Fix datatype issue with Sparse Attention softmax#363
Merged
Conversation
Fixes a dataype issue with softmax where the number of blocks being sent to the Triton kernel source was a torch.Tensor but should have been a python integer. On some environments (e.g., conda) this resulted in triton not knowing how to serialize the input (and crashing in our tests). Once switching to the correct datatype that triton expects this seems to have solved the issue.
jeffra
requested review from
RezaYazdaniAminabadi,
ShadenSmith,
arashashari,
awan-10,
cli99,
conglongli,
eltonzheng,
minjiaz,
niumanar,
samyam and
tjruwase
as code owners
arashashari
approved these changes
Sep 4, 2020
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes a dataype issue with softmax where the number of blocks being sent to the Triton kernel source was a torch.Tensor but should have been a python integer. On some environments (e.g., conda) this resulted in triton not knowing how to serialize the input (and crashing in our tests). Once switching to the correct datatype that triton expects this seems to have solved the issue.
In the future it would be great if we can add python type hints across DeepSpeed to help with these types of issues. https://docs.python.org/3/library/typing.html