Tensorflow: set seed by tensor

I would like to build graph with some random values in it.

Then I would like to evaluate the graph with different seeds.

Here is an example what i'm trying to achieve import tensorflow as tf seed = tf.placeholder(dtype=tf.int32, shape=(), name="seed") randoms = tf.random_normal(shape=8, seed=seed)

Then I was hoping to do this: sess = tf.InteractiveSession() sess.run(randoms, {seed: 1}) sess.run(randoms, {seed: 2})

So two questions: 1. Is there any way to work around this problem 2. Why am I required to use Python-integer and not Tensor. This seems weird limitation to me.