[jit] fix handling of function attributes. #28569
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
Previously, the inclusion of function attributes would "poison" a ConcreteModuleType, because we did not have a way of checking whether they are actually the same function. This PR uses the Python function object to perform that check. This improves our ability to reuse JIT types between modules. Also this PR fixes a bug where we weren't properly adding modules as attributes when converting from ConcreteType -> JIT type (we were adding them after the fact--another reason to switch from using `register_x` to `set_x` during module construction, which is on my to-do list after this). Fixes #28559
suo
added a commit
that referenced
this pull request
Oct 24, 2019
Previously, the inclusion of function attributes would "poison" a ConcreteModuleType, because we did not have a way of checking whether they are actually the same function. This PR uses the Python function object to perform that check. This improves our ability to reuse JIT types between modules. Also this PR fixes a bug where we weren't properly adding modules as attributes when converting from ConcreteType -> JIT type (we were adding them after the fact--another reason to switch from using `register_x` to `set_x` during module construction, which is on my to-do list after this). Fixes #28559 ghstack-source-id: 50dedca Pull Request resolved: #28569
facebook-github-bot
added
the
oncall: jit
Previously, the inclusion of function attributes would "poison" a ConcreteModuleType, because we did not have a way of checking whether they are actually the same function. This PR uses the Python function object to perform that check. This improves our ability to reuse JIT types between modules. Also this PR fixes a bug where we weren't properly adding modules as attributes when converting from ConcreteType -> JIT type (we were adding them after the fact--another reason to switch from using `register_x` to `set_x` during module construction, which is on my to-do list after this). Fixes #28559 Differential Revision: [D18111331](https://our.internmc.facebook.com/intern/diff/D18111331)
suo
added a commit
that referenced
this pull request
Oct 24, 2019
Previously, the inclusion of function attributes would "poison" a ConcreteModuleType, because we did not have a way of checking whether they are actually the same function. This PR uses the Python function object to perform that check. This improves our ability to reuse JIT types between modules. Also this PR fixes a bug where we weren't properly adding modules as attributes when converting from ConcreteType -> JIT type (we were adding them after the fact--another reason to switch from using `register_x` to `set_x` during module construction, which is on my to-do list after this). Fixes #28559 ghstack-source-id: 44bca50 Pull Request resolved: #28569
driazati
approved these changes
Oct 24, 2019
| def __init__(self): | ||
| super().__init__() | ||
| encoder_norm = nn.ReLU() | ||
| self.encoder = N(encoder_norm) |
Contributor
There was a problem hiding this comment.
encoder_norm here is a module instead of a function, modules seemed to already work fine
Member
Author
There was a problem hiding this comment.
oh my attempt to minify this test case went wrong. I'll add back the function attr
| fn1_mod = M(fn) | ||
| fn2_mod = M(fn) | ||
|
|
||
| self.assertSameType(fn1_mod, fn2_mod) |
Contributor
There was a problem hiding this comment.
Can you add a test that they’re not the same if fn is changed to something else?
Member
Author
There was a problem hiding this comment.
that's test_python_function_attribute_different
test/jit/test_type_sharing.py
Outdated
|
|
||
| def test_python_function_attribute_same(self): | ||
| """ | ||
| Different functions passed in should lead to different types |
| cls->addAttribute(name, type, isParameter); | ||
| } | ||
|
|
||
| for (const auto& moduleInfo : modules_) { |
Contributor
There was a problem hiding this comment.
What is this adding to cls? Is modules_ all callables attached to this type or is it just sub-ScriptModules?
Previously, the inclusion of function attributes would "poison" a ConcreteModuleType, because we did not have a way of checking whether they are actually the same function. This PR uses the Python function object to perform that check. This improves our ability to reuse JIT types between modules. Also this PR fixes a bug where we weren't properly adding modules as attributes when converting from ConcreteType -> JIT type (we were adding them after the fact--another reason to switch from using `register_x` to `set_x` during module construction, which is on my to-do list after this). Fixes #28559 Differential Revision: [D18111331](https://our.internmc.facebook.com/intern/diff/D18111331)
Closed
Contributor
thiagocrepaldi
pushed a commit
to thiagocrepaldi/pytorch
that referenced
this pull request
Feb 4, 2020
Summary: Pull Request resolved: pytorch#28569 Previously, the inclusion of function attributes would "poison" a ConcreteModuleType, because we did not have a way of checking whether they are actually the same function. This PR uses the Python function object to perform that check. This improves our ability to reuse JIT types between modules. Also this PR fixes a bug where we weren't properly adding modules as attributes when converting from ConcreteType -> JIT type (we were adding them after the fact--another reason to switch from using `register_x` to `set_x` during module construction, which is on my to-do list after this). Fixes pytorch#28559 Test Plan: Imported from OSS Differential Revision: D18111331 Pulled By: suo fbshipit-source-id: ec2cccf832d3ddd4cd4d28fe19cb265f1275325a
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.
Stack from ghstack:
Previously, the inclusion of function attributes would "poison" a
ConcreteModuleType, because we did not have a way of checking whether
they are actually the same function. This PR uses the Python function
object to perform that check. This improves our ability to reuse JIT
types between modules.
Also this PR fixes a bug where we weren't properly adding modules as
attributes when converting from ConcreteType -> JIT type (we were adding
them after the fact--another reason to switch from using
register_xtoset_xduring module construction, which is on my to-do list afterthis).
Fixes #28559
Pull Request resolved: #28569
Differential Revision: D18111331