botiverse.models.GRUClassifier package#

Submodules#

botiverse.models.GRUClassifier.GRUClassifier module#

class botiverse.models.GRUClassifier.GRUClassifier.BasicGRU(input_size, dropout_p=0.1)[source]#

Bases: Module

An interface for the basic GRU unit

Constructs a BasicGRU instance with specific layer sizes and dropout probability.

Parameters:

  • input_size (int) – The size of the input to the model.

  • dropout_p (float) – A regularization parameter.

Returns:

None

forward(input, hidden)[source]#

Defines the computation performed by the model.

Parameters:

  • input (Tensor) – The provided sequance input.

  • hidden (Tensor) – The provided hidden state.

Returns:

New hidden state.

Return type:

Tensor

initHidden(batch_size)[source]#

Creates a tensor of zeros for the hidden state initialization.

Parameters:

batch_size (int) – The size of the batch for which the hidden state is to be initialized.

Returns:

Tensor of zeros of the shape (batch_size, 1, hidden_size).

Return type:

Tensor

training: bool#
class botiverse.models.GRUClassifier.GRUClassifier.GRUTextClassifier(vocabulary, embedding_size, output_size, dropout_p=0.1)[source]#

Bases: Module

An interface for the GRU text classifier which uses a basic GRU unit with a linear output layer and an input embedding layer

Constructs a GRUTextClassifier instance with specific hyperparameters.

Parameters:

  • vocabulary (int) – The size of vocabulary used in the Embedding layer.

  • embedding_size (int) – The size of each embedding vector.

  • output_size (int) – The size of the output from the model (number of classes).

  • dropout_p (float) – A regularization parameter.

Returns:

None

forward(input)[source]#

Defines the computation performed by the model.

Parameters:

input (Tensor) – The model input.

Returns:

Output after the forward pass (classes probabilities).

Return type:

Tensor

training: bool#

Module contents#