# Bug when indexing tensors using an MPS device

**URL:** <https://discuss.pytorch.org/t/bug-when-indexing-tensors-using-an-mps-device/201613>\
**Category:** mps\
**Created:** [April 25, 2024, 6:39pm UTC](https://discuss.pytorch.org/t/bug-when-indexing-tensors-using-an-mps-device/201613 "2024-04-25T18:39:51Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![joaquimgomez](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/joaquimgomez/32/69577_2.png) [@joaquimgomez](https://discuss.pytorch.org/u/joaquimgomez)\
**Post date:** [April 25, 2024, 6:39pm UTC](https://discuss.pytorch.org/t/bug-when-indexing-tensors-using-an-mps-device/201613/1 "2024-04-25T18:39:51Z")

</div>

Hi there.

I detected this possible bug, but before raising an issue in the repo I’m asking here.

When executing the following code for a CPU or a CUDA device:

```auto
device = torch.device("cpu")

x = torch.tensor([[1, 0, 2, 0, -100, -100, -100],
                  [1, 0, 2, 0, -100, -100, -100]])
mask = torch.tensor([[True, True, True, True, False, False, False],
                     [True, True, True, True, False, False, False]])

x = x.to(device)
mask = mask.to(device)

print(x[mask])

```

x is `tensor([1, 0, 2, 0, 1, 0, 2, 0])`, what it’s correct.

But, when using MPS as a device, the result is `tensor([-100, 0, 2, 0, 1, 0, 2, 0], device='mps:0')`, which is not correct.

This does not happen when x is 1D nor when the mask is in the CPU. But it also happens using `torch.masked_select`.

If this is the expected behavior, could someone explain it to me, please? And is there some workaround?

I’m using PyTorch version 2.2.2 and a MacBook Pro with an M2 Max and macOS Sonoma 14.4.1.

Thanks.