Reputation: 24201
Say I have a tensor A
and a container of values vals
. Is there a clean way of returning a Boolean tensor of the same shape as A
with each element being whether that element of A
is contained within vals
? e.g:
A = torch.tensor([[1,2,3],
[4,5,6]])
vals = [1,5]
# Desired output
torch.tensor([[True,False,False],
[False,True,False]])
Upvotes: 4
Views: 3921
Reputation: 121
Use torch.isin
method is the most convinient way. It's simple as follows: torch.isin(A, vals)
Upvotes: 6
Reputation: 175
You can simply do like this:
result = A.apply_(lambda x: x in vals).bool()
Then result
will contain this tensor:
tensor([[ True, False, False],
[False, True, False]])
Here I simply used a lambda function and the apply_ method that you can find in the official documentation.
Upvotes: 3