Type GRU
Namespace tensorflow.keras.layers
Parent RNN
Interfaces IGRU
Gated Recurrent Unit - Cho et al. 2014. There are two variants. The default one is based on 1406.1078v3 and
has reset gate applied to hidden state before matrix multiplication. The
other one is based on original 1406.1078v1 and has the order reversed. The second variant is compatible with CuDNNGRU (GPU-only) and allows
inference on CPU. Thus it has separate biases for `kernel` and
`recurrent_kernel`. Use `'reset_after'=True` and
`recurrent_activation='sigmoid'`.
Methods
Properties
- activation
- activation_dyn
- activity_regularizer
- activity_regularizer_dyn
- bias_constraint
- bias_constraint_dyn
- bias_initializer
- bias_initializer_dyn
- bias_regularizer
- bias_regularizer_dyn
- built
- cell
- constants_spec
- dropout
- dropout_dyn
- dtype
- dtype_dyn
- dynamic
- dynamic_dyn
- go_backwards
- implementation
- implementation_dyn
- inbound_nodes
- inbound_nodes_dyn
- input
- input_dyn
- input_mask
- input_mask_dyn
- input_shape
- input_shape_dyn
- input_spec
- input_spec_dyn
- kernel_constraint
- kernel_constraint_dyn
- kernel_initializer
- kernel_initializer_dyn
- kernel_regularizer
- kernel_regularizer_dyn
- losses
- losses_dyn
- metrics
- metrics_dyn
- name
- name_dyn
- name_scope
- name_scope_dyn
- non_trainable_variables
- non_trainable_variables_dyn
- non_trainable_weights
- non_trainable_weights_dyn
- outbound_nodes
- outbound_nodes_dyn
- output
- output_dyn
- output_mask
- output_mask_dyn
- output_shape
- output_shape_dyn
- PythonObject
- recurrent_activation
- recurrent_activation_dyn
- recurrent_constraint
- recurrent_constraint_dyn
- recurrent_dropout
- recurrent_dropout_dyn
- recurrent_initializer
- recurrent_initializer_dyn
- recurrent_regularizer
- recurrent_regularizer_dyn
- reset_after
- reset_after_dyn
- return_sequences
- return_state
- state_spec
- stateful
- states
- states_dyn
- submodules
- submodules_dyn
- supports_masking
- time_major
- trainable
- trainable_dyn
- trainable_variables
- trainable_variables_dyn
- trainable_weights
- trainable_weights_dyn
- units
- units_dyn
- unroll
- updates
- updates_dyn
- use_bias
- use_bias_dyn
- variables
- variables_dyn
- weights
- weights_dyn
- zero_output_for_mask
Public static methods
GRU NewDyn(object units, ImplicitContainer<T> activation, ImplicitContainer<T> recurrent_activation, ImplicitContainer<T> use_bias, ImplicitContainer<T> kernel_initializer, ImplicitContainer<T> recurrent_initializer, ImplicitContainer<T> bias_initializer, object kernel_regularizer, object recurrent_regularizer, object bias_regularizer, object activity_regularizer, object kernel_constraint, object recurrent_constraint, object bias_constraint, ImplicitContainer<T> dropout, ImplicitContainer<T> recurrent_dropout, ImplicitContainer<T> implementation, ImplicitContainer<T> return_sequences, ImplicitContainer<T> return_state, ImplicitContainer<T> go_backwards, ImplicitContainer<T> stateful, ImplicitContainer<T> unroll, ImplicitContainer<T> reset_after, IDictionary<string, object> kwargs)
Public properties
object activation get;
object activation_dyn get;
PythonFunctionContainer activity_regularizer get; set;
Optional regularizer function for the output of this layer.