正文共3565个字，预计阅读时间10分钟。

saver = tf.train.Saver() save_path = saver.save(sess, model_path)

saver = tf.train.Saver() load_path = saver.restore(sess, model_path)

import tensorflow as tf

import sys

# load MNIST data

from tensorflow.examples.tutorials.mnist import input_data

mnist = input_data.read_data_sets('data', one_hot=True)

# 一些 hyper parameters

activation = tf.nn.relu batch_size = 100

iteration = 20000

hidden1_units = 30

# 注意！这里是存储路径！

model_path = sys.path[0] + '/simple_mnist.ckpt'

X = tf.placeholder(tf.float32, [None, 784]) y_ = tf.placeholder(tf.float32, [None, 10]) W_fc1 = tf.Variable(tf.truncated_normal([784, hidden1_units], stddev=0.2)) b_fc1 = tf.Variable(tf.zeros([hidden1_units])) W_fc2 = tf.Variable(tf.truncated_normal([hidden1_units, 10], stddev=0.2)) b_fc2 = tf.Variable(tf.zeros([10]))

def inference(img):    fc1 = activation(tf.nn.bias_add(tf.matmul(img, W_fc1), b_fc1))    logits = tf.nn.bias_add(tf.matmul(fc1, W_fc2), b_fc2)

return logits

def loss(logits, labels):    cross_entropy = tf.nn.softmax_cross_entropy_with_logits(logits, labels)    loss = tf.reduce_mean(cross_entropy)

return loss

def evaluation(logits, labels):

correct_prediction = tf.equal(tf.argmax(logits, 1), tf.argmax(labels, 1))    accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

return accuracy logits = inference(X) loss = loss(logits, y_) train_op = tf.train.AdamOptimizer(1e-4).minimize(loss) accuracy = evaluation(logits, y_)

# 先实例化一个Saver()类saver = tf.train.Saver() init = tf.initialize_all_variables()

with tf.Session() as sess:    sess.run(init)    for i in xrange(iteration):        batch = mnist.train.next_batch(batch_size)

if i%1000 == 0 and i:            train_accuracy = sess.run(accuracy, feed_dict={X: batch[0], y_: batch[1]})

print "step %d, train accuracy %g" %(i, train_accuracy)

sess.run(train_op, feed_dict={X: batch[0], y_: batch[1]})

print '[+] Test accuracy is %f' % sess.run(accuracy, feed_dict={X: mnist.test.images, y_: mnist.test.labels})

# 存储训练好的variables

save_path = saver.save(sess, model_path)

print "[+] Model saved in file: %s" % save_path

import tensorflow as tf

import sys

from tensorflow.examples.tutorials.mnist import input_data

mnist = input_data.read_data_sets('data', one_hot=True)

activation = tf.nn.relu hidden1_units = 30

model_path = sys.path[0] + '/simple_mnist.ckpt'

X = tf.placeholder(tf.float32, [None, 784]) y_ = tf.placeholder(tf.float32, [None, 10]) W_fc1 = tf.Variable(tf.truncated_normal([784, hidden1_units], stddev=0.2)) b_fc1 = tf.Variable(tf.zeros([hidden1_units])) W_fc2 = tf.Variable(tf.truncated_normal([hidden1_units, 10], stddev=0.2)) b_fc2 = tf.Variable(tf.zeros([10]))

def inference(img):    fc1 = activation(tf.nn.bias_add(tf.matmul(img, W_fc1), b_fc1))    logits = tf.nn.bias_add(tf.matmul(fc1, W_fc2), b_fc2)

return logits

def evaluation(logits, labels):    correct_prediction = tf.equal(tf.argmax(logits, 1), tf.argmax(labels, 1))    accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

return accuracy logits = inference(X) accuracy = evaluation(logits, y_) saver = tf.train.Saver()with tf.Session() as sess:

# 读取之前训练好的数据    load_path = saver.restore(sess, model_path)

print "[+] Model restored from %s" % load_path

print '[+] Test accuracy is %f' % sess.run(accuracy, feed_dict={X: mnist.test.images, y_: mnist.