首页 文章

如何在tensorflow中交换'mnist.train.next_batch'到'tf.train.batch'？

提问于
浏览 次
1

张量流1.2

我开始使用tf中的mnist数据集学习tensorflow .

我改变了数据集 . 我遇到了将'batch_xs，batch_ys = mnist.train.next_batch（batch_size）'交换为'batch_xs，batch_ys = tf.train.batch（[X，Y]，batch_size = batch_size）'的问题 .

我想知道如何在tensorflow中应用minibatch .

错误消息

TypeError：Feed的值不能是tf.Tensor对象 . 可接受的Feed值包括Python标量，字符串，列表，numpy ndarrays或TensorHandles .

train_data.shape, train_labels.shape # numpy
# ((10000, 20, 20, 3), (10000, 2))


X = tf.placeholder(tf.float32, [10000, 20, 20, 3])
Y = tf.placeholder(tf.float32, [10000, 10])

W1 = tf.Variable(tf.random_normal([4, 4, 3, 32], stddev=0.01))

L1 = tf.nn.conv2d(X, W1, strides=[1, 1, 1, 1], padding = 'SAME')
L1 = tf.nn.relu(L1)
L1 = tf.nn.max_pool(L1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
L1 = tf.reshape(L1, [-1, 10 * 10 * 32])

W2 = tf.get_variable('W2', shape=[10 * 10 * 32, 10], initializer=tf.contrib.layers.xavier_initializer())
b = tf.Variable(tf.random_normal([10]))

hypothesis = tf.matmul(L1, W2) + b

learning_rate = 0.001
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=hypothesis, labels = Y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)

# hyper parameters
learning_rate = 0.001
training_epochs = 5
batch_size = 100

for epoch in range(training_epochs):
    avg_cost = 0
    total_batch = int(10000 / batch_size)
    for i in range(total_batch):
        # batch_xs, batch_ys = tf.train.batch([X, Y], batch_size)
        batch_xs, batch_ys = tf.train.batch([X, Y], batch_size = batch_size)
        feed_dict = {X: batch_xs, Y: batch_ys}
        c, _, = sess.run([cost, optimizer], feed_dict = feed_dict)
        avg_cost += c / total_batch
    print('Epoch: ', '%04d' % (epoch + 1), 'cost: ', '{:.9f}'.format(avg_cost))



TypeError: The value of a feed cannot be a tf.Tensor object. Acceptable feed values include Python scalars, strings, lists, numpy ndarrays, or TensorHandles.
python tensorflow neural-network deep-learning

2 回答

  • 0

    batch_xs, batch_ys = tf.train.batch([X, Y], batch_size = batch_size) 此声明仍处于构建图阶段 . 除非你对他们打电话 sess.run(...) ，否则什么都不会发生 .

    您可以访问tensorflow doc和教程来应用minibatch .

    回复于
  • 0

    您无法在图表中添加任何张量流API，例如您在我的代码中执行的 batch_xs, batch_ys = tf.train.batch([X, Y], batch_size = batch_size) .

    它使batch_xs和batch_ys成为张量 . 您可以做的是在图表中编写 batch_xs, batch_ys = tf.train.batch([X, Y], batch_size = batch_size) ，评估batch_xs和batch_ys，然后在流程中提供 .

    回复于

相关问题