fix bug with sequential backends #10708
Open
+126
−20
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: https://github.com/pytorch/executorch/pull/10584/files#r2070213706 there's a bug described in this PR comment. I add some tests and a fix to cover it. Essentially when sequential partitions go through preprocess_all, the get_item nodes from the first partition in the sequence don't correctly get mapped to the arguments input into the second partition. This is because the name of these nodes change (the original node to a get_item node). Instead of checking for the names, we instead delete the nodes we know must be deleted from the inputspec Additionaly, there is an issue with validation. the _validate fails when there are call_module nodes still in the graph. Since preprocess_multimethod will lower the call_submodule nodes one-by-one calling _validate before all the call_submodule nodes are transformed to call_delegate nodes will fail. We remove the _validate call from unsafe_adjust_original_program and instead call _validate on the original program after all the submodule nodes have been converted to call_delegate Differential Revision: D74226258
mcr229
requested review from
JacobSzwejbka,
tarun292 and
larryliu0820
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/10708
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ❌ 20 New FailuresAs of commit 118f575 with merge base 4dfddf5 ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D74226258 |
This PR needs a
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Summary:
https://github.com/pytorch/executorch/pull/10584/files#r2070213706
there's a bug described in this PR comment. I add some tests and a fix to cover it. Essentially when sequential partitions go through preprocess_all, the get_item nodes from the first partition in the sequence don't correctly get mapped to the arguments input into the second partition. This is because the name of these nodes change (the original node to a get_item node). Instead of checking for the names, we instead delete the nodes we know must be deleted from the inputspec
Additionaly, there is an issue with validation. the _validate fails when there are call_module nodes still in the graph. Since preprocess_multimethod will lower the call_submodule nodes one-by-one calling _validate before all the call_submodule nodes are transformed to call_delegate nodes will fail. We remove the _validate call from unsafe_adjust_original_program and instead call _validate on the original program after all the submodule nodes have been converted to call_delegate
Differential Revision: D74226258