Added dim check for index_copy_ #21617
Conversation
facebook-github-bot
left a comment
facebook-github-bot
left a comment
There was a problem hiding this comment.
@izdeby has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
facebook-github-bot
left a comment
facebook-github-bot
left a comment
There was a problem hiding this comment.
@izdeby has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
|
Test fail seems legit: |
aten/src/ATen/native/Indexing.cpp
Outdated
| if (source.dim() == 0 && numIndices != 1) { | ||
| AT_INDEX_ERROR("index_copy_(): When source is scalar, index should have one element (got ", numIndices, ")"); | ||
| } else if (source.dim() != self.dim()) { | ||
| AT_INDEX_ERROR("index_copy_(): When source is not scalar, its dimensionality must match the dimensionality of the destination"); |
There was a problem hiding this comment.
It's nice to also list the dimensionalities so the user doesn't have to print out the tensors.
Something like:
index_copy_(): When source is not scalar, its dimensionality (3) must match the dimensionality of the destination (2)"
There was a problem hiding this comment.
also, the error message doesn't match what is checked -- the message says source is not a scalar, but this isn't actually checked.
ezyang
requested review from
VitalyFedyunin and
zou3519
and removed request for
ezyang
VitalyFedyunin
left a comment
VitalyFedyunin
left a comment
There was a problem hiding this comment.
Test fail seems legit: ERROR: test_index_copy_scalar_input_dim (__main__.TestAutograd)
izdeby
changed the title
izdeby
changed the title
facebook-github-bot
left a comment
facebook-github-bot
left a comment
There was a problem hiding this comment.
@izdeby has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
|
why does this not have a test? |
Summary: Fixing reported [bug](pytorch/pytorch#20322) The issue was related to not checking the dimensions of source vs destination tensors. Pull Request resolved: pytorch/pytorch#21617 Differential Revision: D15749963 Pulled By: izdeby fbshipit-source-id: acff114c729fd9c0a9a51325e0ebd8b42e1f2fc1
8 participants
Fixing reported bug
The issue was related to not checking the dimensions of source vs destination tensors.