Gradients Error Using TensorArray Tensorflow
i am trying to implement multidimentional lstm in tensorflow, I am using TensorArray to remember previous states, i am using a complicated way to get two neigbours state(above and
Solution 1:
You should add parameter parallel_iterations=1
to your while loop call.
Such as:
result, outputs_ta, states_ta = tf.while_loop(
condition, body, [time,outputs_ta,states_ta], parallel_iterations=1)
This is required because inside body you perform read and write operations on the same tensor array (states_ta
). And in case of parallel loop execution(parallel_iterations > 1) some thread may try to read info from tensorArray, that was not written to it by another one.
I've test your code snippet with parallel_iterations=1 on tensorflow 0.12.1 and it works as expected.
Post a Comment for "Gradients Error Using TensorArray Tensorflow"