TensorBoard可视化

基本用法

import tensorflow as tf

input1 = tf.constant([1.,2.,3.],name='input1')
input2 = tf.Variable(tf.ran  dom_uniform([3]),name='input2')
output = input1 + input2

writer = tf.summary.FileWriter('./log',tf.get_default_graph())
writer.close()

运行以下命令:

tensorboard –logdir=log

在浏览器打开：localhost:6006

获取在某一次运行时所消耗的时间和内存

    with tf.Session()