functional API测试失败

Question

为了进行测试，我尝试将https://colab.research.google.com/github/tensorflow/docs/blob/master/site/en/r2/tutorials/generative/dcgan.ipynb#scrollTo=6bpTcDqoLWjY的生成器转换为functional API，但它不起作用。ValueError:使用序列设置数组元素。

有人知道我做错了什么吗？

我将生成器代码替换为：

def make_generator_model():
  inputs = tf.keras.Input(shape=(100,))

  l = layers.Dense(7*7*256, use_bias=False)(inputs)
  #l1 = layers.BatchNormalization(l)
  l2 = layers.LeakyReLU(l)
  l3 = layers.Reshape((7, 7, 256))(l2)
  l4 = layers.Conv2DTranspose(128, (5, 5), strides=(1, 1), padding='same', use_bias=False)(l3)
  #l5 = layers.BatchNormalization()(l4)
  l6 = layers.LeakyReLU()(l4)
  l7 = layers.Conv2DTranspose(64, (5, 5), strides=(2, 2), padding='same', use_bias=False)(l6)
  #l8 = layers.BatchNormalization()(l7)
  l9 = layers.LeakyReLU()(l7)
  l10 = layers.Conv2DTranspose(1, (5, 5), strides=(2, 2), padding='same', use_bias=False, activation='tanh')(l9)
  return tf.keras.Model(inputs=inputs, outputs=l10)

Answer 1

错误来自于LeakyRely () ()之后缺少()。工作代码是：

def make_generator_model():
  inputs = tf.keras.Input(shape=(100,))

  l = layers.Dense(7*7*256, use_bias=False)(inputs)
  l1 = layers.BatchNormalization()(l)
  l2 = layers.LeakyReLU()(l1)
  l3 = layers.Reshape((7, 7, 256))(l2)
  l4 = layers.Conv2DTranspose(128, (5, 5), strides=(1, 1), padding='same', use_bias=False)(l3)
  l5 = layers.BatchNormalization()(l4)
  l6 = layers.LeakyReLU()(l5)
  l7 = layers.Conv2DTranspose(64, (5, 5), strides=(2, 2), padding='same', use_bias=False)(l6)
  l8 = layers.BatchNormalization()(l7)
  l9 = layers.LeakyReLU()(l8)
  l10 = layers.Conv2DTranspose(1, (5, 5), strides=(2, 2), padding='same', use_bias=False, activation='tanh')(l9)
  return tf.keras.Model(inputs=inputs, outputs=l10)