# sharing variables # Tensorflow supports two ways of sharing variables # 1、Explicitly passing tf.Variable objects around # 2、Implicitly wrapping tf.Variable objects within tf.variable_scope objects # For example , let's write a function to create a convolutional / relu layer import tensorflow as tf def conv_relu(input, kernel_shape, bias_shape): # Create variable named "weights" weights = tf.get_variable("weights", kernel_shape, initializer=tf.random_normal_initializer()) # Create variable named "biases" biases = tf.get_variable("biases", bias_shape, initializer=tf.constant_initializer(0.0)) # stride表示步长 , weight表示卷积核 , padding 表示卷积核是否可以停留在图像的边缘 conv = tf.nn.conv2d(input, weights, strides=[1, 1, 1, 1], padding='SAME') return tf.nn.relu(conv + biases) # 这个函数使用了简写的weights和biases # 在声明的时候是有帮助的, # 但是在真实的模型中,我们更想要以下的卷积层,重复调用这个函数是不会起作用的 input1 = tf.random_normal([1, 10, 10, 32]) input2 = tf.random_normal([1, 20, 20, 32]) x = conv_relu(input1, kernel_shape=[5, 5, 32, 32], bias_shape=32) # This fails 不支持重复调用 , 因为想要的行为还不清楚 # 是创建新的变量还是使用现有的变量 # x = conv_relu(x, kernel_shape=[5, 5, 32, 32], bias_shape=[32])