对比 tf.layers.conv2d_transpose和tf.nn.conv2d_transpose区别

都表示反卷积，但是参数不同。使用方法对比如下：

import tensorflow as tf  

x1 = tf.ones(shape=[64,7,7,256])  

y1 = tf.layers.conv2d_transpose(x1, 128, 3, strides=2, padding='SAME')  

w=tf.ones([3,3,128,256])
y2 = tf.nn.conv2d_transpose(x1, w, output_shape=[64,14,14,128], strides=[1,2,2,1], padding='SAME')


x2=tf.nn.conv2d(y2,w,strides=[1,2,2,1],padding='SAME')
with tf.Session() as sess:  
    sess.run(tf.global_variables_initializer())  
    y1_value,y2_value,x2_value=sess.run([y1,y2,x2])