PMSPLayer

Bases: tf.keras.layers.Layer

PMSP sim 3 model RNN layer.

Unrolling the PMSPCell for a fixed number of time steps.

See Plaut, McClelland, Seidenberg and Patterson (1996), simulation 3.

Parameters:

Name	Type	Description	Default
tau	float	Time-averaging parameter, from 0 to 1.	required
h_units	int	Number of units in the hidden layer.	required
p_units	int	Number of units in the phonological layer.	required
c_units	int	Number of units in the cleanup layer.	required
h_noise	float	Gaussian noise parameter (in stddev) for hidden layer.	0.0
p_noise	float	Gaussian noise parameter (in stddev) for phonological layer.	0.0
c_noise	float	Gaussian noise parameter (in stddev) for cleanup layer.	0.0
connections	List[str]	List of connections to use, each connection consists of two letters (from, to). Default is ["oh", "ph", "hp", "pp", "cp", "pc"].	None
zero_out_rates	Dict[str, float]	Dictionary of zero-out rates for each connection. Default is {c: 0.0 for c in self.connections}. See PMSP for more details.	None
l2	float	L2 regularization parameter, apply to all trainable weights and biases.	0.0

build(input_shape)

Build the layers.

call(inputs, training=False, return_internals=False)

Forward pass, unrolling the RNN cell.

Parameters:

Name	Type	Description	Default
inputs	tf.Tensor	Input tensor with shape (batch_size, max_ticks, input_units).	required
training	bool	Whether to run in training mode or not.	False
return_internals	bool	Whether to return intermediate inputs to each connection.	False

Returns:

Name	Type	Description
outputs	Dict[str, tf.Tensor]	Activations with shape (batch_size, max_ticks, units) in each layer. E.g.: {"phonology": ...}, if return_internals=True, also return intermediate inputs with shape (batch_size, max_ticks, units) in each connection. E.g., {"oh": last_o @ w_{oh}, ...}