import tensorflow as tf

v1 = tf.Variable(1.0,dtype=tf.float32)

v2 = tf.Variable(2.0,dtype=tf.float32)

x = v1 + v2

saver = tf.train.Saver()

with tf.Session() as sess:

    sess.run(tf.global_variables_initializer())

    result = sess.run(x)

    #将模型保存在model文件夹下

    saver.save(sess,'./model/test.model')

    print('result:{}'.format(result))

import tensorflow as tf

saver = tf.train.import_meta_graph('./model/test.model.meta')

with tf.Session() as sess:

    saver.restore(sess,'./model/test.model')

    print(sess.run(tf.get_default_graph().get_tensor_by_name('add:0')))

import tensorflow as tf

a = tf.Variable(5.0,dtype=tf.float32,name='a')

b = tf.Variable(6.0,dtype=tf.float32,name='b')

x = a + b

saver = tf.train.Saver({'v1':a,'v2':b})

with tf.Session() as sess:

    saver.restore(sess,'./model/test.model')

    print(sess.run([x]))