TensorFlow 度量张量和张量或者和零之间的误差值

　　用于一个回归任务或者正则问题

 # l2损失，output= sum(x ** 2)/2
 inputdata = tf.Variable(np.random.rand(2,3), dtype=np.float32)
 loss = tf.nn.l2_loss(inputdata)    # 标量
 with tf.Session() as sess:
     sess.run(tf.global_variables_initializer())
     print (sess.run(inputdata))
     print (sess.run(loss))

输出：

[[ 0.99520844 0.57105571 0.4224354 ]
[ 0.77398151 0.40279421 0.6592328 ]]
1.34544