在CIFAR10
示例中,conv2
定義如下。如何知道shape=[5,5,64,64]
在kernel = _variable_with_weight_decay中應該給出這些值,例如5,5,64,64
另外,在biases = _variable_on_cpu('biases', [64], tf.constant_initializer(0.1))
中,shape還被定義爲[64]
,如何獲取這些值?關於CIFAR10的卷積層中的形狀值示例
# conv2
with tf.variable_scope('conv2') as scope:
kernel = _variable_with_weight_decay('weights',
shape=[5, 5, 64, 64],
stddev=5e-2,
wd=0.0)
conv = tf.nn.conv2d(norm1, kernel, [1, 1, 1, 1], padding='SAME')
biases = _variable_on_cpu('biases', [64], tf.constant_initializer(0.1))
bias = tf.nn.bias_add(conv, biases)
conv2 = tf.nn.relu(bias, name=scope.name)
_activation_summary(conv2)