[RFC] Export Operator List from Mobile CompilationUnit instead of from TorchScript Model #49385

dhruvbird · 2020-12-15T03:16:47Z

Stack from ghstack:

[RFC] Export Operator List from Mobile CompilationUnit instead of from TorchScript Model #49385 [RFC] Export Operator List from Mobile CompilationUnit instead of from TorchScript Model

Currently, the API to export operator lists accepts a torch::jit::Module object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile.

What we need to to instead is glean the list of operators from the mobile model itself (bytecode.pkl specifically), and expose that instead.

Also updated the logic in converter.

Before this change:

Get operator List from Torch Script Model
Convert to bytecode mobile model

After this change:

Convert to bytecode mobile model
Use this converted mobile model to get the list of operators for each method on the model

Differential Revision: D24690094

NOTE FOR REVIEWERS: This PR has internal Facebook specific changes or comments, please review them on Phabricator!

…m TorchScript Model Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)! [ghstack-poisoned]

…tead of from TorchScript Model" Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)! [ghstack-poisoned]

…tead of from TorchScript Model Pull Request resolved: #49385 Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model ghstack-source-id: 118589313 Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)!

…tead of from TorchScript Model" Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)! [ghstack-poisoned]

facebook-github-bot · 2020-12-15T05:12:22Z

💊 CI failures summary and remediations

As of commit 90f7c08 (more details on the Dr. CI page):

2/2 failures possibly* introduced in this PR
- 1/2 non-CircleCI failure(s)

🕵️ 1 new failure recognized by patterns

The following CI failures do not appear to be due to upstream breakages:

pytorch_windows_vs2019_py36_cuda11.1_build (1/1)

Step: "Build" (full log | diagnosis details | 🔁 rerun)

CMake Error at third_party/gloo/cmake/Dependencies.cmake:46 (find_library):

This warning is for project developers.  Use -Wno-dev to suppress it.

-- MSVC detected
-- Set USE_REDIS OFF
-- Set USE_IBVERBS OFF
-- Set USE_NCCL OFF
-- Set USE_RCCL OFF
-- Set USE_LIBUV ON
-- Only USE_LIBUV is supported on Windows
-- Gloo build as SHARED library
CMake Error at third_party/gloo/cmake/Dependencies.cmake:46 (find_library):
  Could not find libuv_LIBRARY using the following names: uv, libuv
Call Stack (most recent call first):
  third_party/gloo/CMakeLists.txt:102 (include)


-- Configuring incomplete, errors occurred!
See also "C:/Users/circleci/project/build/CMakeFiles/CMakeOutput.log".
See also "C:/Users/circleci/project/build/CMakeFiles/CMakeError.log".
-- Building version 1.8.0a0
cmake -GNinja -DBUILD_ENVIRONMENT=pytorch-win-vs2019-cuda11-cudnn8-py3 -DBUILD_PYTHON=True -DBUILD_TEST=True -DBUILD_TYPE=release -DCMAKE_BUILD_TYPE=Release -DCMAKE_GENERATOR=Ninja -DCMAKE_INCLUDE_PATH=C:\Users\circleci\project\build\win_tmp\mkl\include -DCMAKE_INSTALL_PREFIX=C:\Users\circleci\project\torch -DCMAKE_PREFIX_PATH=C:\Jenkins\Miniconda3\Lib\site-packages -DCMAKE_VERBOSE_MAKEFILE=1 -DCUDA_NVCC_EXECUTABLE=C:\Users\circleci\project\build\win_tmp\bin\randomtemp.exe -DCUDNN_LIBRARY=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.1\lib\x64 -DJAVA_HOME=C:\Program Files\OpenJDK\jdk-12.0.2 -DNUMPY_INCLUDE_DIR=C:\Jenkins\Miniconda3\lib\site-packages\numpy\core\include -DPYTHON_EXECUTABLE=C:\Jenkins\Miniconda3\python.exe -DPYTHON_INCLUDE_DIR=C:\Jenkins\Miniconda3\include -DPYTHON_LIBRARY=C:\Jenkins\Miniconda3/libs/python36.lib -DTORCH_BUILD_VERSION=1.8.0a0 -DUSE_CUDA=1 -DUSE_NUMPY=True C:\Users\circleci\project

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 to the (internal) Dr. CI Users group.

This comment has been revised 56 times.

