scimba_torch.neural_nets.coordinates_based_nets.scimba_module

Base module for Scimba neural networks.

Classes

ScimbaModule(in_size, out_size, **kwargs)

Abstract class for Scimba neural networks.
class ScimbaModule(in_size, out_size, **kwargs)[source]

Bases: Module

Abstract class for Scimba neural networks.

Parameters:

  • in_size (int) – Input dimension.

  • out_size (int) – Output dimension.

  • **kwargs – Additional keyword arguments.

in_size

Input dimension.

out_size

Output dimension.

output_layer

Output layer module (to be set by subclasses).

parameters(flag_scope='all', flag_format='list')[source]

Get parameters of the neural net.

Parameters:

  • flag_scope (str) – Specifies which parameters to return. Options: ‘all’, ‘last_layer’, ‘except_last_layer’.

  • flag_format (str) – Specifies the format Options: ‘list’, ‘tensor’.

Returns:

list[nn.Parameter] or torch.Tensor

Raises:

ValueError – If flag_scope is not one of the supported options.

set_parameters(new_params, flag_scope='all')[source]

Set parameters.

Parameters:

  • new_params (Tensor) – new parameters.

  • flag_scope (str) – ‘all’, ‘last_layer’, ‘except_last_layer’

Raises:

ValueError – If flag_scope is not one of the supported options.