autoencoder - Feeding a hidden tensor in Tensorflow -
i have autoencoder
. model not important now. suppose model takes input image , output reconstructed image. after training, see effect of 1 tensor on output. in addition, images being fed autoencoder
through fifoqueue
. therefore, when running following peace of code:
reconstructed_image = sess.run([deconv_image], feed_dict={mu:my_vector})
where deconv_image
output tensor
of model , mu
hidden tensor inside model; automatically feed model image queue
.
my question is: value inside mu
replaced whatever should come input image, or, takes vector fed using feed_dict
argument.
any appreciated!!
when running final tensor, is, evaluating last tensor of graph, run tensors depends on. if have y3
operation depends on y2
, y2
depends on y1
, then, running final tensor in graph cause y1
run first, y2
evaluated after gets input y1
, finally, output of y2
feed y3
. graph follows: y1 -> y2 -> y3
on other hand, can run (evaluate) y3
feeding inputs directly using feed_dict
argument. in case, y2
, y1
evaluated.
ex:
import tensorflow tf import numpy np x = np.array([1.0, 2.0, 3.0]) x_var = tf.variable(x, dtype=tf.float32) y1 = tf.square(x_var) y2 = tf.subtract(y1, tf.constant(1.0)) init_op = tf.global_variables_initializer() tf.session() sess: sess.run(init_op) print(sess.run(y2)) # output: [ 0. 3. 8.] print(sess.run(y2, feed_dict={y1: [1.0, 1.0, 1.0]}))# output: [ 0. 0. 0.]
Comments
Post a Comment