conv1 = tf.nn.conv1d(input,w,2,'VALID') #2为步长. print(conv1.shape)#宽度计算(width-kernel_size+1)/strides ,(10-3+1)/2=4 (64,4,32).
確定! 回上一頁