tensorflow笔记--placeholder

Dependencies:
tensorflow: 1.1.0
"""
import tensorflow as tf

x1 = tf.placeholder(dtype=tf.float32, shape=None)
y1 = tf.placeholder(dtype=tf.float32, shape=None)
z1 = x1 + y1

x2 = tf.placeholder(dtype=tf.float32, shape=[2, 1])
y2 = tf.placeholder(dtype=tf.float32, shape=[1, 2])
z2 = tf.matmul(x2, y2)

with tf.Session() as sess:
    # when only one operation to run
    z1_value = sess.run(z1, feed_dict={x1: 1, y1: 2})

    # when run multiple operations
    z1_value, z2_value = sess.run(
        [z1, z2],       # run them together
        feed_dict={
            x1: 1, y1: 2,
            x2: [[2], [2]], y2: [[3, 3]]
        })
    print(z1_value)
    print(z2_value)