Question

我正在学习TensorFLow。因此，要了解如何制作某些内容，我尝试从源代码中复制一些代码并执行它。但是我遇到了错误信息。所以我尝试了一些来自这个网站的解决方案，但它不起作用（我在评论中保留了我的测试）。

    """programme 1 """
import tensorflow as tf 
import numpy as np

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, 28, 28, 1]) #28 * 28 taille image 1 = 1pixel car noir et blanc "X" valeur
W = tf.Variable(tf.zeros([784, 10])) # 28*28 = 784 , 10 -> 0 à 9  "W" = weight = poid
b = tf.Variable(tf.zeros([10])) #chiffre de 0 à 9 a reconnaitre "b" = constante 
init = tf.initialize_all_variables()

#model
Y = tf.nn.softmax(tf.matmul(tf.reshape(X,[-1, 784]), W) + b) #fonction "matmul": produit matriciel "-1": reussite obligatoire

#Place holder
Y_ = tf.placeholder(tf.float32, [None, 10])

#loss function
cross_entropy = -1 * tf.reduce_sum(Y_ * tf.log(Y)) #formule

# % of correct annwer found in batch
is_correct = tf.equal(tf.argmax(Y,1),tf.argmax(Y_,1))
accuracy = tf.reduce_mean(tf.cast(is_correct,tf.float32))

#training step
optimizer = tf.train.GradientDescentOptimizer(0.003) #petit pas
train_step = optimizer.minimize(cross_entropy)

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

for i in range(10000):
    #load batch of image and ocrrects answer
    batch_X, batch_Y = mnist.train.next_batch(100)
    batch_X = np.reshape(batch_X, (-1, 784))
    #batch_Y = np.reshape(batch_Y, (-1, 784))

    train_data = {X: batch_X, Y_: batch_Y}

    #train
    sess.run(train_step, feed_dict=train_data)

    a,c = sess.run([accuracy,cross_entropy],feed = train_data)

    test_data = {X:mnist.test.images, Y_:mnist.test.labels}
    a,c = sess.run([accuracy,cross_entropy],feed = test_data)

日志：

    Traceback (most recent call last):
  File "d:\tensorflow\test1.py", line 46, in <module>
    sess.run(train_step, feed_dict=train_data)
  File "C:\Users\Proprietaire\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\client\session.py", line 895, in run
    run_metadata_ptr)
  File "C:\Users\Proprietaire\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\client\session.py", line 1100, in _run
    % (np_val.shape, subfeed_t.name, str(subfeed_t.get_shape())))
ValueError: Cannot feed value of shape (100, 784) for Tensor 'Placeholder:0', which has shape '(?, 28, 28, 1)'
2017-08-30 19:07:37.406994: W C:\tf_jenkins\home\workspace\rel-win\M\windows\PY\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.

第46行是

sess.run(train_step, feed_dict=train_data)

我该怎么做才能解决此错误？

Answer 1

您收到该错误是因为您正在喂食的形状与TensorFlow期望的形状不匹配。要解决此问题，您可能需要在placeholder:0 batch_X重新整形数据 (?, 28, 28, 1)。例如，您将执行以下操作：

batch_X = np.reshape(batch_X, (-1, 28, 28, 1))

Answer 2

你需要重塑X。

    X = tf.placeholder(tf.float32 , [None ,28 , 28 , 1])
    X = tf.reshape(X , [-1 , 784])

Answer 3

在TensorFlow中定义占位符时，会话期间输入的形状应与占位符的形状相同。

在.archive-link中，batch_X, batch_Y = mnist.train.next_batch(100)是 2D 像素值数组，其形状为batch_x。

在[batch_size, 28*28]中，输入占位符定义为 4D 形状X = tf.placeholder(tf.float32,[None, 28, 28, 1])

你可以1）在喂养TensorFlow之前重塑[batch_size, 28, 28, 1]。例如batch_x或2）更改占位符的形状。例如batch_x = np.reshape(batch_x, [-1, 28, 28, 1])

我建议2），因为这样可以避免在TensorFlow图形内外进行任何重塑操作。

TensorFlow无法为Tensor＆＃39; Placeholder：0＆＃39;提供形状值（100,784）。

3 个答案: