TF生成数据的方法

拼搏现实的明天。 2022-06-12 22:19 151阅读 0赞

转自：http://blog.csdn.net/phdat101/article/details/52442738

正常情况下，使用tf.initialize\_all\_variables()初始化变量，在完全构建好模型并加载之后才运行这个操作。生成数据的主要方法如下

1）如果需要利用已经初始化的参数给其他变量赋值

TF的变量有个initialized\_value()属性，就是初始化的值，使用方法如下：

##### [ 来自CODE的代码片][CODE]  init\_with\_other\_init.py #####

2）生成tensor的一些方法

生成tensor：

### `tf.zeros(shape, dtype=tf.float32, name=None)` ###

### `tf.zeros_like(tensor, dtype=None, name=None)` ###

### `tf.constant(value, dtype=None, shape=None, name='Const')` ###

### `tf.fill(dims, value, name=None)` ###

### `tf.ones_like(tensor, dtype=None, name=None)` ###

### `tf.ones(shape, dtype=tf.float32, name=None)` ###

生成序列

### `tf.range(start, limit, delta=1, name='range')` ###

### `tf.linspace(start, stop, num, name=None)` ###

生成随机数

### `tf.random_normal(shape, mean=0.0, stddev=1.0, dtype=tf.float32, seed=None, name=None)` ###

### `tf.truncated_normal(shape, mean=0.0, stddev=1.0, dtype=tf.float32, seed=None, name=None)` ###

### `tf.random_uniform(shape, minval=0.0, maxval=1.0, dtype=tf.float32, seed=None, name=None)` ###

### `tf.random_shuffle(value, seed=None, name=None)` ###

效果程序：

<table style="border-collapse:collapse; border:none; margin:0px; padding:0px; table-layout:fixed"> 
 <tbody> 
 <tr> 
 <td style="border:none; margin:0px; padding:0px; vertical-align:top; width:50px; background:rgb(238,238,238)"> <pre style="white-space: pre-wrap; word-wrap: break-word; border: none; border-radius: 0px; font-family: Menlo, 'Liberation Mono', Consolas, 'Courier New', 'andale mono', 'lucida console', monospace; margin-top: 0px; margin-bottom: 0px; padding: 10px 6px 10px 0px; width: auto; color: rgb(102, 102, 102); text-align: right; line-height: 16px !important; background: rgb(238, 238, 238);"><a href="http://blog.csdn.net/phdat101/article/details/52442738#L1" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 1</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L2" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 2</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L3" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 3</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L4" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 4</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L5" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 5</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L6" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 6</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L7" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 7</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L8" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 8</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L9" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 9</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L10" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 10</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L11" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 11</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L12" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 12</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L13" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 13</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L14" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 14</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L15" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 15</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L16" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 16</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L17" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 17</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L18" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 18</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L19" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 19</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L20" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 20</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L21" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 21</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L22" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 22</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L23" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 23</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L24" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 24</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L25" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 25</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L26" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 26</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L27" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 27</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L28" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 28</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L29" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 29</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L30" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 30</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L31" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 31</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L32" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 32</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L33" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 33</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L34" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 34</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L35" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 35</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L36" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 36</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L37" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 37</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L38" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 38</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L39" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 39</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L40" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 40</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L41" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 41</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L42" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 42</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L43" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 43</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L44" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 44</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L45" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 45</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L46" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 46</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L47" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 47</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L48" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 48</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L49" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 49</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L50" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 50</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L51" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 51</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L52" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 52</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L53" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 53</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L54" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 54</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L55" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 55</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L56" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 56</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L57" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 57</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L58" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 58</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L59" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 59</a>
<a href="http://blog.csdn.net/phdat101/article/details/52442738#L60" style="color: rgb(102, 102, 102); text-decoration: none;" rel="nofollow"> 60</a></pre></td> 
 <td style="border:none; margin:0px; padding:0px; vertical-align:top"> 
 <div style="border-left-width:1px; border-left-style:solid; border-left-color:rgb(222,226,227); overflow-x:auto; overflow-y:hidden"> 
 <pre style="word-wrap:normal; color:rgb(51,51,51); border:none; font-family:Menlo,'Liberation Mono',Consolas,'Courier New','andale mono','lucida console',monospace; margin-top:0px; margin-bottom:0px; padding:10px 0px; line-height:16px!important; background-color:rgb(255,255,255)">
 
