description	Default
indices	list[int] | int	The indices to take from the input array.	required
axis	int	The axis along which to take the indices. Defaults to -1.	-1

Stack

Bases: MiniMLModel

A MiniML model that stacks multiple MiniML models sequentially.

__init__(models, loss=None)

Initialize the Stack model.

Parameters:

Name	Type	Description	Default
models	list[MiniMLModel]	The models to stack.	required
loss	LossFunction	The loss function to use. Defaults to None.	None

Raises:

Type	Description
ValueError	If the model list is empty.

Parallel

Bases: MiniMLModel

A MiniML model that applies multiple MiniML models in parallel and sums or concatenates their outputs.

__init__(models, mode, concat_axis=-1, loss=None)

Initialize the Parallel model.

Parameters:

Name	Type	Description	Default
models	list[MiniMLModel]	The models to apply in parallel.	required
mode	Literal['sum', 'concat']	The mode of combining outputs.	required
concat_axis	int	The axis to concatenate along if mode is "concat". Defaults to -1.	-1
loss	LossFunction | None	The loss function to use. Defaults to None.	None

Raises:

Type	Description
ValueError	If models list is empty or mode is invalid.
ValueError	If mode is not a valid option.