layer_torch_module_wrapper {keras3}

Torch module wrapper layer.

Description

layer_torch_module_wrapper is a wrapper class that can turn any torch.nn.Module into a Keras layer, in particular by making its parameters trackable by Keras.

layer_torch_module_wrapper() is only compatible with the PyTorch backend and cannot be used with the TensorFlow or JAX backends.

Usage

layer_torch_module_wrapper(object, module, name = NULL, ...)

Arguments

object

Object to compose the layer with. A tensor, array, or sequential model.

module

torch.nn.Module instance. If it's a LazyModule instance, then its parameters must be initialized before passing the instance to layer_torch_module_wrapper (e.g. by calling it once).

name

The name of the layer (string).

...

For forward/backward compatability.

Value

The return value depends on the value provided for the first argument. If object is:

• a keras_model_sequential(), then the layer is added to the sequential model (which is modified in place). To enable piping, the sequential model is also returned, invisibly.

• a keras_input(), then the output tensor from calling layer(input) is returned.

• NULL or missing, then a Layer instance is returned.

Example

Here's an example of how the layer_torch_module_wrapper() can be used with vanilla PyTorch modules.

# reticulate::py_install(
#   packages = c("torch", "torchvision", "torchaudio"),
#   envname = "r-keras",
#   pip_options = c("--index-url https://download.pytorch.org/whl/cpu")
# )
library(keras3)
use_backend("torch")
torch <- reticulate::import("torch")
nn <- reticulate::import("torch.nn")
nnf <- reticulate::import("torch.nn.functional")

Classifier(keras$Model) \%py_class\% {
  initialize <- function(...) {
    super$initialize(...)

    self$conv1 <- layer_torch_module_wrapper(module = nn$Conv2d(
      in_channels = 1L,
      out_channels = 32L,
      kernel_size = tuple(3L, 3L)
    ))
    self$conv2 <- layer_torch_module_wrapper(module = nn$Conv2d(
      in_channels = 32L,
      out_channels = 64L,
      kernel_size = tuple(3L, 3L)
    ))
    self$pool <- nn$MaxPool2d(kernel_size = tuple(2L, 2L))
    self$flatten <- nn$Flatten()
    self$dropout <- nn$Dropout(p = 0.5)
    self$fc <-
      layer_torch_module_wrapper(module = nn$Linear(1600L, 10L))
  }

  call <- function(inputs) {
    x <- nnf$relu(self$conv1(inputs))
    x <- self$pool(x)
    x <- nnf$relu(self$conv2(x))
    x <- self$pool(x)
    x <- self$flatten(x)
    x <- self$dropout(x)
    x <- self$fc(x)
    nnf$softmax(x, dim = 1L)
  }
}
model <- Classifier()
model$build(shape(1, 28, 28))
cat("Output shape:", format(shape(model(torch$ones(1L, 1L, 28L, 28L)))))

model |> compile(loss = "sparse_categorical_crossentropy",
                 optimizer = "adam",
                 metrics = "accuracy")

model |> fit(train_loader, epochs = 5)

See Also

Other wrapping layers:
layer_flax_module_wrapper()
layer_jax_model_wrapper()

Other layers:
Layer()
layer_activation()
layer_activation_elu()
layer_activation_leaky_relu()
layer_activation_parametric_relu()
layer_activation_relu()
layer_activation_softmax()
layer_activity_regularization()
layer_add()
layer_additive_attention()
layer_alpha_dropout()
layer_attention()
layer_average()
layer_average_pooling_1d()
layer_average_pooling_2d()
layer_average_pooling_3d()
layer_batch_normalization()
layer_bidirectional()
layer_category_encoding()
layer_center_crop()
layer_concatenate()
layer_conv_1d()
layer_conv_1d_transpose()
layer_conv_2d()
layer_conv_2d_transpose()
layer_conv_3d()
layer_conv_3d_transpose()
layer_conv_lstm_1d()
layer_conv_lstm_2d()
layer_conv_lstm_3d()
layer_cropping_1d()
layer_cropping_2d()
layer_cropping_3d()
layer_dense()
layer_depthwise_conv_1d()
layer_depthwise_conv_2d()
layer_discretization()
layer_dot()
layer_dropout()
layer_einsum_dense()
layer_embedding()
layer_feature_space()
layer_flatten()
layer_flax_module_wrapper()
layer_gaussian_dropout()
layer_gaussian_noise()
layer_global_average_pooling_1d()
layer_global_average_pooling_2d()
layer_global_average_pooling_3d()
layer_global_max_pooling_1d()
layer_global_max_pooling_2d()
layer_global_max_pooling_3d()
layer_group_normalization()
layer_group_query_attention()
layer_gru()
layer_hashed_crossing()
layer_hashing()
layer_identity()
layer_integer_lookup()
layer_jax_model_wrapper()
layer_lambda()
layer_layer_normalization()
layer_lstm()
layer_masking()
layer_max_pooling_1d()
layer_max_pooling_2d()
layer_max_pooling_3d()
layer_maximum()
layer_mel_spectrogram()
layer_minimum()
layer_multi_head_attention()
layer_multiply()
layer_normalization()
layer_permute()
layer_random_brightness()
layer_random_contrast()
layer_random_crop()
layer_random_flip()
layer_random_rotation()
layer_random_translation()
layer_random_zoom()
layer_repeat_vector()
layer_rescaling()
layer_reshape()
layer_resizing()
layer_rnn()
layer_separable_conv_1d()
layer_separable_conv_2d()
layer_simple_rnn()
layer_spatial_dropout_1d()
layer_spatial_dropout_2d()
layer_spatial_dropout_3d()
layer_spectral_normalization()
layer_string_lookup()
layer_subtract()
layer_text_vectorization()
layer_tfsm()
layer_time_distributed()
layer_unit_normalization()
layer_upsampling_1d()
layer_upsampling_2d()
layer_upsampling_3d()
layer_zero_padding_1d()
layer_zero_padding_2d()
layer_zero_padding_3d()
rnn_cell_gru()
rnn_cell_lstm()
rnn_cell_simple()
rnn_cells_stack()