Model Class Reference

Detailed Description

A thin wrapper around neml2::Model

Public Member Functions
dict[str, Model]	dependency (self)
TensorValue	get_parameter (self, str arg0)
tensors.LabeledAxis	input_axis (self)
tensors.TensorType	input_type (self, tensors.LabeledAxisAccessor variable)
dict[str, TensorValue]	named_buffers (self)
dict[str, TensorValue]	named_parameters (self)
dict[str, Model]	named_submodels (self)
tensors.LabeledAxis	output_axis (self)
tensors.TensorType	output_type (self, tensors.LabeledAxisAccessor variable)
None	reinit (self, tuple[int,...] batch_shape=(), int deriv_order=0, torch.device device=..., torch.dtype dtype=...)
None	set_parameter (self, str arg0, tensors.Tensor arg1)
None	set_parameters (self, dict[str, tensors.Tensor] arg0)
typing.Any	value (self, typing.Any arg0)
tuple	value_and_dvalue (self, typing.Any arg0)
bool	is_AD_enabled (self)
str	name (self)
str	type (self)

Member Function Documentation

dependency()

dict[str, Model] dependency

(

self

)

Get the dictionary describing this model's dependency information, if any.

get_parameter()

TensorValue get_parameter	(		self,
		str	arg0 )

Get a model parameter given its name

input_axis()

tensors.LabeledAxis input_axis

(

self

)

Input axis of the model. The axis contains information on variable names and their associated slicing indices.

input_type()

tensors.TensorType input_type	(		self,
		tensors.LabeledAxisAccessor	variable )

Introspect the underlying tensor type of an input variable. @returns tensors.TensorType

is_AD_enabled()

bool is_AD_enabled

(

self

)

Whether automatic differentiation is enabled for this model. This property cannot be modified once the model is created. Use the `enable_AD` option of base.load_model or base.get_model to control this property.

name()

str name

(

self

)

Name of the model

named_buffers()

dict[str, TensorValue] named_buffers

(

self

)

Get the model buffers. The keys of the returned dictionary are the buffers' names.

named_parameters()

dict[str, TensorValue] named_parameters

(

self

)

Get the model parameters. The keys of the returned dictionary are the parameters' names.

named_submodels()

dict[str, Model] named_submodels

(

self

)

Get the sub-models registered to this model

output_axis()

tensors.LabeledAxis output_axis

(

self

)

Input axis of the model. The axis contains information on variable names and their associated slicing indices.

output_type()

tensors.TensorType output_type	(		self,
		tensors.LabeledAxisAccessor	variable )

Introspect the underlying tensor type of an output variable. @returns tensors.TensorType

reinit()

None reinit	(		self,
		tuple[int, ...]	batch_shape = (),
		int	deriv_order = 0,
		torch.device	device = ...,
		torch.dtype	dtype = ... )

(Re)initialize the model with given batch shape, derivative order, device, and dtype.

:param batch_shape: Batch shape used to allocate input, output, and derivative storage
:param deriv_order: An integer ranging from 0-2. When set to 0, only the output storage
    will be allocated; when set to 1, both the output and the first derivative storage
    are allocated; when set to 2, the second derivative storage is additionally allocated.
:param device:      Device on which the model will be evaluated. All parameters, buffers,
    and custom data are synced to the given device.
:param dtype:       Floating point scalar type used throughout the model.

set_parameter()

None set_parameter	(		self,
		str	arg0,
		tensors.Tensor	arg1 )

Set the value for a model parameter

set_parameters()

None set_parameters	(		self,
		dict[str, tensors.Tensor]	arg0 )

Set the values for multiple model parameters

type()

str type

(

self

)

Type of the model

value()

typing.Any value	(		self,
		typing.Any	arg0 )

Evaluate the model with given input and return the output. Note that the input can either be of type torch.Tensor, tensors.Tensor, or tensors.LabeledVector. The returned output will be of the covariant type of the input.

value_and_dvalue()

tuple value_and_dvalue	(		self,
		typing.Any	arg0 )

Evaluate the model with given input and return the output as well the first derivative. Note that the input can either be of type torch.Tensor, tensors.Tensor, or tensors.LabeledVector. The returned output and derivative will be of the covariant type of the input.