phlower.nn.Accessor

class phlower.nn.Accessor(nodes=None, activation='identity', index=0)

Bases: IPhlowerCoreModule, Module

Accessor is a neural network module that provides access to specific tensor data with optional activation.

This module is designed to access and process specific elements from input tensor collections. It can extract data at a specified index and apply an activation function to the result.

Parameters:

• nodes (list[int] | None (optional)) – List of feature dimension sizes (The last value of tensor shape). Defaults to None.

• activation (str (optional)) – Name of the activation function to apply to the output. Defaults to “identity” (no activation).

• index (int (optional)) – The index to access from the input tensor. Defaults to 0.

Examples

>>> accessor = Accessor(activation="relu", index=2)
>>> output = accessor(input_data)  # Applies activation to data[2]

Methods

forward(data, *[, field_data])	forward function which overloads torch.nn.Module
from_setting(setting)	Create Accessor from setting object
get_nn_name()	Return name of Accessor

Attributes

T_destination
call_super_init
dump_patches
training

forward(data, *, field_data=None, **kwards)

forward function which overloads torch.nn.Module

Parameters:

• data (IPhlowerTensorCollections) – IPhlowerTensorCollections data which receives from predecessors

• field_data (ISimulationField | None) – ISimulationField | None Constant information through training or prediction

Returns:

Tensor object

Return type:

PhlowerTensor

classmethod from_setting(setting)

Create Accessor from setting object

Parameters:

setting (AccessorSetting) – setting object

Returns:

Accessor

Return type:

Self

classmethod get_nn_name()

Return name of Accessor

Returns:

name

Return type:

str