Tensorflow GRU 层调用（）参数——TypeError：调用（）得到了一个意外的关键字参数“reset_after”

Posted 2023-02-16

【中文标题】Tensorflow GRU 层调用（）参数——TypeError：调用（）得到了一个意外的关键字参数“reset_after”

【英文标题】：Tensorflow GRU layer call() arguments -- TypeError: call() got an unexpected keyword argument 'reset_after'

【发布时间】：2021-11-15 21:27:05

【问题描述】：

我实现了一个带有 GRU 层的模型，该模型及其训练工作正常

   class MyModel(tf.keras.Model):
      def __init__(self, vocab_size, embedding_dim, rnn_units):
        super().__init__(self)
        self.embedding = tf.keras.layers.Embedding(vocab_size, embedding_dim)
        self.gru = tf.keras.layers.GRU(rnn_units,
                                       return_sequences=True,
                                       return_state=True)
        self.dense = tf.keras.layers.Dense(vocab_size)
    
      def call(self, inputs, states=None, return_state=False, training=False):
        x = inputs
        x = self.embedding(x, training=training)
        if states is None:
          states = self.gru.get_initial_state(x)
        x, states = self.gru(x, initial_state=states, training=training)
        x = self.dense(x, training=training)
    
        if return_state:
          return x, states
        else:
          return x

我只是更改了 GRU 层的定义，使其 (1) CuDNN 兼容 (2) 添加 dropout

在我保留的模型定义中

    self.gru = tf.keras.layers.GRU(rnn_units,
                                   return_sequences=True,
                                   return_state=True)

在我设置的调用函数中

    if states is None:
      states = self.gru.get_initial_state(x)
    x, states = self.gru(x, initial_state=states, training=training,
                         reset_after=True, recurrent_activation='sigmoid', # to make it more GPU friendly
                         recurrent_dropout=0.2, dropout=0.2 # to add some dropout to it
                         )

似乎遵守了Keras 或Tensorflow 准则，但我收到此错误

Traceback (most recent call last):
  File "rnn_train_004.py", line 125, in <module>
    example_batch_predictions = model(input_example_batch)
  File "/usr/local/lib/python3.6/dist-packages/keras/engine/base_layer.py", line 1037, in __call__
    outputs = call_fn(inputs, *args, **kwargs)
  File "rnn_train_004.py", line 107, in call
    recurrent_dropout=0.2, dropout=0.2 # to add some dropout to it
  File "/usr/local/lib/python3.6/dist-packages/keras/layers/recurrent.py", line 716, in __call__
    return super(RNN, self).__call__(inputs, **kwargs)
  File "/usr/local/lib/python3.6/dist-packages/keras/engine/base_layer.py", line 1037, in __call__
    outputs = call_fn(inputs, *args, **kwargs)
TypeError: call() got an unexpected keyword argument 'reset_after'

【问题讨论】：

大部分参数（reset_after、recurrent_activation、recurrent_dropout、dropout）必须传递给构造函数。您将它们传递给call。

【参考方案1】：

将参数传递给构造函数而不是 call() 方法

   class MyModel(tf.keras.Model):
      def __init__(self, vocab_size, embedding_dim, rnn_units):
        super().__init__(self)
        self.embedding = tf.keras.layers.Embedding(vocab_size, embedding_dim)
        self.gru = tf.keras.layers.GRU(rnn_units,
                                       return_sequences=True,
                                       return_state=True,
                                       reset_after=True,
                                       recurrent_activation='sigmoid', # to make it more GPU friendly
                                       recurrent_dropout=0.2,
                                       dropout=0.2 # to add some dropout to it
                                       )
        self.dense = tf.keras.layers.Dense(vocab_size)
    
      def call(self, inputs, states=None, return_state=False, training=False):
        x = inputs
        x = self.embedding(x, training=training)
        if states is None:
          states = self.gru.get_initial_state(x)
        x, states = self.gru(x, initial_state=states, training=training)
        x = self.dense(x, training=training)
    
        if return_state:
          return x, states
        else:
          return x