适用于4-D张量的TensorFlow LSTM

Question

如果我有一个形状为[None, None, None, 10] ，我想做下面的事情，

input = tf.placeholder([None, None, None, 10], dtype=tf.float32)
length = tf.placeholder([None, None], dtype=tf.int32)
cell = tf.nn.rnn_cell.BasicLSTMCell(num_units=10)
def fn(inp): 
  output, _ = tf.dynamic_rnn(cell, inp[0], sequence_length=inp[1])
  return output

tf.map_fn(fn, (input, length), dtype=tf.float32)

但这不适用于自动梯度推导，它会产生如下错误，

Cannot use 'XXX' as input to 'gradients/YYY/while/TensorArrayWrite/TensorArrayWriteV3_grad/TensorArrayReadV3/f_acc' 
because 'XXX' is in a while loop.

这是TensorFlow的局限性吗？它不能处理嵌套的while_loop （ dynamic_rnn和map_fn都在while_loop运行while_loop ），还是有一种解决方法？

Answer 1

TensorFlow 1.6已解决此问题