Remove aten::binary_cross_entropy_with_logits from ATen Fallback#12301
Merged
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
PyTorch removed "binary_cross_entropy_with_logits_backward" from its native function list so that we cannot fallback aten::binary_cross_entropy_with_logits to ATen since we cannot build the gradient graph using that function easily. We have exporter logic for "binary_cross_entropy_with_logits ", when swtiched back to use the original exporter logic, it still works well for both forward and backward.
The ATen fallback for this function is from this PR: #9670. At that time we saw customer issue when using original exporter logic. Not sure if this issue is still valid for the customer, if yes, we need to revisit this, fix it, either by changing model script, or exporter logic. @zhijxu-MS