def __init__(self, signal, config):
self.input = signal
self.config = config
model = tf.keras.layers.Conv1D(filters=256, kernel_size=1, strides=1, use_bias=False, padding='same')(signal)
for i in range(5):
model = blocks.pre_activation_residual_block(model)
max_dilation = 64
i = 1
while i <= max_dilation:
model = blocks.tcn_identity_block(model, i)
i = i * 2
self.logits = tf.keras.layers.Dense(5)(model)
def __init__(self, signal, params):
self.input = signal
self.params = params
max_dilation = 128
model = tf.keras.layers.Conv1D(filters=256, kernel_size=1, strides=1, use_bias=False, padding='same')(signal)
for i in range(12):
model = blocks.pre_activation_residual_block(signal)
i = 1
while i <= max_dilation:
model, _ = blocks.wavenet_weird_block(model, i)
model, _ = blocks.wavenet_weird_block(model, i)
i = i * 2
self.logits = tf.keras.layers.Dense(5)(model)
def __init__(self, signal, config):
hidden_num = 100
self.input = signal
self.config = config
model = signal
for i in range(5):
model = blocks.pre_activation_residual_block(model)
model = blocks.lstm_block(model)
weight_bi = tf.Variable(tf.truncated_normal([2, hidden_num], stddev=np.sqrt(2.0 / (2*hidden_num))))
bias_bi = tf.Variable(tf.zeros([hidden_num]))
model = tf.reshape(model, [tf.shape(model)[0], 300, 2, hidden_num])
model = tf.nn.bias_add(tf.reduce_sum(tf.multiply(model, weight_bi), axis=2), bias_bi)
model = tf.keras.layers.Dense(5)(model)
self.logits = model
def __init__(self, signal, config):
hidden_num = 128
self.input = signal
self.config = config
model = tf.keras.layers.Conv1D(filters=256, kernel_size=1, strides=1, use_bias=False, padding='same')(signal)
for i in range(5):
model = blocks.pre_activation_residual_block(model)
model = blocks.lstm_identity_block(model)
weight_bi = tf.Variable(tf.truncated_normal([2, hidden_num], stddev=np.sqrt(2.0 / (2*hidden_num))))
bias_bi = tf.Variable(tf.zeros([hidden_num]))
model = tf.reshape(model, [tf.shape(model)[0], 300, 2, hidden_num])
model = tf.nn.bias_add(tf.reduce_sum(tf.multiply(model, weight_bi), axis=2), bias_bi)
model = tf.keras.layers.Dense(5)(model)
self.logits = model
def __init__(self, signal, params):
self.input = signal
self.params = params
max_dilation = 128
model = tf.keras.layers.Conv1D(filters=256, kernel_size=1, strides=1, use_bias=False, padding='same')(signal)
for i in range(5):
model = blocks.pre_activation_residual_block(model)
i = 1
skip_connections = []
while i <= max_dilation:
model, skip = blocks.wavenet_bidirectional_block(model, i)
skip_connections.append(skip)
i = i * 2
skip_sum = tf.keras.layers.Add()(skip_connections)
self.logits = tf.keras.layers.Dense(5)(skip_sum)