时间戳

首页 > 解决方案 > 编码器-解码器 LSTMCell/RNN 网络的输入

neural-network - 编码器-解码器 LSTMCell/RNN 网络的输入

问题描述

我正在使用 Keras 创建一个 LSTM 编码器-解码器网络，遵循此处提供的代码：https ://github.com/LukeTonin/keras-seq-2-seq-signal-prediction 。我所做的唯一更改是将 GRUCell 替换为 LSTMCell。基本上编码器和解码器都包含 2 层，35 个 LSTMCell。这些层使用 RNN 层相互堆叠（并组合）。

LSTMCell 返回 2 个状态，而 GRUCell 返回 1 个状态。这是我遇到错误的地方，因为我不知道如何为 LSTMCell 的 2 个返回状态编码。

我创建了两个模型：第一个是编码器-解码器模型。第二，预测模型。我在编码器-解码器模型中没有遇到任何问题，而是在预测模型的解码器中遇到问题。

我得到的错误是： ValueError: Layer rnn_4 expects 9 inputs, but it received 3 input tensors. Input received: [<tf.Tensor 'input_4:0' shape=(?, ?, 1) dtype=float32>, <tf.Tensor 'input_11:0' shape=(?, 35) dtype=float32>, <tf.Tensor 'input_12:0' shape=(?, 35) dtype=float32>]

当在预测模型中运行下面的这一行时，会发生此错误：

decoder_outputs_and_states = decoder(
        decoder_inputs, initial_state=decoder_states_inputs)

这适合的代码部分是：

encoder_predict_model = keras.models.Model(encoder_inputs,
                                           encoder_states)

decoder_states_inputs = []

# Read layers backwards to fit the format of initial_state
# For some reason, the states of the model are order backwards (state of the first layer at the end of the list)
# If instead of a GRU you were using an LSTM Cell, you would have to append two Input tensors since the LSTM has 2 states.
for hidden_neurons in layers[::-1]:
    # One state for GRU, but two states for LSTMCell
    decoder_states_inputs.append(keras.layers.Input(shape=(hidden_neurons,)))
decoder_outputs_and_states = decoder(
        decoder_inputs, initial_state=decoder_states_inputs)

decoder_outputs = decoder_outputs_and_states[0]
decoder_states = decoder_outputs_and_states[1:]

decoder_outputs = decoder_dense(decoder_outputs)

decoder_predict_model = keras.models.Model(
        [decoder_inputs] + decoder_states_inputs,
        [decoder_outputs] + decoder_states)

有人可以帮我处理上面的 for 循环，然后我应该通过解码器的初始状态吗？

标签： neural-networklstmrecurrent-neural-networkdecoderencoder-decoder

解决方案

我有一个类似的错误，我只是按照他说的解决了，添加了另一个输入张量：

# If instead of a GRU you were using an LSTM Cell, you would have to append two Input tensors since the LSTM has 2 states.
for hidden_neurons in layers[::-1]:
    # One state for GRU
    decoder_states_inputs.append(keras.layers.Input(shape=(hidden_neurons,)))
    decoder_states_inputs.append(keras.layers.Input(shape=(hidden_neurons,)))

在这里它解决了问题......

推荐阅读