tensorflow保存模型和导入模型

基于tf 1.12.0版本
保存模型：

import tensorflow as tf 
import numpy as np

## 保存模型
W = tf.Variable([[1,2,3],[3,2,1]],dtype=tf.float32, name="weights")
b = tf.Variable([[7,7,7]], dtype=tf.float32, name="biases")
saver = tf.train.Saver()
with tf.Session() as sess:
    # 重要：需要初始化变量
    sess.run(tf.global_variables_initializer())
    save_path = saver.save(sess, "model_save/model.ckpt") # 保存之后返回 保存的路径
    print("save to path:", save_path)

save to path: model_save/model.ckpt

$ ll model_save/
total 16
-rw-r--r-- 1 root root   77 Jan 13 10:49 checkpoint
-rw-r--r-- 1 root root   36 Jan 13 10:49 model.ckpt.data-00000-of-00001
-rw-r--r-- 1 root root  161 Jan 13 10:49 model.ckpt.index
-rw-r--r-- 1 root root 3440 Jan 13 10:49 model.ckpt.meta

导入模型：

import tensorflow as tf 
import numpy as np

## 导入模型：
## 因为保存的只是参数，所以要先建立结构一样的模型
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")
# 不需要初始化这一步
saver = tf.train.Saver()
with tf.Session() as sess:
    saver.restore(sess, "model_save/model.ckpt")
    print("Weights:", sess.run(W))
    print("biases:", sess.run(b))

Weights: [[1. 2. 3.]
          [3. 2. 1.]]
biases: [[7. 7. 7.]]