TF0003、_Fetch和_Feed

import tensorflow as tf
import warnings
warnings.filterwarnings('ignore')
# Fetch:可以在session中同时计算多个tensor或执行多个操作op
# 定义三个常量
input1 = tf.constant(3.0)
input2 = tf.constant(2.0)
input3 = tf.constant(5.0)
# 加法op
add = tf.add(input2,input3)
# 乘法op
mul = tf.multiply(input1, add)

with tf.Session() as sess:
    result1,result2 = sess.run([add,mul]) #顺序不影响
    print(result1,result2)

输出：

7.0 21.0

# Feed:先定义占位符，等需要的时候再传入数据
input1 = tf.placeholder(tf.float32)
input2 = tf.placeholder(tf.float32)
# 乘法op
output = tf.multiply(input1, input2)

with tf.Session() as sess:
    print(sess.run(output, feed_dict={input1:8.0,input2:2.0}))

输出：

16.0