《tensorflow实战google》第二章代码

import tensorflow as tf

g1 = tf.Graph()
with g1.as_default():
    v = tf.get_variable("v", initializer=tf.zeros_initializer(), shape=[3])

g2 = tf.Graph()
with g2.as_default():
    v = tf.get_variable("v", initializer=tf.ones_initializer(), shape=[5])

with tf.Session(graph=g1) as sess:
    tf.initialize_all_variables().run()
    with tf.variable_scope("", reuse=True):
        print(sess.run(tf.get_variable("v")))

with tf.Session(graph=g2) as sess:
    tf.initialize_all_variables().run()
    with tf.variable_scope("", reuse=True):
        print(sess.run(tf.get_variable("v")))