Keras Concatenate and Lambda not recognized as Layer, TypeError: The added layer must be an instance of class 层

Question

我一直无法找到解决问题的方法。 以下（使用和不使用 Lambda）无法识别 Concatenate as a Layer 的结果：

tensors = []
for item in items:
    tensor = tf.constant(item, dtype=tf.string, shape=[1])
    tensors.append(tensor)

def constant_layer(tensor):
    return tf.keras.layers.Input(tensor=tensor, shape=tensor.shape)

input_layers = []
for tensor in tensors:
    input_layers.append(constant_layer(tensor))

model = tf.keras.Sequential()
model.add(tf.keras.layers.Lambda(lambda x: tf.keras.layers.Concatenate()(x))(input_layers))
model.add(tf.keras.layers.Dense(10, activation='relu'))
model.add(tf.keras.layers.Dense(10, activation='relu'))
model.add(tf.keras.layers.Dense(1))

# Compile model
model.compile(optimizer='adam', loss='mean_squared_error')

我收到的错误如下：

  File "script.py", line 25
    model.add(tf.keras.layers.Lambda(lambda x: tf.keras.layers.Concatenate()(x))(input_layers))
  File "./lib/python3.7/site-packages/tensorflow/python/trackable/base.py", line 205, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "./lib/python3.7/site-packages/keras/utils/traceback_utils.py", line 70, in error_handler
    raise e.with_traceback(filtered_tb) from None
  File "./lib/python3.7/site-packages/keras/engine/sequential.py", line 187, in add
    "The added layer must be an instance of class Layer. "
TypeError: The added layer must be an instance of class Layer. Received: layer=KerasTensor(type_spec=TensorSpec(shape=(1,), dtype=tf.string, name=None), name='lambda/concatenate/concat/concat:0', description="created by layer 'lambda'") of type <class 'keras.engine.keras_tensor.KerasTensor'>.

我正在尝试将 TF 变量应用于 model，输入是一个字符串列表。 有什么建议么？

Answer 1

为我修复它的解决方法是将连接的结果添加到输入层，以下对我来说成功了：

concat_layer = tf.keras.layers.Concatenate()(tensors)
reshape_layer = tf.keras.layers.Reshape((1,))(concat_layer)
input_layer = tf.keras.layers.Input(tensor=reshape_layer, shape=reshape_layer.shape, dtype=tf.string)

model = tf.keras.Sequential()
model.add(input_layer)
model.add(tf.keras.layers.Dense(10, activation='relu'))
model.add(tf.keras.layers.Dense(10, activation='relu'))
model.add(tf.keras.layers.Dense(1))

...