[ONNX] Utilize ONNX shape inference for ONNX exporter#40628
Closed
BowenBao wants to merge 5 commits intopytorch:masterfrom
Closed
[ONNX] Utilize ONNX shape inference for ONNX exporter#40628BowenBao wants to merge 5 commits intopytorch:masterfrom
BowenBao wants to merge 5 commits intopytorch:masterfrom
Conversation
💊 CI failures summary and remediationsAs of commit 58158a0 (more details on the Dr. CI page): 💚 💚 Looks good so far! There are no failures yet. 💚 💚 This comment was automatically generated by Dr. CI (expand for details).Follow this link to opt-out of these comments for your Pull Requests.Please report bugs/suggestions on the GitHub issue tracker or post in the (internal) Dr. CI Users group. This comment has been revised 120 times. |
|
@BowenBao - Does this also include type propagation, or is that a separate PR? |
Collaborator
Author
Yes this includes type as well. |
houseroad
reviewed
Jul 2, 2020
neginraoof
reviewed
Jul 7, 2020
spandantiwari
approved these changes
Jul 20, 2020
spandantiwari
left a comment
There was a problem hiding this comment.
LGTM. If you could please capture any of the upcoming work items related to this in a comment or the opening description that will be great for context.
neginraoof
approved these changes
Jul 21, 2020
bzinodev
approved these changes
Jul 22, 2020
Contributor
facebook-github-bot
left a comment
There was a problem hiding this comment.
@bzinodev has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
Contributor
facebook-github-bot
left a comment
There was a problem hiding this comment.
@bzinodev has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
03c59c8 to
10155aa
Compare
26dd024 to
ea7a49f
Compare
Contributor
facebook-github-bot
left a comment
There was a problem hiding this comment.
@bzinodev has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
ea7a49f to
ae98158
Compare
Contributor
facebook-github-bot
left a comment
There was a problem hiding this comment.
@bzinodev has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
0b56320 to
5d6f8cf
Compare
clang tidy add helper func remove useless comment add missing type fix flake and clang-tidy update clang-tidy skip files address comments clang-tidy add more comments rebase with master clang format clang-tidy on unrelated file clang-format newline
5d6f8cf to
58158a0
Compare
Collaborator
Author
|
@bzinodev CI is green, please help import this one. |
houseroad
reviewed
Aug 26, 2020
Member
houseroad
left a comment
There was a problem hiding this comment.
Awesome, this is happening!
@houseroad - Yes, it is a first PR :-). We still have some ground to cover, but a good start. Thanks for all the brainstorming and discussions on this. |
Contributor
facebook-github-bot
left a comment
There was a problem hiding this comment.
@bzinodev has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
Contributor
facebook-github-bot
pushed a commit
that referenced
this pull request
Sep 17, 2020
…43991) Summary: Fixes the `true_divide` symbolic to cast tensors correctly. The logic depends on knowing input types at export time, which is a known gap for exporting scripted modules. On that end we are improving exporter by enabling ONNX shape inference #40628, and starting to increase coverage for scripting support. Pull Request resolved: #43991 Reviewed By: mruberry Differential Revision: D23674614 Pulled By: bzinodev fbshipit-source-id: 1b1b85340eef641f664a14c4888781389c886a8b
xuzhao9
pushed a commit
that referenced
this pull request
Sep 18, 2020
…43991) Summary: Fixes the `true_divide` symbolic to cast tensors correctly. The logic depends on knowing input types at export time, which is a known gap for exporting scripted modules. On that end we are improving exporter by enabling ONNX shape inference #40628, and starting to increase coverage for scripting support. Pull Request resolved: #43991 Reviewed By: mruberry Differential Revision: D23674614 Pulled By: bzinodev fbshipit-source-id: 1b1b85340eef641f664a14c4888781389c886a8b
laurentdupin
pushed a commit
to laurentdupin/pytorch
that referenced
this pull request
Apr 24, 2026
Summary: in `_jit_pass_onnx`, symbolic functions are called for each node for conversion. However, there are nodes that cannot be converted without additional context. For example, the number of outputs from split (and whether it is static or dynamic) is unknown until the point where it is unpacked by listUnpack node. This pass does a preprocess, and prepares the nodes such that enough context can be received by the symbolic function. * After preprocessing, `_jit_pass_onnx` should have enough context to produce valid ONNX nodes, instead of half baked nodes that replies on fixes from later postpasses. * `_jit_pass_onnx_peephole` should be a pass that does ONNX specific optimizations instead of ONNX specific fixes. * Producing more valid ONNX nodes in `_jit_pass_onnx` enables better utilization of the ONNX shape inference pytorch#40628. Pull Request resolved: pytorch#41832 Reviewed By: ZolotukhinM Differential Revision: D22968334 Pulled By: bzinodev fbshipit-source-id: 8226f03c5b29968e8197d242ca8e620c6e1d42a5
laurentdupin
pushed a commit
to laurentdupin/pytorch
that referenced
this pull request
Apr 24, 2026
Summary: It is often that the conversion from torch operator to onnx operator requires input rank/dtype/shape to be known. Previously, the conversion depends on tracer to provide these info, leaving a gap in conversion of scripted modules. We are extending the export with support from onnx shape inference. If enabled, onnx shape inference will be called whenever an onnx node is created. This is the first PR introducing the initial look of the feature. More and more cases will be supported following this PR. * Added pass to run onnx shape inference on a given node. The node has to have namespace `onnx`. * Moved helper functions from `export.cpp` to a common place for re-use. * This feature is currently experimental, and can be turned on through flag `onnx_shape_inference` in internal api `torch.onnx._export`. * Currently skipping ONNX Sequence ops, If/Loop and ConstantOfShape due to limitations. Support will be added in the future. Pull Request resolved: pytorch#40628 Reviewed By: mrshenli Differential Revision: D22709746 Pulled By: bzinodev fbshipit-source-id: b52aeeae00667e66e0b0c1144022f7af9a8b2948
laurentdupin
pushed a commit
to laurentdupin/pytorch
that referenced
this pull request
Apr 24, 2026
…ytorch#43991) Summary: Fixes the `true_divide` symbolic to cast tensors correctly. The logic depends on knowing input types at export time, which is a known gap for exporting scripted modules. On that end we are improving exporter by enabling ONNX shape inference pytorch#40628, and starting to increase coverage for scripting support. Pull Request resolved: pytorch#43991 Reviewed By: mruberry Differential Revision: D23674614 Pulled By: bzinodev fbshipit-source-id: 1b1b85340eef641f664a14c4888781389c886a8b
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
9 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.
It is often that the conversion from torch operator to onnx operator requires input rank/dtype/shape to be known. Previously, the conversion depends on tracer to provide these info, leaving a gap in conversion of scripted modules.
We are extending the export with support from onnx shape inference. If enabled, onnx shape inference will be called whenever an onnx node is created. This is the first PR introducing the initial look of the feature. More and more cases will be supported following this PR.
onnx.export.cppto a common place for re-use.onnx_shape_inferencein internal apitorch.onnx._export.