Tensor Flow - LSTM - 'Tensor' object not iterable

Question

Hi I am using the following function for lstm rnn cell.

def LSTM_RNN(_X, _istate, _weights, _biases):
    # Function returns a tensorflow LSTM (RNN) artificial neural network from given parameters. 
    # Note, some code of this notebook is inspired from an slightly different 
    # RNN architecture used on another dataset: 
    # https://tensorhub.com/aymericdamien/tensorflow-rnn

    # (NOTE: This step could be greatly optimised by shaping the dataset once
    # input shape: (batch_size, n_steps, n_input)
    _X = tf.transpose(_X, [1, 0, 2])  # permute n_steps and batch_size

    # Reshape to prepare input to hidden activation
    _X = tf.reshape(_X, [-1, n_input]) # (n_steps*batch_size, n_input)

    # Linear activation
    _X = tf.matmul(_X, _weights['hidden']) + _biases['hidden']

    # Define a lstm cell with tensorflow
    lstm_cell = rnn_cell.BasicLSTMCell(n_hidden, forget_bias=1.0)


    # Split data because rnn cell needs a list of inputs for the RNN inner loop
    _X = tf.split(0, n_steps, _X) # n_steps * (batch_size, n_hidden)

    # Get lstm cell output
    outputs, states = rnn.rnn(lstm_cell, _X, initial_state=_istate)

    # Linear activation
    # Get inner loop last output
    return tf.matmul(outputs[-1], _weights['out']) + _biases['out']

The function's output is stored under pred variable.

pred = LSTM_RNN(x, istate, weights, biases)

But its showing the following error. (which states that tensor object is not iterable.)

Here is the ERROR image link - http://imgur.com/a/NhSFK

Please help me with this and I apologize if this question seems silly as I am fairly new to the lstm and tensor flow library.

Thanks.

Solution 1

The error happened when it's trying to unpack state with statement c, h=state. Depending on which version of tensorflow you are using (you can check the version info by typing import tensorflow; tensorflow.__version__ in python interpreter), in version prior to r0.11, the default setting for the state_is_tuple argument when you initialize the rnn_cell.BasicLSTMCell(n_hidden, forget_bias=1.0) is set to be False. See the documentation here.

Since tensorflow version r0.11 (or the master version), the default setting for state_is_tuple is set to be True. See the documentation here.

If you installed r0.11 or the master version of tensorflow, try change the BasicLSTMCell initialization line into: lstm_cell = rnn_cell.BasicLSTMCell(n_hidden, forget_bias=1.0, state_is_tuple=False). The error you are encountering should go away. Although, their page does say that the state_is_tuple=False behavior will be deprecated soon.

Solution 2

I happened to met the same question at same time. I just describe my circumstance which may do help for u

it state as follow

c1_ex, T1_ex = tf. ones(10,tf. int 32)
 raise Type Error ...

I find that left side of '=' have been set two name of vector in advance

while the other side just return a vector

sorry for my inefficiency of English

your problem actually appear in line 146 not the line 193