Question

我是使用tensorflow的新手，并且在处理它时遇到了一些困难。我尝试使用类似于MNIS示例的softmax模型进行简单的分类工作。

我尝试创建批量数据并将dem放入run方法中。我的第一种方法是使用

sess.run(train_step, feed_dict={x: feature_batch, y_: labels_batch})

导致了不能将张量传递给feed_dict的错误。

经过一些研究，我发现我应该使用。

feat, lab = sess.run([feature_batch, feature_batch])
sess.run(train_step, feed_dict={x: feat, y_: lab})

尝试后，我的脚本不会终止计算，但也不会打印任何错误。

有没有人提示为什么它不起作用？

孔文件如下：

def input_pipeline(filename='dataset.csv', batch_size=30, num_epochs=None):
    filename_queue = tf.train.string_input_producer([filename], num_epochs=num_epochs, shuffle=True)
    features, labels = read_from_cvs(filename_queue)

    min_after_dequeue = 10000
    capacity = min_after_dequeue + 3 * batch_size
    feature_batch, label_batch = tf.train.shuffle_batch(
        [features, labels], batch_size=batch_size, capacity=capacity,
        min_after_dequeue=min_after_dequeue)
    return feature_batch, label_batch


def tensorflow():
    x = tf.placeholder(tf.float32, [None, num_attributes])
    W = tf.Variable(tf.zeros([num_attributes, num_types]))
    b = tf.Variable(tf.zeros([num_types]))

    y = tf.nn.softmax(tf.matmul(x, W) + b)
    y_ = tf.placeholder(tf.float32, [None, num_types])
    cross_entropy = tf.reduce_mean(
        tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y))
    train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)

    sess = tf.InteractiveSession()

    tf.global_variables_initializer().run()

    coord = tf.train.Coordinator()
    threads = tf.train.start_queue_runners(coord=coord)

    feature_batch, label_batch = input_pipeline()

    for _ in range(1200):
        feat, lab = sess.run([feature_batch, feature_batch])
        sess.run(train_step, feed_dict={x: feat, y_: lab})

    coord.request_stop()
    coord.join(threads)

    correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
    accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
    #print(sess.run(accuracy, feed_dict={x: feature_batch, y_: label_batch}))

Answer 1

您可以在模型定义中直接使用张量。例如：

def tensorflow():
    x, y_ = input_pipeline()
    W = tf.Variable(tf.zeros([num_attributes, num_types]))
    b = tf.Variable(tf.zeros([num_types]))

    y = tf.nn.softmax(tf.matmul(x, W) + b)
    cross_entropy = tf.reduce_mean(
        tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y))
    train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)

    sess = tf.InteractiveSession()

    tf.global_variables_initializer().run()

    coord = tf.train.Coordinator()
    threads = tf.train.start_queue_runners(coord=coord)



    for _ in range(1200):
        sess.run(train_step)

或者您应该在tf.train.shuffle_batch中使用占位符。例如：

#...omit
features_placeholder = tf.placeholder(...)
labels_placeholder = tf.placeholder(...)
x, y_ = tf.train.shuffle_batch(
        [features_placeholder, labels_placeholder], batch_size=batch_size, capacity=capacity,
        min_after_dequeue=min_after_dequeue)
W = tf.Variable(tf.zeros([num_attributes, num_types]))
b = tf.Variable(tf.zeros([num_types]))
#...omit
for _ in range(1200):
    sess.run(train_step, feed_dict={features_placeholder: ..., labels_placeholder: ...})

Answer 2

我怀疑问题是这两行的顺序：

threads = tf.train.start_queue_runners(coord=coord)

feature_batch, label_batch = input_pipeline()

对tf.train.start_queue_runners()的调用将为已定义到那一点的所有输入管道阶段启动后台线程。对input_pipeline()的调用会创建两个新的输入管道阶段（在tf.train.string_input_producer()和tf.train.shuffle_batch()的调用中）。这意味着两个新阶段的后台线程将不会启动，程序将挂起。

解决方案是颠倒这些行的顺序：

feature_batch, label_batch = input_pipeline()

threads = tf.train.start_queue_runners(coord=coord)

Tensorflow不会使用批次终止

2 个答案: