TensorFlow：记住下一批的 LSTM 状态（stateful LSTM）

Question

给定一个经过训练的 LSTM 模型，我想对单个时间步执行推理，即下面示例中的 seq_length = 1。在每个时间步之后，需要为下一个 'batch' 记住内部 LSTM（记忆和隐藏）状态。对于推理的最开始，内部 LSTM 状态 init_c, init_h 是在给定输入的情况下计算的。然后将这些存储在传递给 LSTM 的 LSTMStateTuple 对象中。在训练期间，每个时间步都会更新此状态。但是为了推断，我希望 state 在批次之间保存，即只需要在最开始计算初始状态，然后在每个 'batch' 之后保存 LSTM 状态（n= 1).

我发现了这个相关的 Whosebug 问题：. However this only works if state_is_tuple=False, but this behavior is soon to be deprecated by TensorFlow (see rnn_cell.py). Keras seems to have a nice wrapper to make stateful LSTMs possible but I don't know the best way to achieve this in TensorFlow. This issue on the TensorFlow GitHub is also related to my question: https://github.com/tensorflow/tensorflow/issues/2838

有人对构建有状态 LSTM 模型有什么好的建议吗？

inputs  = tf.placeholder(tf.float32, shape=[None, seq_length, 84, 84], name="inputs")
targets = tf.placeholder(tf.float32, shape=[None, seq_length], name="targets")

num_lstm_layers = 2

with tf.variable_scope("LSTM") as scope:

    lstm_cell  = tf.nn.rnn_cell.LSTMCell(512, initializer=initializer, state_is_tuple=True)
    self.lstm  = tf.nn.rnn_cell.MultiRNNCell([lstm_cell] * num_lstm_layers, state_is_tuple=True)

    init_c = # compute initial LSTM memory state using contents in placeholder 'inputs'
    init_h = # compute initial LSTM hidden state using contents in placeholder 'inputs'
    self.state = [tf.nn.rnn_cell.LSTMStateTuple(init_c, init_h)] * num_lstm_layers

    outputs = []

    for step in range(seq_length):

        if step != 0:
            scope.reuse_variables()

        # CNN features, as input for LSTM
        x_t = # ... 

        # LSTM step through time
        output, self.state = self.lstm(x_t, self.state)
        outputs.append(output)

Answer 1

Tensorflow，在 RNN 中保存状态的最佳方式？实际上是我最初的问题。下面的代码是我如何使用状态元组。

with tf.variable_scope('decoder') as scope:
    rnn_cell = tf.nn.rnn_cell.MultiRNNCell \
    ([
        tf.nn.rnn_cell.LSTMCell(512, num_proj = 256, state_is_tuple = True),
        tf.nn.rnn_cell.LSTMCell(512, num_proj = WORD_VEC_SIZE, state_is_tuple = True)
    ], state_is_tuple = True)

    state = [[tf.zeros((BATCH_SIZE, sz)) for sz in sz_outer] for sz_outer in rnn_cell.state_size]

    for t in range(TIME_STEPS):
        if t:
            last = y_[t - 1] if TRAINING else y[t - 1]
        else:
            last = tf.zeros((BATCH_SIZE, WORD_VEC_SIZE))

        y[t] = tf.concat(1, (y[t], last))
        y[t], state = rnn_cell(y[t], state)

        scope.reuse_variables()

我没有使用 tf.nn.rnn_cell.LSTMStateTuple，而是创建了一个工作正常的列表列表。在这个例子中，我没有保存状态。然而，您可以很容易地从变量中创建状态，并只使用赋值来保存值。

Answer 2

我发现将所有层的整个状态保存在一个占位符中是最简单的。

init_state = np.zeros((num_layers, 2, batch_size, state_size))

...

state_placeholder = tf.placeholder(tf.float32, [num_layers, 2, batch_size, state_size])

然后在使用原生tensorflow RNN Api.

之前解压它并创建一个LSTMStateTuples元组

l = tf.unpack(state_placeholder, axis=0)
rnn_tuple_state = tuple(
[tf.nn.rnn_cell.LSTMStateTuple(l[idx][0], l[idx][1])
 for idx in range(num_layers)]
)

RNN 传入 API:

cell = tf.nn.rnn_cell.LSTMCell(state_size, state_is_tuple=True)
cell = tf.nn.rnn_cell.MultiRNNCell([cell]*num_layers, state_is_tuple=True)
outputs, state = tf.nn.dynamic_rnn(cell, x_input_batch, initial_state=rnn_tuple_state)

state - 变量将作为占位符提供给下一批。

TensorFlow：记住下一批的 LSTM 状态（stateful LSTM）

TensorFlow: Remember LSTM state for next batch (stateful LSTM)

python

stateful

lstm

tensorflow

recurrent-neural-network