def call(self, inputs, mask=None):
pos = K.relu(inputs)
if K.backend() == 'theano':
neg = (K.pattern_broadcast(self.alpha, self.param_broadcast) *
(inputs - K.abs(inputs)) * 0.5)
else:
neg = -self.alpha * K.relu(-inputs)
return pos + neg
def call(self, inputs, mask=None):
pos = K.relu(inputs)
if K.backend() == 'theano':
neg = (K.pattern_broadcast(self.alpha, self.param_broadcast) *
(inputs - K.abs(inputs)) * 0.5)
else:
neg = -self.alpha * K.relu(-inputs)
return pos + neg
def _attention_mlp(self, memory):
for i in range(self.attention_mlp_layers):
memory = self._linear(memory, self.mlp_kernels[i],
self.mlp_biases[i])
if i < self.attention_mlp_layers - 1:
memory = K.relu(memory)
return memory
def relu(x, alpha=0., max_value=None): return K.relu(x, alpha=alpha, max_value=max_value)
def call(self, inputs): return K.relu(inputs, alpha=self.alpha)
def relu6(x):
return K.relu(x, max_value=6)
def relu6(x): return K.relu(x, max_value=6)
def relu(x, alpha=0., max_value=None): return K.relu(x, alpha=alpha, max_value=max_value)
def call(self, inputs):
return K.relu(inputs, alpha=self.alpha)
