根据numpy的reshape文档
One shape dimension can be -1. In this case, the value is inferred from the length of the array and remaining dimensions.
举个例子
z = np.array([[1, 2, 3, 4],
[5, 6, 7, 8],
[9, 10, 11, 12],
[13, 14, 15, 16]])
z.shape
(4, 4)
z.reshape(-1)
array([ 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16])
z.reshape(-1,1)
array([[ 1],
[ 2],
[ 3],
[ 4],
[ 5],
[ 6],
[ 7],
[ 8],
[ 9],
[10],
[11],
[12],
[13],
[14],
[15],
[16]])
Numpy自动计算出有16行,新的数组shape属性为(16, 1),与原来的(4, 4)配套。
z.reshape(-1, 2)
array([[ 1, 2],
[ 3, 4],
[ 5, 6],
[ 7, 8],
[ 9, 10],
[11, 12],
[13, 14],
[15, 16]])
同理。