[JIT] parse prim::Constant[value=annotate()] and prim::Constant[value={0}] #76875
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
…={0}]
irparser previously didn't support these, which would cause failures in
log_extract.py
[ghstack-poisoned]
🔗 Helpful links
❌ 1 New FailuresAs of commit f4d51a9 (more details on the Dr. CI page): Expand to see more
🕵️ 1 new failure recognized by patternsThe following CI failures do not appear to be due to upstream breakages
|
…stant[value={0}]"
irparser previously didn't support these, which would cause failures in
log_extract.py
[ghstack-poisoned]
|
@davidberard98 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
…stant[value={0}]"
irparser previously didn't support these, which would cause failures in
log_extract.py
Differential Revision: [D36156699](https://our.internmc.facebook.com/intern/diff/D36156699)
[ghstack-poisoned]
…stant[value={0}]"
irparser previously didn't support these, which would cause failures in
log_extract.py
Differential Revision: [D36156699](https://our.internmc.facebook.com/intern/diff/D36156699)
[ghstack-poisoned]
|
@davidberard98 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
eellison
reviewed
May 5, 2022
| graph = torch._C.parse_ir(ir, True) | ||
| func = torch._C._create_function_from_graph("forward", graph) | ||
| ret = func() | ||
| self.assertTrue(ret.numel() == 1) |
There was a problem hiding this comment.
can you also check that len(ret.size() == 0)
There was a problem hiding this comment.
actually it's ret.size() == 1 for this tensor - added a check
| %3 : int[] = prim::Constant[value=annotate(List[int], [])]() | ||
| return (%3) | ||
| """ | ||
| graph = torch._C.parse_ir(ir, True) |
There was a problem hiding this comment.
nit, check result somehow ? maybe with
%1 : bool = prim::isinstance[types=[int[]]](%3)
return (%1)
There was a problem hiding this comment.
hmm that didn't work because... it can't parse List[int] 😄
I'm just checking == [] right now, lmk if you have another suggestion on how to do this
torch/csrc/jit/ir/irparser.cpp
Outdated
| << "Unexpected annotation (only List and Dict can be parsed)"; | ||
| } | ||
| L.next(); | ||
| // use the profiling annotation instead of the `annotate()` type. |
There was a problem hiding this comment.
profiling is too specific, since types could come elsewhere. I would say something like ignore the annotations on the IValue constants, and recover type from the Node output
…stant[value={0}]"
irparser previously didn't support these, which would cause failures in
log_extract.py
Differential Revision: [D36156699](https://our.internmc.facebook.com/intern/diff/D36156699)
[ghstack-poisoned]
|
@davidberard98 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
|
@pytorchbot merge this please |
|
@pytorchbot revert this pls It broke Torchvision per the linked issue. |
pytorchmergebot
added a commit
that referenced
this pull request
May 11, 2022
…nt[value={0}]"
This reverts commit 31d3ce7.
Reverted #76875 on behalf of https://github.com/janeyx99
davidberard98
added a commit
that referenced
this pull request
May 12, 2022
…t[value={0}]"
Retry of #76875. It was reverted
due to torchvision failures, but it turned out that the failures were
caused by a different PR.
irparser previously didn't support these, which would cause failures in
log_extract.py
[ghstack-poisoned]
pytorchmergebot
pushed a commit
that referenced
this pull request
May 13, 2022
…t[value={0}]"
Retry of #76875. It was reverted
due to torchvision failures, but it turned out that the failures were
caused by a different PR.
irparser previously didn't support these, which would cause failures in
log_extract.py
Pull Request resolved: #77377
Approved by: https://github.com/datumbox
facebook-github-bot
pushed a commit
that referenced
this pull request
May 13, 2022
…={0}] (#76875)
Summary:
irparser previously didn't support these, which would cause failures in
log_extract.py
Pull Request resolved: #76875
Approved by: https://github.com/eellison, https://github.com/tugsbayasgalan
Test Plan: contbuild & OSS CI, see https://hud.pytorch.org/commit/pytorch/pytorch/31d3ce70005251860d78d3994b73894c14b067c5
Reviewed By: eellison, tugsbayasgalan
Differential Revision: D36156699
fbshipit-source-id: 501df8f0f7744c55ce572468e38f61367108131a
facebook-github-bot
pushed a commit
that referenced
this pull request
May 13, 2022
…nt[value={0}]"
Summary:
This reverts commit 31d3ce7.
Reverted #76875 on behalf of https://github.com/janeyx99
Test Plan: contbuild & OSS CI, see https://hud.pytorch.org/commit/pytorch/pytorch/2083b16f68963e563c6fbf6bf1084e78a7f66139
Reviewed By: atalman
Differential Revision: D36322437
Pulled By: atalman
fbshipit-source-id: 2d48ffa603e5b1f85f2bc493741b3279af474a96
facebook-github-bot
pushed a commit
that referenced
this pull request
May 15, 2022
…t[value={0}]" (#77377)
Summary:
Retry of #76875. It was reverted
due to torchvision failures, but it turned out that the failures were
caused by a different PR.
irparser previously didn't support these, which would cause failures in
log_extract.py
Pull Request resolved: #77377
Approved by: https://github.com/datumbox
Test Plan: contbuild & OSS CI, see https://hud.pytorch.org/commit/pytorch/pytorch/fa44e165ff5f616ae69c0d8dbca757f8d8a9159f
Reviewed By: atalman
Differential Revision: D36386459
Pulled By: davidberard98
fbshipit-source-id: b2ada187e94b8280020f4454e44981f6e45922e9
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:
irparser previously didn't support these, which would cause failures in
log_extract.py
Differential Revision: D36156699