tensorflowでcnnをしようとモデルを作っていたのですがうまくできていないので質問させていただきます.

以下コードです

import numpy as np
import tensorflow as tf
import tensorflow.python.platform

def inference(image_placeholder, keep_prob):

  #define weight functioln
  def weight_variable(shape):
    initial = tf.truncated_normal(shape, stddev = 0.1)
      return tf.Variable(initial)

  #define bias function
  def bias_variable(shape):
    initial = tf.constant(0.1, shape = shape)
    return tf.Variable(initial)

  #define ConvNet function
  def conv2d(x,W):
    return tf.nn.conv2d(x,W,strides = [1,1,1,1],padding = 'VALID')

  #define max pooling
  def max_pool_nxn(x,n):
    return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')

  #　reshape input array to 1,32,32,3
  x_images = tf.reshape(image_placeholder, [-1,32,32,3])

  #define model

  #ConvNet 1
  with tf.name_scope('conv1') as scope:
    W_conv1 = weight_variable([3, 3, 3, 32])
    b_conv1 = bias_variable([32])
    h_conv1 = tf.nn.relu(
            conv2d(x_images, W_conv1) +  b_conv1
        )

  #Pooling1
  with tf.name_scope('pool1') as scope:
    h_pool1 = max_pool_nxn(h_conv1,2)

  #ConvNet2
  with tf.name_scope('conv2') as scope:
    W_conv2 = weight_variable([3,3,32,64])
    b_conv2 = bias_variable([64])
    h_conv2 = tf.nn.relu(
            conv2d(h_pool1, W_conv2) +  b_conv2
        )

  #Pooling2
  with tf.name_scope('pool2') as scope:
    h_pool2 = max_pool_nxn(h_conv2,2)

  #　fully conected layer 1
  with tf.name_scope('fc1') as scope:
    W_fc1 = weight_variable([6 * 6 * 64, 1024])
    b_fc1 = bias_variable([1024])
    h_pool2_flat = tf.reshape(h_pool2, [-1, 6 * 6 * 64])
    h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
    h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)

  # fully conected layer 2
  with tf.name_scope('fc2') as scope:
    W_fc2 = weight_variable([1024,20])
    b_fc2 = bias_variable([20])

  # softmax layer
  with tf.name_scope('softmax') as scope:
    y_conv = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)

  return y_conv

def loss(logits, labels):
  cross_entropy = -tf.reduce_sum(labels * 
  tf.log(tf.clip_by_value(logits, 1e-10, 1.0)))
  tf.summary.scalar("cross_entropy", cross_entropy)
  return cross_entropy

def training(loss):
  train_step = 
  tf.train.AdamOptimizer(0.01).minimize(loss)
  return train_step

def accuracy(logits, labels):
  correct_prediction = tf.equal(tf.argmax(logits, 1), 
  tf.argmax(labels, 1))
  accuracy = tf.reduce_mean(tf.cast(correct_prediction, 
  "float"))
  tf.summary.scalar("accuracy", accuracy)
  return accuracy

以下学習のコードです

image_placeholder = tf.placeholder(tf.float32, shape = (400, 32 * 32 * 3))
label_placeholder = tf.placeholder(tf.float32, shape = (400, 20))
keep_prob = tf.placeholder(tf.float32)


logits = inference(image_placeholder, keep_prob)
loss_value = loss(logits,label_placeholder)
train_op = training(loss_value)
acc = accuracy(logits, label_placeholder)



with tf.Session() as sess:
  sess.run(tf.global_variables_initializer())
  for epoch in range(epochs):
    for batch in range(1250):
        x,t = train_generator.__next__()
        sess.run(train_op, feed_dict = {
            image_placeholder : x,
            label_placeholder   : t,
            keep_prob : 0.5
        })

    if epoch % 10 == 0:
        x_validarion, t_validation = validation_generator.__next__()
        train_accuracy = sess.run(acc, feed_dict = {
            image_placeholder : x_validation,
            label_placeholder : t_validation,
            keep_prob : 1.
        })

        print("step: {} , accracy{}".format(epoch + 1, train_accuracy))

x_test, t_test = test_generator.__next__()
test_acc = sess.run(acc, feed_dict = {
    image_placeholder : x_test,
    label_placeholder : t_test,
    keep_prob : 1.0
})
print("test accuracy : {}".format(test_acc))

以下エラーです.

ValueError                                Traceback 
(most recent call last)
<ipython-input-19-1528997ebea5> in <module>()
19                 image_placeholder : x,
20                 label_placeholder   : t,
---> 21                 keep_prob : 0.5
22             })
23 
/usr/local/lib/python3.6/dist- 
packages/tensorflow/python/client/session.py in run(self, fetches, feed_dict, options, run_metadata)
927     try:
928       result = self._run(None, fetches, feed_dict, options_ptr,
--> 929                          run_metadata_ptr)
930       if run_metadata:
931         proto_data = 
tf_session.TF_GetBuffer(run_metadata_ptr)

/usr/local/lib/python3.6/dist- 
packages/tensorflow/python/client/session.py in _run(self, handle, fetches, feed_dict, options, run_metadata)
1126                              'which has shape %r' %
1127                              (np_val.shape, subfeed_t.name,
-> 1128                               
str(subfeed_t.get_shape())))
1129           if not self.graph.is_feedable(subfeed_t):
1130             raise ValueError('Tensor %s may not be fed.' % subfeed_t)

ValueError: Cannot feed value of shape (400, 32, 32, 3) for Tensor 'Placeholder_21:0', which has shape '(400, 3072)'

モデルのでconv2dで出力のシェイプが悪いと思われるのですが,リファレンスを読んでみても何が間違っているのかわかりませんでした.
よろしくお願いいたします.