MNIST For ML Beginners - Tensorflow Tutorials

mnist.py

# -*- coding: utf-8 -*-
import numpy as np
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)

x = tf.placeholder(tf.float32, [None, 784])
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))

y = tf.nn.softmax(tf.matmul(x, W) + b)

y_ = tf.placeholder(tf.float32, [None, 10])
cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices=[1]))
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)

init = tf.global_variables_initializer()

sess = tf.Session()
sess.run(init)

for i in range(1000):
    batch_xs, batch_ys = mnist.train.next_batch(100)
    sess.run(train_step, feed_dict={x: batch_xs, y_: batch_ys})
    if i % 100 == 0:
        correct_prediction = tf.equal(tf.argmax(y,1), tf.argmax(y_,1))
        accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
        print("{0:3d} times\taccuracy: {1:.10f} %".format(i+100, sess.run(accuracy, feed_dict={x: mnist.test.images, y_: mnist.test.labels})*100))
        

L28 のprint().format()は、print("".format())でした。

http://stackoverflow.com/questions/28378257/attributeerror-nonetype-object-has-no-attribute-format