我想用if ... else循环组成自定义图层，然后针对一些特殊图层权重计算梯度。该怎么做？
我们可以使用以下方法（https://www.tensorflow.org/tutorials/eager/custom_layers）组成自定义图层
class ResnetIdentityBlock(tf.keras.Model):
def __init__(self, kernel_size, filters):
super(ResnetIdentityBlock, self).__init__(name='')
filters1, filters2, filters3 = filters
self.conv2a = tf.keras.layers.Conv2D(filters1, (1, 1))
self.bn2a = tf.keras.layers.BatchNormalization()
self.conv2b = tf.keras.layers.Conv2D(filters2, kernel_size, padding='same')
self.bn2b = tf.keras.layers.BatchNormalization()
def call(self, input_tensor, training=False):
x = self.conv2a(input_tensor)
x = self.bn2a(x, training=training)
x = tf.nn.relu(x)
x = self.conv2b(x)
x = self.bn2b(x, training=training)
x = tf.nn.relu(x)
x += input_tensor
return tf.nn.relu(x)
我可以使用以下方法吗？
class ResnetIdentityBlock(tf.keras.Model):
def __init__(self, kernel_size, filters):
super(ResnetIdentityBlock, self).__init__(name='')
filters1, filters2, filters3 = filters
if trk=='rnn' and req==True:
self.conv2a = tf.keras.layers.Conv2D(filters1, (1, 1))
self.bn2a = tf.keras.layers.BatchNormalization()
if trk=!'rnn' and inf==True:
self.conv2b = tf.keras.layers.Conv2D(filters2, kernel_size, padding='same')
self.bn2b = tf.keras.layers.BatchNormalization()
def call(self, input_tensor, training=False):
if trk=='rnn' and req==True:
x = self.conv2a(input_tensor)
x = self.bn2a(x, training=training)
x = tf.nn.relu(x)
if trk=!'rnn' and inf==True:
x = self.conv2b(x)
x = self.bn2b(x, training=training)
x = tf.nn.relu(x)
x += input_tensor
return tf.nn.relu(x)
如果可以的话，如何计算一些特殊图层权重的梯度？ self.conv2b？如果要使用以下方法计算梯度，如何更改model.trainable_variables？
model=ResnetIdentityBlock()
def grad(model, inputs, targets):
with tf.GradientTape() as tape:
loss_value = loss(model, inputs, targets)
return loss_value, tape.gradient(loss_value, model.trainable_variables)