 
 <div style="padding:0px 10px">
 
 
 import 
 
 
 tensorflow 
 
 
 as 
 
 
 tf
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 import 
 
 
 numpy 
 
 
 as 
 
 
 np
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px">
 
 
 # 生成0和1矩阵
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v1 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 Variable
 
 
 (
 
 
 tf
 
 
 .
 
 
 zeros
 
 
 ([
 
 
 3
 
 
 ,
 
 
 3
 
 
 ,
 
 
 3
 
 
 ]), 
 
 
 name
 
 
 =
 
 
 "v1"
 
 
 )
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v2 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 Variable
 
 
 (
 
 
 tf
 
 
 .
 
 
 ones
 
 
 ([
 
 
 10
 
 
 ,
 
 
 5
 
 
 ]), 
 
 
 name
 
 
 =
 
 
 "v2"
 
 
 )
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px">
 
 
 #填充单值矩阵
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v3 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 Variable
 
 
 (
 
 
 tf
 
 
 .
 
 
 fill
 
 
 ([
 
 
 2
 
 
 ,
 
 
 3
 
 
 ], 
 
 
 9
 
 
 ))
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px">
 
 
 #常量矩阵
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v4_1 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 constant
 
 
 ([
 
 
 1
 
 
 , 
 
 
 2
 
 
 , 
 
 
 3
 
 
 , 
 
 
 4
 
 
 , 
 
 
 5
 
 
 , 
 
 
 6
 
 
 , 
 
 
 7
 
 
 ])
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v4_2 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 constant
 
 
 (
 
 
 -
 
 
 1.0
 
 
 , 
 
 
 shape
 
 
 =
 
 
 [
 
 
 2
 
 
 , 
 
 
 3
 
 
 ])
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px">
 
 
 #生成等差数列
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v6_1 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 linspace
 
 
 (
 
 
 10.0
 
 
 , 
 
 
 12.0
 
 
 , 
 
 
 30
 
 
 , 
 
 
 name
 
 
 =
 
 
 "linspace"
 
 
 )
 
 
 #float32 or float64
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v7_1 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 range
 
 
 (
 
 
 10
 
 
 , 
 
 
 20
 
 
 , 
 
 
 3
 
 
 )
 
 
 #just int32
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px">
 
 
 #生成各种随机数据矩阵
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v8_1 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 Variable
 
 
 (
 
 
 tf
 
 
 .
 
 
 random_uniform
 
 
 ([
 
 
 2
 
 
 ,
 
 
 4
 
 
 ], 
 
 
 minval
 
 
 =
 
 
 0.0
 
 
 , 
 
 
 maxval
 
 
 =
 
 
 2.0
 
 
 , 
 
 
 dtype
 
 
 =
 
 
 tf
 
 
 .
 
 
 float32
 
 
 , 
 
 
 seed
 
 
 =
 
 
 1234
 
 
 , 
 
 
 name
 
 
 =
 
 
 "v8_1"
 
 
 ))
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v8_2 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 Variable
 
 
 (
 
 
 tf
 
 
 .
 
 
 random_normal
 
 
 ([
 
 
 2
 
 
 ,
 
 
 3
 
 
 ], 
 
 
 mean
 
 
 =
 
 
 0.0
 
 
 , 
 
 
 stddev
 
 
 =
 
 
 1.0
 
 
 , 
 
 
 dtype
 
 
 =
 
 
 tf
 
 
 .
 
 
 float32
 
 
 , 
 
 
 seed
 
 
 =
 
 
 1234
 
 
 , 
 
 
 name
 
 
 =
 
 
 "v8_2"
 
 
 ))
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v8_3 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 Variable
 
 
 (
 
 
 tf
 
 
 .
 
 
 truncated_normal
 
 
 ([
 
 
 2
 
 
 ,
 
 
 3
 
 
 ], 
 
 
 mean
 
 
 =
 
 
 0.0
 
 
 , 
 
 
 stddev
 
 
 =
 
 
 1.0
 
 
 , 
 
 
 dtype
 
 
 =
 
 
 tf
 
 
 .
 
 
 float32
 
 
 , 
 
 
 seed
 
 
 =
 
 
 1234
 
 
 , 
 
 
 name
 
 
 =
 
 
 "v8_3"
 
 
 ))
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v8_4 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 Variable
 
