[ts_converter] Fix prim::If buffer names #136648

angelayi · 2024-09-25T17:33:54Z

Summary:
We previously incorrectly handled the following graph, specifically for the node w.3 in block0:

 graph(%x.1 : Float(3, strides=[1], requires_grad=0, device=cpu),
       %y.1 : int):
   %2 : __torch__.___torch_mangle_1.M = prim::CreateObject()
   %3 : int = prim::Constant[value=20](), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:747:34
   %4 : int = prim::Constant[value=10](), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:746:34
   %5 : int = prim::Constant[value=1](), scope: M::
   %w.1 : int = prim::GetAttr[name="w"](%2), scope: M::
   %7 : int = aten::mul(%w.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:746:25
    = prim::SetAttr[name="w"](%2, %7), scope: M::
   %h.1 : int = prim::GetAttr[name="h"](%2), scope: M::
   %9 : int = aten::mul(%h.1, %3), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:747:25
    = prim::SetAttr[name="h"](%2, %9), scope: M::
   %10 : bool = aten::gt(%y.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:749:19
   %res.37 : Tensor = prim::If(%10), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:749:16
     block0():
       %w.3 : int = prim::GetAttr[name="w"](%2), scope: M::
       %res.1 : Tensor = aten::add(%x.1, %w.3, %5), scope: M:: # <string>:5:9
       -> (%res.1)
     block1():
       %h.3 : int = prim::GetAttr[name="h"](%2), scope: M::
       %res.3 : Tensor = aten::add(%x.1, %h.3, %5), scope: M:: # <string>:5:9
       -> (%res.3)
   %16 : bool = aten::lt(%y.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:754:19
   %res : Tensor = prim::If(%16), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:754:16
     block0():
       %w : int = prim::GetAttr[name="w"](%2), scope: M::
       %res.15 : Tensor = aten::add(%res.37, %w, %5), scope: M:: # <string>:5:9
       -> (%res.15)
     block1():
       %h : int = prim::GetAttr[name="h"](%2), scope: M::
       %res.21 : Tensor = aten::add(%res.37, %h, %5), scope: M:: # <string>:5:9
       -> (%res.21)
   return (%res)

Test Plan: CI

Differential Revision: D63399064

pytorch-bot · 2024-09-25T17:33:58Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/136648

📄 Preview Python docs built from this PR
📄 Preview C++ docs built from this PR
❓ Need help or want to give feedback on the CI? Visit the bot commands wiki or our office hours

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 New Failure, 1 Unrelated Failure

As of commit 96c2827 with merge base 99eb47f ():

NEW FAILURE - The following job has failed:

inductor-periodic / cuda12.1-py3.10-gcc9-sm80 / test (inductor_torchbench_smoketest_perf, 1, 1, linux.gcp.a100) (gh)
##[error]Process completed with exit code 1.

FLAKY - The following job failed but was likely due to flakiness present on trunk:

pull / linux-focal-py3.11-clang10 / test (dynamo, 3, 3, lf.linux.2xlarge) (gh) (disabled by #134602)
test_transformers.py::TestSDPAPrivateUse1Only::test_scaled_dot_product_fused_attention_overrideable_backward

This comment was automatically generated by Dr. CI and updates every 15 minutes.

facebook-github-bot · 2024-09-25T17:34:17Z

This pull request was exported from Phabricator. Differential Revision: D63399064

SherlockNoMad · 2024-09-26T09:46:57Z

torch/_export/converter.py


        # Mapping from torchscript node output name to attribute fully qualified name
-        self.name_to_attribute_fqn: Dict[str, str] = {}
+        self.name_to_attribute_fqn: Dict[str, str] = name_to_attribute_fqn


I am starting to loose track of the meaning of

name_to_param, name_to_buffer, name_to_node, name_to_constant, name_to_attribute_fqn.

Do you mind adding some comments on so that we don't confuse our future selves?

facebook-github-bot · 2024-09-26T17:59:34Z

This pull request was exported from Phabricator. Differential Revision: D63399064