…tead of from TorchScript Model" Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)! [ghstack-poisoned]

…tead of from TorchScript Model Pull Request resolved: #49385 Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model ghstack-source-id: 118596248 Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)!

…tead of from TorchScript Model" Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)! [ghstack-poisoned]

…tead of from TorchScript Model Pull Request resolved: #49385 Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model ghstack-source-id: 118634218 Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)!

…tead of from TorchScript Model" Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)! [ghstack-poisoned]

…tead of from TorchScript Model Pull Request resolved: #49385 Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model ghstack-source-id: 118689226 Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)!

…tead of from TorchScript Model" Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)! [ghstack-poisoned]

…tead of from TorchScript Model Pull Request resolved: #49385 Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model ghstack-source-id: 118790588 Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)!

…tead of from TorchScript Model" Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)! [ghstack-poisoned]

…tead of from TorchScript Model Pull Request resolved: #49385 Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model ghstack-source-id: 118796752 Differential Revision: [D24690094](https://our.internmc.facebook.com/intern/diff/D24690094/) **NOTE FOR REVIEWERS**: This PR has internal Facebook specific changes or comments, please review them on [Phabricator](https://our.internmc.facebook.com/intern/diff/D24690094/)!

facebook-github-bot · 2020-12-18T21:21:13Z

This pull request has been merged in 4a870f6.

facebook-github-bot · 2020-12-18T21:21:28Z

This pull request has been merged in 4a870f6.

…tead of from TorchScript Model (pytorch#49385) Summary: Pull Request resolved: pytorch#49385 Currently, the API to export operator lists accepts a `torch::jit::Module` object, and spits out an operator list. The operator list is practically used only for mobile. This is not ideal because the set of root operators may change by the time the model is subsequently optmized and exported for mobile. What we need to to instead is glean the list of operators from the mobile model itself (`bytecode.pkl` specifically), and expose that instead. Also updated the logic in `converter`. ### Before this change: 1. Get operator List from Torch Script Model 2. Convert to bytecode mobile model ### After this change: 1. Convert to bytecode mobile model 2. Use this converted mobile model to get the list of operators for each method on the model ghstack-source-id: 118796752 Test Plan: Added a unit test in `test_lite_interpreter.cpp` to ensure that all model referenced operators show up in the exported operator list. Also make `test_lite_interpreter.cpp` runnable from `xplat/caffe2/BUCK` since this is where the production code will be built from. Verified that the list of operators produced before and after this change for an example model (segmentation) are the same. {P147863234} Also verified that the operator lists for BI-Xray model is different (we have been having problems with missing operators for this one): {P154903132} Reviewed By: iseeyuan Differential Revision: D24690094 fbshipit-source-id: 0426a6ef90456a811010cfe337c415882ae2deff

facebook-github-bot added cla signed oncall: jit Add this issue/PR to JIT oncall triage queue labels Dec 15, 2020

facebook-github-bot closed this in 4a870f6 Dec 18, 2020

facebook-github-bot added the Merged label Dec 18, 2020

facebook-github-bot deleted the gh/dhruvbird/21/head branch December 22, 2020 15:17

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[RFC] Export Operator List from Mobile CompilationUnit instead of from TorchScript Model #49385

[RFC] Export Operator List from Mobile CompilationUnit instead of from TorchScript Model #49385

Uh oh!

dhruvbird commented Dec 15, 2020 •

edited

Loading

Uh oh!

facebook-github-bot commented Dec 15, 2020 •

edited

Loading

Uh oh!

facebook-github-bot commented Dec 18, 2020

Uh oh!

facebook-github-bot commented Dec 18, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

[RFC] Export Operator List from Mobile CompilationUnit instead of from TorchScript Model #49385

[RFC] Export Operator List from Mobile CompilationUnit instead of from TorchScript Model #49385

Uh oh!

Conversation

dhruvbird commented Dec 15, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Before this change:

After this change:

Uh oh!

facebook-github-bot commented Dec 15, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

💊 CI failures summary and remediations

🕵️ 1 new failure recognized by patterns

pytorch_windows_vs2019_py36_cuda11.1_build (1/1)

Uh oh!

facebook-github-bot commented Dec 18, 2020

Uh oh!

facebook-github-bot commented Dec 18, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

dhruvbird commented Dec 15, 2020 •

edited

Loading

facebook-github-bot commented Dec 15, 2020 •

edited

Loading