tf.varible_scope()和tf.AUTO_REUSE

痛定思痛。 2022-05-14 11:24 165阅读 0赞

class G():
        def __init__(self):
            with tf.variable_scope('a'):
                self.a=tf.Variable([[[0.1,0.2,0.5,0.2],
                    [0.1,0.1,0.1,0.7]],
    
                   [[0.2,0.3,0.2,0.3],
                    [0.2,0.2,0.2,0.4]]])
            with tf.variable_scope('c'):
                outputs = self.conv(self.a)
                self.outputs = self.conv(outputs)
                outputs = self.conv(outputs)
                self.outputs1 = self.conv(outputs)
        def conv(self,inputs,scope="m",reuse=tf.AUTO_REUSE):
            # with tf.variable_scope(scope, reuse=reuse):
                outputs=tf.layers.conv1d(inputs=inputs,filters=3,kernel_size=1,activation=None,use_bias=False,
                                         kernel_initializer=tf.ones_initializer())
                outputs = tf.layers.conv1d(inputs=outputs, filters=4, kernel_size=1, activation=None, use_bias=False,kernel_initializer=tf.ones_initializer())
                # outputs=outputs #+ inputs
                return outputs
    
        def normalize(self,inputs, epsilon=1e-8, scope="ln", reuse=None):
    
            with tf.variable_scope(scope, reuse=reuse):
                inputs_shape = inputs.get_shape()
                params_shape = inputs_shape[-1:]
    
                mean, variance = tf.nn.moments(inputs, [-1], keep_dims=True)
                beta = tf.Variable(tf.zeros(params_shape))
                gamma = tf.Variable(tf.ones(params_shape))
                normalized = (inputs - mean) / ((variance + epsilon) ** (.5))
                outputs = gamma * normalized + beta
    
            return outputs
    g=G()
    a=g.a
    outputs=g.outputs
    outputs1=g.outputs1
    
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        print(a.name)
        print(sess.run(outputs))
        print('o',outputs.name)
        print(sess.run(outputs1))
        print('o1',outputs1.name)