Tensorflow 1.0:老司机立下的Flag

作者: Double_E | 来源:发表于2017-04-05 23:24 被阅读1472次

    老司机怎么能不会立FLAG呢·

    • 如何利用FLAG设置参数(包括超参数和一些路径设置),详细语法介绍见代码注释
    • 利用name_scope管理变量
      ** 1. 方便查看tensorboard,比上一讲的清晰多啦
      **
      ** 2.方便。。
      **
    # -*- coding: utf-8 -*-
    """
    Created on Mon Apr  3 19:15:24 2017
    
    @author: Jhy_BUPT
    README:
    
    REF:
    
    """
    import tensorflow as tf
    # 载入数据
    from tensorflow.examples.tutorials.mnist import input_data
    mnist = input_data.read_data_sets("C:\\tmp\\data", one_hot=True)
    
    # 老司机立下FLAGS
    FLAGS = tf.app.flags.FLAGS
    # 根据参数类型不同,可以定义字符串DEFINE_string,可以定义整数DEFINE_integer
    #DEFINE_integer('参数名称',  参数值, '参数的解释')
    tf.app.flags.DEFINE_integer('batch_size', 100, 'batch size')
    tf.app.flags.DEFINE_float('learning_rate', 0.1, 'l;earning rate')
    tf.app.flags.DEFINE_integer('epoch', 1000, 'how many epoch tf will run')
    tf.app.flags.DEFINE_string('ckp_dir', 'C:\\tmp\\d44', 'where to save tensorboard')
    
    with tf.name_scope('input'):
        x = tf.placeholder(tf.float32, [None, 784], name='x_input')
        y_ = tf.placeholder(tf.float32, [None, 10], name='y_input')
    
    with tf.name_scope('nn'):
        with tf.name_scope('weight'):
            W = tf.Variable(tf.zeros([784, 10]))
        with tf.name_scope('bias'):
            b = tf.Variable(tf.zeros([10]))
        y = tf.nn.softmax(tf.matmul(x, W) + b)
    
    with tf.name_scope('loss'):
        cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y))
    
    with tf.name_scope('trian'):
        train_step = tf.train.GradientDescentOptimizer(FLAGS.learning_rate).minimize(cross_entropy)
    
    # 开启会话
    sess = tf.InteractiveSession()
    tf.global_variables_initializer().run()
    
    for _ in range(FLAGS.epoch):
        batch_xs, batch_ys = mnist.train.next_batch(FLAGS.batch_size)
        fd = {x: batch_xs, y_: batch_ys}
        sess.run(train_step, feed_dict=fd)
    correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
    accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
    fd2 = {x: mnist.test.images, y_: mnist.test.labels}
    print(sess.run(accuracy, feed_dict=fd2))
    
    merged = tf.summary.merge_all()
    train_writer = tf.summary.FileWriter(FLAGS.ckp_dir, sess.graph)
    

    tensorboard 变量展示,还可打开哦 开箱有惊喜

    Paste_Image.png Paste_Image.png

    相关文章

      网友评论

        本文标题:Tensorflow 1.0:老司机立下的Flag

        本文链接:https://www.haomeiwen.com/subject/puyxattx.html