Ensure ScalarToTensorPass returns correct modified value. #16058
+55
−41
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
Summary: This change adds support for the `eliminate_quant_dequant_pairs` flag in the remove clone ops transform. This flag allows users to control whether quantization-dequantization pairs should be eliminated during the clone removal optimization pass. We need this to control the functionality of this pass in the subsequent diffs. Differential Revision: D88092217
Summary: This change fixes an issue where the `RemovePermutesAroundElementwiseOps` pass was not correctly returning the modified flag to indicate whether the graph was changed during the optimization. The modified flag is crucial for the pass manager to determine if additional passes need to be run or if the optimization has reached a fixed point. Without properly returning this flag, the pass manager may not iterate correctly or may run unnecessary additional passes. Differential Revision: D88109865
…rted Summary: This change eliminates code duplication by replacing the local `RemoveCloneOpPass` implementation with the imported `RemoveCloneOpsTransformImported` from the executorch transforms module. Previously, there were multiple implementations of the clone removal pass scattered across the codebase. This consolidation improves maintainability by ensuring a single source of truth for the clone removal optimization logic. Differential Revision: D88109862
Summary: This change reorganizes the pass execution order by moving the remove passes (clone removal, permute removal, etc.) into the iterative section of the pass manager. Previously, these passes were executed in a single-shot manner. By moving them to the iterative section, the pass manager can now run these optimization passes repeatedly until a fixed point is reached, where no further optimizations can be applied. This is important because removing one operation may enable additional removal opportunities that weren't visible before. Differential Revision: D88109852
Differential Revision: D88215978
eigen-k
requested review from
JacobSzwejbka,
kimishpatel and
larryliu0820
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/16058
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New FailureAs of commit 59cbddc with merge base 144a37d ( NEW FAILURE - The following job has failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
This PR needs a
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Differential Revision: D88215978