获取连续最小比较的索引

Question

说我是否有张量

values = torch.tensor([5., 4., 8., 3.])

我想在每2个连续值中取最小值，这意味着

min(5., 4.) = 4.
min(8., 3.) = 3.

是否有矢量化的方式来做到这一点，但仍能获得最小值的相对指数？我要作为输出的意思是：

min_index = [1, 1]
#min_index[0] == 1 as 4. is the minimum of (5., 4.) and is in index 1 of (5., 4.)
#min_index[1] == 1 as 3. is the minimum of (8., 3.) and is in index 1 of (8., 3.) 

Answer 1

我认为重塑张量将使它变得容易得多。 之后，max-width: 100vw; overflow-x: hidden; 自动返回最小值和索引。

torch.min

Answer 2

这里有一个numpy实现：

Supplier<Stream<? extends Long>> longStreamSupplier = ..