ResNetEmbedding2D

class ResNetEmbedding2D(c_in, c_out=20, c_hidden_fc=1000, n_stages=4, blocks_per_stage=None, c_stages=None, activation=<class 'torch.nn.modules.activation.ReLU'>, change_c_mode='conv', construct_mapping=<function construct_simple_conv_net>, construct_mapping_kwargs=None, residual_block_kwargs=None)

Bases: Module

Residual neural network mapping image-like data to a fixed sized vector.

Parameters:

• c_in (int)

• c_out (int)

• c_hidden_fc (int)

• n_stages (int)

• blocks_per_stage (List | None)

• c_stages (List | None)

• activation (Type[Module])

• change_c_mode (str)

• construct_mapping (Callable)

• construct_mapping_kwargs (Dict | None)

• residual_block_kwargs (Dict | None)

forward(x)

Forward pass of the residual network.

Parameters:

x (Tensor) – Input tensor

Returns:

Output tensor

Return type:

z