Summary: Pull Request resolved: pytorch#136648 We previously incorrectly handled the following graph, specifically for the node `w.3` in `block0`: ``` graph(%x.1 : Float(3, strides=[1], requires_grad=0, device=cpu), %y.1 : int): %2 : __torch__.___torch_mangle_1.M = prim::CreateObject() %3 : int = prim::Constant[value=20](), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:747:34 %4 : int = prim::Constant[value=10](), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:746:34 %5 : int = prim::Constant[value=1](), scope: M:: %w.1 : int = prim::GetAttr[name="w"](%2), scope: M:: %7 : int = aten::mul(%w.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:746:25 = prim::SetAttr[name="w"](%2, %7), scope: M:: %h.1 : int = prim::GetAttr[name="h"](%2), scope: M:: %9 : int = aten::mul(%h.1, %3), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:747:25 = prim::SetAttr[name="h"](%2, %9), scope: M:: %10 : bool = aten::gt(%y.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:749:19 %res.37 : Tensor = prim::If(%10), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:749:16 block0(): %w.3 : int = prim::GetAttr[name="w"](%2), scope: M:: %res.1 : Tensor = aten::add(%x.1, %w.3, %5), scope: M:: # <string>:5:9 -> (%res.1) block1(): %h.3 : int = prim::GetAttr[name="h"](%2), scope: M:: %res.3 : Tensor = aten::add(%x.1, %h.3, %5), scope: M:: # <string>:5:9 -> (%res.3) %16 : bool = aten::lt(%y.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:754:19 %res : Tensor = prim::If(%16), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:754:16 block0(): %w : int = prim::GetAttr[name="w"](%2), scope: M:: %res.15 : Tensor = aten::add(%res.37, %w, %5), scope: M:: # <string>:5:9 -> (%res.15) block1(): %h : int = prim::GetAttr[name="h"](%2), scope: M:: %res.21 : Tensor = aten::add(%res.37, %h, %5), scope: M:: # <string>:5:9 -> (%res.21) return (%res) ``` Test Plan: CI Reviewed By: SherlockNoMad Differential Revision: D63399064

facebook-github-bot · 2024-10-01T21:04:26Z

This pull request was exported from Phabricator. Differential Revision: D63399064

Summary: Pull Request resolved: pytorch#136648 We previously incorrectly handled the following graph, specifically for the node `w.3` in `block0`: ``` graph(%x.1 : Float(3, strides=[1], requires_grad=0, device=cpu), %y.1 : int): %2 : __torch__.___torch_mangle_1.M = prim::CreateObject() %3 : int = prim::Constant[value=20](), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:747:34 %4 : int = prim::Constant[value=10](), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:746:34 %5 : int = prim::Constant[value=1](), scope: M:: %w.1 : int = prim::GetAttr[name="w"](%2), scope: M:: %7 : int = aten::mul(%w.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:746:25 = prim::SetAttr[name="w"](%2, %7), scope: M:: %h.1 : int = prim::GetAttr[name="h"](%2), scope: M:: %9 : int = aten::mul(%h.1, %3), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:747:25 = prim::SetAttr[name="h"](%2, %9), scope: M:: %10 : bool = aten::gt(%y.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:749:19 %res.37 : Tensor = prim::If(%10), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:749:16 block0(): %w.3 : int = prim::GetAttr[name="w"](%2), scope: M:: %res.1 : Tensor = aten::add(%x.1, %w.3, %5), scope: M:: # <string>:5:9 -> (%res.1) block1(): %h.3 : int = prim::GetAttr[name="h"](%2), scope: M:: %res.3 : Tensor = aten::add(%x.1, %h.3, %5), scope: M:: # <string>:5:9 -> (%res.3) %16 : bool = aten::lt(%y.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:754:19 %res : Tensor = prim::If(%16), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:754:16 block0(): %w : int = prim::GetAttr[name="w"](%2), scope: M:: %res.15 : Tensor = aten::add(%res.37, %w, %5), scope: M:: # <string>:5:9 -> (%res.15) block1(): %h : int = prim::GetAttr[name="h"](%2), scope: M:: %res.21 : Tensor = aten::add(%res.37, %h, %5), scope: M:: # <string>:5:9 -> (%res.21) return (%res) ``` Test Plan: CI Reviewed By: SherlockNoMad Differential Revision: D63399064

facebook-github-bot · 2024-10-01T21:14:02Z

This pull request was exported from Phabricator. Differential Revision: D63399064

angelayi · 2024-10-02T00:05:10Z

@pytorchbot merge -f "errors seem unrelated?"

pytorchmergebot · 2024-10-02T00:07:41Z

Merge started

Your change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use -f as last resort and instead consider -i/--ignore-current to continue the merge ignoring current failures. This will allow currently pending tests to finish and report signal before the merge.

