Eval () and run () in a tensor stream

Question

I mean the Deep MNIST tutorial for experts defined by the tensor flow. I have a problem in Teach and evaluate part of this tutorial. There they gave a code example as follows.

cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y_conv),reduction_indices=[1])) train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy) correct_prediction = tf.equal(tf.argmax(y_conv,1), tf.argmax(y_,1)) accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32)) sess.run(tf.initialize_all_variables()) for i in range(20000): batch = mnist.train.next_batch(50) if i%100 == 0: train_accuracy = accuracy.eval(feed_dict={x:batch[0], y_: batch[1], keep_prob: 1.0}) print("step %d, training accuracy %g"%(i, train_accuracy)) train_step.run(feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5}) print("test accuracy %g"%accuracy.eval(feed_dict={x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0})) 

So, in this code segment, they used accuracy.eval() at a time. And other time train_step.run() . As I know, both of them are tensor variables.

And in some cases, I saw how

 sess.run(variable, feed_dict) 

So my question is what are the differences between these three implementations. And how can I know what to use when ...?

Thanks!

Answer 1

If you have only one session by default, they are basically the same.

From https://www.tensorflow.org/versions/r0.10/api_docs/python/framework.html#Operation :

op.run () is a shortcut to calling tf.get_default_session (). run (op)

From https://www.tensorflow.org/versions/r0.10/api_docs/python/framework.html#Tensor :

t.eval () is a shortcut to calling tf.get_default_session (). run (t)

Why are these differences between Tensor and Operation? From https://www.tensorflow.org/versions/r0.10/api_docs/python/framework.html#Tensor :

Note: The tensor class will be replaced by a pin in the future. Currently, these two are aliases for each other.

Answer 2

The difference lies in the operation against the tensor. Use run () operations and tensors use eval ().

There seems to be a link to this question in the TensorFlow FAQ: https://www.tensorflow.org/programmers_guide/faq#running_a_tensorflow_computation

This section discusses the following question: what is the difference between Session.run () and Tensor.eval ()?