 
 (
 
 
 tf
 
 
 .
 
 
 random_uniform
 
 
 ([
 
 
 2
 
 
 ,
 
 
 3
 
 
 ], 
 
 
 minval
 
 
 =
 
 
 0.0
 
 
 , 
 
 
 maxval
 
 
 =
 
 
 1.0
 
 
 , 
 
 
 dtype
 
 
 =
 
 
 tf
 
 
 .
 
 
 float32
 
 
 , 
 
 
 seed
 
 
 =
 
 
 1234
 
 
 , 
 
 
 name
 
 
 =
 
 
 "v8_4"
 
 
 ))
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 v8_5 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 random_shuffle
 
 
 ([[
 
 
 1
 
 
 ,
 
 
 2
 
 
 ,
 
 
 3
 
 
 ],[
 
 
 4
 
 
 ,
 
 
 5
 
 
 ,
 
 
 6
 
 
 ],[
 
 
 6
 
 
 ,
 
 
 6
 
 
 ,
 
 
 6
 
 
 ]], 
 
 
 seed
 
 
 =
 
 
 134
 
 
 , 
 
 
 name
 
 
 =
 
 
 "v8_5"
 
 
 )
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px">
 
 
 # 初始化
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 init_op 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 initialize_all_variables
 
 
 ()
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px">
 
 
 # 保存变量，也可以指定保存的内容
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 saver 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 train
 
 
 .
 
 
 Saver
 
 
 ()
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 #saver = tf.train.Saver({"my_v2": v2})
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px">
 
 
 #运行
 
 
 </div>
 
 
 <div style="padding:0px 10px">
 
 
 with 
 
 
 tf
 
 
 .
 
 
 Session
 
 
 () 
 
 
 as 
 
 
 sess
 
 
 :
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 sess
 
 
 .
 
 
 run
 
 
 (
 
 
 init_op
 
 
 )
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 # 输出形状和值
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 print 
 
 
 tf
 
 
 .
 
 
 Variable
 
 
 .
 
 
 get_shape
 
 
 (
 
 
 v1
 
 
 )
 
 
 #shape
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 print 
 
 
 sess
 
 
 .
 
 
 run
 
 
 (
 
 
 v1
 
 
 )
 
 
 #vaule
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 # numpy保存文件
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 np
 
 
 .
 
 
 save
 
 
 (
 
 
 "v1.npy"
 
 
 ,
 
 
 sess
 
 
 .
 
 
 run
 
 
 (
 
 
 v1
 
 
 ))
 
 
 #numpy save v1 as file
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 test_a 
 
 
 = 
 
 
 np
 
 
 .
 
 
 load
 
 
 (
 
 
 "v1.npy"
 
 
 )
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 print 
 
 
 test_a
 
 
 [
 
 
 1
 
 
 ,
 
 
 2
 
 
 ]
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px"> 
 
 
 #一些输出
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 print 
 
 
 sess
 
 
 .
 
 
 run
 
 
 (
 
 
 v3
 
 
 )
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px"> 
 
 
 v5 
 
 
 = 
 
 
 tf
 
 
 .
 
 
 zeros_like
 
 
 (
 
 
 sess
 
 
 .
 
 
 run
 
 
 (
 
 
 v1
 
 
 ))
 
 
 </div>
 
 
 <div style="padding:0px 10px">

</div>
 
 
 <div style="padding:0px 10px"> 
 
 
 print 
 
 
 sess
 
 
 .
 
 
 run
 
 
 (
 
 
 v8_5
 
 
 )
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 #保存图的变量
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 save_path 
 
 
 = 
 
 
 saver
 
 
 .
 
 
 save
 
 
 (
 
 
 sess
 
 
 , 
 
 
 "/tmp/model.ckpt"
 
 
 )
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 #加载图的变量
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 #saver.restore(sess, "/tmp/model.ckpt")
 
 
 </div>
 
 
 <div style="padding:0px 10px"> 
 
 
 print 
 
 
 "Model saved in file: "
 
 
 , 
 
 
 save_path
 
 
 </div></pre> 
 </div> </td> 
 </tr> 
 </tbody> 
</table>

##### [ 来自CODE的代码片][CODE 1]  variables.py      #####

[CODE]: https://code.csdn.net/snippets/1867350
[CODE 1]: https://code.csdn.net/snippets/1867529