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging

Check the merge workflow status
here

Summary: We previously incorrectly handled the following graph, specifically for the node `w.3` in `block0`: ``` graph(%x.1 : Float(3, strides=[1], requires_grad=0, device=cpu), %y.1 : int): %2 : __torch__.___torch_mangle_1.M = prim::CreateObject() %3 : int = prim::Constant[value=20](), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:747:34 %4 : int = prim::Constant[value=10](), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:746:34 %5 : int = prim::Constant[value=1](), scope: M:: %w.1 : int = prim::GetAttr[name="w"](%2), scope: M:: %7 : int = aten::mul(%w.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:746:25 = prim::SetAttr[name="w"](%2, %7), scope: M:: %h.1 : int = prim::GetAttr[name="h"](%2), scope: M:: %9 : int = aten::mul(%h.1, %3), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:747:25 = prim::SetAttr[name="h"](%2, %9), scope: M:: %10 : bool = aten::gt(%y.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:749:19 %res.37 : Tensor = prim::If(%10), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:749:16 block0(): %w.3 : int = prim::GetAttr[name="w"](%2), scope: M:: %res.1 : Tensor = aten::add(%x.1, %w.3, %5), scope: M:: # <string>:5:9 -> (%res.1) block1(): %h.3 : int = prim::GetAttr[name="h"](%2), scope: M:: %res.3 : Tensor = aten::add(%x.1, %h.3, %5), scope: M:: # <string>:5:9 -> (%res.3) %16 : bool = aten::lt(%y.1, %4), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:754:19 %res : Tensor = prim::If(%16), scope: M:: # /data/users/angelayi/pytorch/test/export/test_converter.py:754:16 block0(): %w : int = prim::GetAttr[name="w"](%2), scope: M:: %res.15 : Tensor = aten::add(%res.37, %w, %5), scope: M:: # <string>:5:9 -> (%res.15) block1(): %h : int = prim::GetAttr[name="h"](%2), scope: M:: %res.21 : Tensor = aten::add(%res.37, %h, %5), scope: M:: # <string>:5:9 -> (%res.21) return (%res) ``` Test Plan: CI Differential Revision: D63399064 Pull Request resolved: pytorch#136648 Approved by: https://github.com/SherlockNoMad

angelayi requested review from avikchaudhuri, tugsbayasgalan, ydwu4 and zhxchen17 as code owners September 25, 2024 17:33

pytorch-bot bot added the ciflow/inductor label Sep 25, 2024

facebook-github-bot added the fb-exported label Sep 25, 2024

angelayi added topic: not user facing topic category release notes: export labels Sep 25, 2024

SherlockNoMad reviewed Sep 26, 2024

View reviewed changes

SherlockNoMad approved these changes Sep 26, 2024

View reviewed changes

pytorch-bot bot added the ciflow/trunk Trigger trunk jobs on your pull request label Sep 26, 2024

angelayi force-pushed the export-D63399064 branch from efc9cdf to 6f1d128 Compare September 26, 2024 17:59

angelayi force-pushed the export-D63399064 branch from 6f1d128 to 6dbb2fc Compare October 1, 2024 21:04

angelayi force-pushed the export-D63399064 branch from 6dbb2fc to 96c2827 Compare October 1, 2024 21:14

pytorchmergebot added the merging label Oct 2, 2024

pytorchmergebot added the Merged label Oct 2, 2024

pytorchmergebot closed this in d725758 Oct 2, 2024

pytorchmergebot removed the merging label Oct 2, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[ts_converter] Fix prim::If buffer names #136648

[ts_converter] Fix prim::If buffer names #136648

Uh oh!

angelayi commented Sep 25, 2024

Uh oh!

pytorch-bot bot commented Sep 25, 2024 •

edited

Loading

Uh oh!

facebook-github-bot commented Sep 25, 2024

Uh oh!

SherlockNoMad Sep 26, 2024

Uh oh!

facebook-github-bot commented Sep 26, 2024

Uh oh!

facebook-github-bot commented Oct 1, 2024

Uh oh!

facebook-github-bot commented Oct 1, 2024

Uh oh!

angelayi commented Oct 2, 2024

Uh oh!

pytorchmergebot commented Oct 2, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

[ts_converter] Fix prim::If buffer names #136648

[ts_converter] Fix prim::If buffer names #136648

Uh oh!

Conversation

angelayi commented Sep 25, 2024

Uh oh!

pytorch-bot bot commented Sep 25, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/136648

❌ 1 New Failure, 1 Unrelated Failure

Uh oh!

facebook-github-bot commented Sep 25, 2024

Uh oh!

SherlockNoMad Sep 26, 2024

Choose a reason for hiding this comment

Uh oh!

facebook-github-bot commented Sep 26, 2024

Uh oh!

facebook-github-bot commented Oct 1, 2024

Uh oh!

facebook-github-bot commented Oct 1, 2024

Uh oh!

angelayi commented Oct 2, 2024

Uh oh!

pytorchmergebot commented Oct 2, 2024

Merge started

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

pytorch-bot bot commented Sep 25, 2024 •

edited

Loading