tensorflow学习（4）之saver保存读取

Bertha 。 2022-02-20 05:15 198阅读 0赞

莫凡2017tensorflow（使用更简便版）[https://github.com/MorvanZhou/Tensorflow-Tutorial][https_github.com_MorvanZhou_Tensorflow-Tutorial]

怎样用 Tensorflow 中的 saver 保存和加载

[https://morvanzhou.github.io/tutorials/machine-learning/tensorflow/5-06-save/][https_morvanzhou.github.io_tutorials_machine-learning_tensorflow_5-06-save]

如果保存搭建好的神经网络：

1.保存

`import`所需的模块, 然后建立神经网络当中的 `W` 和 `b`, 并初始化变量.

import tensorflow as tf
    import numpy as np
    
    ## Save to file
    # remember to define the same dtype and shape when restore
    W = tf.Variable([[1,2,3],[3,4,5]], dtype=tf.float32, name='weights')
    b = tf.Variable([[1,2,3]], dtype=tf.float32, name='biases')
    
    # init= tf.initialize_all_variables() # tf 马上就要废弃这种写法
    # 替换成下面的写法:
    init = tf.global_variables_initializer()

保存时, 首先要建立一个 `tf.train.Saver()` 用来保存, 提取变量. 再创建一个名为`my_net`的文件夹, 用这个 `saver` 来保存变量到这个目录 `"my_net/save_net.ckpt"`.

saver = tf.train.Saver()
    
    with tf.Session() as sess:
        sess.run(init)
        save_path = saver.save(sess, "my_net/save_net.ckpt")
        print("Save to path: ", save_path)
    
    """    
    Save to path:  my_net/save_net.ckpt
    """

2.提取

提取时, 先建立零时的`W` 和 `b`容器. 找到文件目录, 并用`saver.restore()`我们放在这个目录的变量.

# 先建立 W, b 的容器
    W = tf.Variable(np.arange(6).reshape((2, 3)), dtype=tf.float32, name="weights")
    b = tf.Variable(np.arange(3).reshape((1, 3)), dtype=tf.float32, name="biases")
    
    # 这里不需要初始化步骤 init= tf.initialize_all_variables()
    
    saver = tf.train.Saver()
    with tf.Session() as sess:
        # 提取变量
        saver.restore(sess, "my_net/save_net.ckpt")
        print("weights:", sess.run(W))
        print("biases:", sess.run(b))
    
    """
    weights: [[ 1.  2.  3.]
              [ 3.  4.  5.]]
    biases: [[ 1.  2.  3.]]
    """

[https_github.com_MorvanZhou_Tensorflow-Tutorial]: https://github.com/MorvanZhou/Tensorflow-Tutorial
[https_morvanzhou.github.io_tutorials_machine-learning_tensorflow_5-06-save]: https://morvanzhou.github.io/tutorials/machine-learning/tensorflow/5-06-save/