如何使用tensorboard

知道如何启动tensorboard后，就该了解一下怎么使用它。

1. 运行python示例，test.py

import tensorflow as tf

with tf.name_scope('graph') as scope:
     matrix1 = tf.constant([[3., 3.]],name ='matrix1')  #1 row by 2 column
     matrix2 = tf.constant([[2.],[2.]],name ='matrix2') # 2 row by 1 column
     product = tf.matmul(matrix1, matrix2,name='product')
  
sess = tf.Session()
writer = tf.summary.FileWriter("logs/", sess.graph)
init = tf.global_variables_initializer()

sess.run(init)

2. 启动tensorboard

>tensorboard --logdir=d:\develop\python\ex\logs

注：这里logdir的值和第一步logs/的路径是同一个

3. 用浏览器访问http://192.168.x.x:6006，打开GRAPH菜单如下：