support non-persistent buffers #1817
Closed
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
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/1817
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (22 Unrelated Failures)As of commit a97ce2d with merge base b2636ff ( FLAKY - The following jobs failed but were likely due to flakiness present on trunk:
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: D53340041 |
Summary: Basic support for non-persistent buffers, which are buffers that do not show up in the state dict. One weird twist is that most of our other systems (FX, aot_export, dynamo) have completely buggy handling of non-persistent buffers. I tried to go on a wild goose chase to fix them all, but it got to be too much. So I introduced some sad rewrite passes in `_export` make the final state dict correctly align with the original module's state dict. This exposed some bugs/ambiguous handling of parameters/buffers in existing test code. For example, `TestSaveLoad.test_save_buffer` traced over a module that was not in the root module hierarchy and caused some weird behavior. I think we should error explicitly on use cases like this: pytorch/pytorch#118410. For now I just rewrote the tests or skipped them. As a side effect, this diff tightened up quite a few sloppy behaviors around state dict handling: - Tensor attributes were getting promoted to be buffers—bad! - Tracing through a module not in the children of the root module would add its parameters/buffers to the state dict—bad! This behavior is unlikely to show up in user code since the model would be totally broken, but did show up in a bunch of tests. #buildmore Differential Revision: D53340041
|
This pull request was exported from Phabricator. Differential Revision: D53340041 |
huydhn
approved these changes
Feb 2, 2024
suo
added a commit
to suo/pytorch
that referenced
this pull request
Feb 2, 2024
Summary: Pull Request resolved: pytorch#118969 X-link: pytorch/executorch#1817 Basic support for non-persistent buffers, which are buffers that do not show up in the state dict. One weird twist is that most of our other systems (FX, aot_export, dynamo) have completely buggy handling of non-persistent buffers. I tried to go on a wild goose chase to fix them all, but it got to be too much. So I introduced some sad rewrite passes in `_export` make the final state dict correctly align with the original module's state dict. This exposed some bugs/ambiguous handling of parameters/buffers in existing test code. For example, `TestSaveLoad.test_save_buffer` traced over a module that was not in the root module hierarchy and caused some weird behavior. I think we should error explicitly on use cases like this: pytorch#118410. For now I just rewrote the tests or skipped them. As a side effect, this diff tightened up quite a few sloppy behaviors around state dict handling: - Tensor attributes were getting promoted to be buffers—bad! - Tracing through a module not in the children of the root module would add its parameters/buffers to the state dict—bad! This behavior is unlikely to show up in user code since the model would be totally broken, but did show up in a bunch of tests. #buildmore Test Plan: unit tests sandcastle Differential Revision: D53340041 fbshipit-source-id: 0dcba4843608a7bb6017b54289ee88d3da8ab41c
|
This pull request has been merged in c49e1ee. |
pytorchmergebot
pushed a commit
to pytorch/pytorch
that referenced
this pull request
Feb 2, 2024
Summary: X-link: pytorch/executorch#1817 Basic support for non-persistent buffers, which are buffers that do not show up in the state dict. One weird twist is that most of our other systems (FX, aot_export, dynamo) have completely buggy handling of non-persistent buffers. I tried to go on a wild goose chase to fix them all, but it got to be too much. So I introduced some sad rewrite passes in `_export` make the final state dict correctly align with the original module's state dict. This exposed some bugs/ambiguous handling of parameters/buffers in existing test code. For example, `TestSaveLoad.test_save_buffer` traced over a module that was not in the root module hierarchy and caused some weird behavior. I think we should error explicitly on use cases like this: #118410. For now I just rewrote the tests or skipped them. As a side effect, this diff tightened up quite a few sloppy behaviors around state dict handling: - Tensor attributes were getting promoted to be buffers—bad! - Tracing through a module not in the children of the root module would add its parameters/buffers to the state dict—bad! This behavior is unlikely to show up in user code since the model would be totally broken, but did show up in a bunch of tests. #buildmore Test Plan: unit tests sandcastle Differential Revision: D53340041 Pull Request resolved: #118969 Approved by: https://github.com/guangy10, https://github.com/huydhn, https://github.com/titaiwangms
pytorch-bot bot
pushed a commit
to pytorch/pytorch
that referenced
this pull request
Feb 8, 2024
Summary: X-link: pytorch/executorch#1817 Basic support for non-persistent buffers, which are buffers that do not show up in the state dict. One weird twist is that most of our other systems (FX, aot_export, dynamo) have completely buggy handling of non-persistent buffers. I tried to go on a wild goose chase to fix them all, but it got to be too much. So I introduced some sad rewrite passes in `_export` make the final state dict correctly align with the original module's state dict. This exposed some bugs/ambiguous handling of parameters/buffers in existing test code. For example, `TestSaveLoad.test_save_buffer` traced over a module that was not in the root module hierarchy and caused some weird behavior. I think we should error explicitly on use cases like this: #118410. For now I just rewrote the tests or skipped them. As a side effect, this diff tightened up quite a few sloppy behaviors around state dict handling: - Tensor attributes were getting promoted to be buffers—bad! - Tracing through a module not in the children of the root module would add its parameters/buffers to the state dict—bad! This behavior is unlikely to show up in user code since the model would be totally broken, but did show up in a bunch of tests. #buildmore Test Plan: unit tests sandcastle Differential Revision: D53340041 Pull Request resolved: #118969 Approved by: https://github.com/guangy10, https://github.com/huydhn, https://github.com/titaiwangms
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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.
Summary:
Basic support for non-persistent buffers, which are buffers that do not show up in the state dict.
One weird twist is that most of our other systems (FX, aot_export, dynamo) have completely buggy handling of non-persistent buffers. I tried to go on a wild goose chase to fix them all, but it got to be too much. So I introduced some sad rewrite passes in
_exportmake the final state dict correctly align with the original module's state dict.This exposed some bugs/ambiguous handling of parameters/buffers in existing test code. For example,
TestSaveLoad.test_save_buffertraced over a module that was not in the root module hierarchy and caused some weird behavior. I think we should error explicitly on use cases like this: pytorch/pytorch#118410. For now I just rewrote the tests or skipped them.As a side effect, this diff tightened up quite a few sloppy behaviors around state dict handling:
This behavior is unlikely to show up in user code since the model would be totally broken, but did show up in a bunch of tests.
#buildmore
Differential Revision: D53340041