TensorFlow-模型的保存和调用（ckpt方式）

版权声明：本文为博主原创文章，欢迎转载。 https://blog.csdn.net/samylee/article/details/85066855

TensorFlow-模型的保存和调用（ckpt方式）

硬件：NVIDIA-GTX1080

软件：Windows7、python3.6.5、tensorflow-gpu-1.4.0

一、基础知识

1、checkpoint：模型文本信息

2、meta：模型graph，调用时可重载入

3、index、data：模型数据

二、代码展示

1、保存模型

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.global_variables_initializer()

#define saver
saver = tf.train.Saver()

with tf.Session() as sess:
    sess.run(init)

    #save ckpt
    save_path = saver.save(sess, "my_net/save_net.ckpt")
    print("Save to path: ", save_path)

2、调用模型

import tensorflow as tf
import numpy as np

with tf.Session() as sess:
    # restore graph
    saver = tf.train.import_meta_graph('my_net/save_net.ckpt.meta')
    
    #restore ckpt
    saver.restore(sess, "my_net/save_net.ckpt")

    # check variable W and b, like weight or bias
    print("weights:", sess.run('weights:0'))
    print("biases:", sess.run('biases:0'))

任何问题请加唯一QQ2258205918（名称samylee）！