Pytorch mask tensor with boolean numpy array

Question

I have a 84x84 pytorch tensor named target. I need to mask it with an 84x84 boolean numpy array which consists of True and False.

When I do target = target[mask], I get the error TypeError: can't convert np.ndarray of type numpy.bool_. The only supported types are: double, float, float16, int64, int32, and uint8.

Surprisingly, I get this error only when running on a GPU. When running on a CPU, everything works fine. How can I fix this?

What about the answer provided here: discuss.pytorch.org/t/… ? — s.ouchène
– s.ouchène, Commented Jan 5, 2020 at 23:11

sascha10000 · Accepted Answer · 2020-01-06 11:19:17Z

2

I think there is some confusion with the types. But this works.

import torch
tensor = torch.randn(84,84)
c = torch.randn(tensor.size()).bool()
c[1, 2:5] = False
x = tensor[c].size()

For testing I created a tensor with random values. Afterwards 3 elements are set to False. In the last step I look get the size 7053 resulting from 84^2 - 3.

Hope that helps somehow.

answered Jan 6, 2020 at 11:19

sascha10000

1,2401 gold badge10 silver badges23 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Pytorch mask tensor with boolean numpy array

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related