T5-tensorboard 可視化之圖層

Tensorboard 可視化之圖層

Tensorflow 自帶的 tensorboard 可以構建我們的神經網絡圖層, 讓我們看看自己的神經網絡長什么樣.

開始構建圖層結構啦

我們要用到前面用到的代碼來構建神經網絡圖像

首先是數據的輸入 input :

# 我們先給輸入和輸出的占位符指定名稱
# 指定的名稱會在可視化的圖層 input 中顯示
xs = tf.placeholder(tf.float32, [None, 1], name='x_in')
ys = tf.placeholder(tf.float32, [None, 1], name='y_in')

圖層可以包含子圖層, 所以, 我們要用到 with tf.name_scope('inputs') 將xs和ys包含起來, 作為輸入層. (inputs 就是圖層的名字, 可任意命名)

with tf.name_scope('inputs'):
	xs = tf.placeholder(tf.float32, [None, 1], name='x_in')
	ys = tf.placeholder(tf.float32, [None, 1], name='y_in')

接下來, 就是layer了, 我們前面用了add_layer函數來添加圖層, 這里我可以直接在add_layer函數里面構建圖層結構. ( 記得 name_scope 可以嵌套的哦

def add_layer(inputs, in_size, out_size, activation_function=None):
	# 每一個圖層名為 `layer`
	with tf.name_scope('layer'):
		# 添加層里面的小部件也需要定義
		with tf.name_scope('weights'):
			Weights = tf.Variable(tf.random_normal([in_size, out_size]))
		with tf.name_scope('biases'):
			biases = tf.Variable(tf.zeros([1, out_size]) + 0.1)
		with tf.name_scope('wx_plus_b'):
			Wx_plus_b = tf.add(tf.matmul(inputs, Weights), biases)
		if activation_function is None:
			outputs = Wx_plus_b
		else:
			outputs = activation_function(Wx_plus_b, )
		return outputs	

最后是loss和training部分了, 同樣為他們各自取名

with tf.name_scope('loss'):
	loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys-prediction),
							reduction_indices=[1]))
with tf.name_scope('train'):
	    train_setp = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

繪制我們的神經網絡圖啦

繪制方法為tf.summary.FileWriter(dir, sess.graph), 第一個參數為圖的儲存路徑, 第二個參數是將前面定義好的框架信息收集起來, 最后放到dir目錄中, 因此需要先獲得 Session

sess = tf.Session()
# 執行 python3 filename.py 之后會自動創建 graph 文件夾
# 並把生成的圖層圖像信息保存在 graph 下, 需要用瀏覽器觀看
writer = tf.summary.FileWriter('graph/', sess.graph)

運行完整代碼后「[完整代碼] (#code)」, 會自動生成圖片信息並保存到 graph 目錄中, 然后什么在 graph 上一級目錄執行下面這條命令, 它會輸出一條地址, 我們在瀏覽器上打開http://127.0.1.6006:1

Ubuntu ~#  tensorboard --logdir='./graph/'
Starting TensorBoard b'41' on port 6006
(You can navigate to http://127.0.1.1:6006)
...

這個網頁有多個選項卡, 因為我們只定義了sess.graph, 所以我們切換到GRAPH, 可以看到我們的神經網絡的基本結構

我們再點開inputs圖層看看, 里面有x_in和y_in兩個輸入, 這兩個名字是我們取的, 其他的可以自己看看啦

---

完整代碼

最后把輸入輸出圖層也加上了名字, 看下完整代碼

# !/usr/bin/env python3
# -*- coding: utf-8 -*-

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt 


# Define add layer function.
def add_layer(inputs, in_size, out_size, activation_function=None):
    # add one more layer and return the output of this layer
    with tf.name_scope('layer'):
        with tf.name_scope('weights'):
            Weights = tf.Variable(tf.random_normal([in_size, out_size]))
        with tf.name_scope('biases'):
            biases = tf.Variable(tf.zeros([1, out_size]) + 0.1)
        with tf.name_scope('wx_plus_b'):
            Wx_plus_b = tf.add(tf.matmul(inputs, Weights), biases)
        if activation_function is None:
            outputs = Wx_plus_b
        else:
            outputs = activation_function(Wx_plus_b, )
        return outputs

# Define palceholder for inputs to network.
# Use [with] including xs & ys:
with tf.name_scope('inputs'):
    xs = tf.placeholder(tf.float32, [None, 1], name='x_in') # Add name
    ys = tf.placeholder(tf.float32, [None, 1], name='y_in')


# Add hidden layer
with tf.name_scope('hidden_layer'):
    l1 = add_layer(xs, 1, 10, activation_function=tf.nn.relu)
# Add output layer
with tf.name_scope('output_layer'):
    prediction = add_layer(l1, 10, 1, activation_function=None)

# The error between prediction and real data
with tf.name_scope('loss'):
    loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys-prediction),
        reduction_indices=[1]))

with tf.name_scope('train'):
    train_setp = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

sess = tf.Session()
# ** Add frame to file
writer = tf.summary.FileWriter('./graph/', sess.graph)

# Important step
sess.run(tf.initialize_all_variables())

最后效果: