tensorflow 1.14 的 demo 02 —— tensorboard 远程访问

tensorflow 1.14.0， 提供远程访问 tensorboard 服务的方法

第一步生成 events 文件：

在上一篇demo的基础上加了一句，如下，

 tf.summary.FileWriter("./tmp/summary", graph=sess1.graph)

hello_tensorboard_remote.py

import tensorflow as tf
import os
os.environ['TF_CPP_MIN_LOG_LEVEL']='2'

def tf114_demo():
        a = 3
        b = 4
        c = a + b
        print("a + b in py =",c)
        a_t = tf.constant(3)
        b_t = tf.constant(4)
        c_t = a_t + b_t
        print("TensorFlow add a_t + b_t =", c_t)
        with tf.Session() as sess:
                c_t_value = sess.run(c_t)
                print("c_t_value= ", c_t_value)

        return None

def graph_demo():
        a_t = tf.constant(3)
        b_t = tf.constant(4)
        c_t = a_t + b_t
        print("TensorFlow add a_t + b_t =", c_t)
        default_g = tf.get_default_graph()
        print("default_g:\n",default_g)
        print("a_t g:", a_t.graph)
        print("c_t g:", c_t.graph)
        with tf.Session() as sess:
                c_t_value = sess.run(c_t)
                print("c_t_value= ", c_t_value)
                print("sess g:", sess.graph)
        new_g = tf.Graph()

        with new_g.as_default():
                a_new = tf.constant(20)
                b_new = tf.constant(30)
                c_new = a_new + b_new
                print("c_new:", c_new)
                print("a_new g:",a_new.graph)
                print("b_new g:",c_new.graph)

        with tf.Session() as sess1:
                c_t_value = sess1.run(c_t)
#               print("c_new_value:", c_new_value)
                print("sess1 g:", sess1.graph)
                tf.summary.FileWriter("./tmp/summary", graph=sess1.graph)

        with tf.Session(graph=new_g) as new_sess:
                c_new_value = new_sess.run((c_new))
                print("c_new_value:", c_new_value)
                print("new_sess graph properties:", new_sess.graph)
#               return None



if __name__ == "__main__":
#       tf114_demo()
        graph_demo()

运行 tensorflow1 的 app：

python3 hello_tensorboard_remote.py

ls ./tmp/summary/ 

启动 tensorboard 网络服务：

tensorboard --logdir="./tmp/summary" --port 6789

6789是自己选定的端口号，尝试任选；

运行状态如下：

远程访问tensorboard：

在同一个网络内的主机网页浏览器的地址栏中输入：

http://10.208.14.37:6789

效果如下，显示出来了示例中非常简单的一个计算图：

 

如果是本机访问，则在地址栏里输入

http://127.0.0.1:6006