[quant][fx] Support some default ops in the native backend config #74600

jerryzh168 · 2022-03-23T06:10:06Z

Stack from ghstack (oldest at bottom):

Summary:
Following #74210, this PR adds the support for some ops
using the DefaultNodeQuantizeHandler in the backend_config_dict defintion for pytorch native backend

TODO: There is still a few ops we didn't handle with backend_config_dict path: gelu and softmax, need to discuss if we still need them, if so we can change the test
to use backend_config_dict and remove the DefaultNodeQuantizeHandler after that

Test Plan:
python test/test_quantization.py TestQuantizeFxOps
python test/test_quantization.py TestFXNumericSuiteCoreAPIs
python test/test_quantization.py TestFXGraphMatcher

Reviewers:

Subscribers:

Tasks:

Tags:

Differential Revision: D35071437

Summary: Following #74210, this PR adds the support for some ops using the DefaultNodeQuantizeHandler in the backend_config_dict defintion for pytorch native backend TODO: There is still a few ops we didn't handle with backend_config_dict path: gelu and softmax, need to discuss if we still need them, if so we can change the test to use backend_config_dict and remove the DefaultNodeQuantizeHandler after that Test Plan: python test/test_quantization.py TestQuantizeFxOps Reviewers: Subscribers: Tasks: Tags: [ghstack-poisoned]

facebook-github-bot · 2022-03-23T06:10:12Z

🔗 Helpful links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/74600
↩️ [fb-only] Re-run with SSH instructions
Need help or want to give feedback on the CI? Visit our office hours

💊 CI failures summary and remediations

As of commit 007eb11 (more details on the Dr. CI page):

4/4 failures introduced in this PR

🕵️ 3 new failures recognized by patterns

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

pull / linux-xenial-py3.7-gcc5.4 / test (docs_test, 1, 1, linux.2xlarge) (1/3)

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

2022-03-24T19:31:51.3295900Z ImportError: canno...da/lib/python3.7/site-packages/jinja2/__init__.py)

2022-03-24T19:31:51.3292445Z   File "/opt/conda/lib/python3.7/site-packages/sphinx/cmd/build.py", line 25, in <module>
2022-03-24T19:31:51.3292742Z     from sphinx.application import Sphinx
2022-03-24T19:31:51.3293114Z   File "/opt/conda/lib/python3.7/site-packages/sphinx/application.py", line 43, in <module>
2022-03-24T19:31:51.3293419Z     from sphinx.registry import SphinxComponentRegistry
2022-03-24T19:31:51.3293802Z   File "/opt/conda/lib/python3.7/site-packages/sphinx/registry.py", line 24, in <module>
2022-03-24T19:31:51.3294074Z     from sphinx.builders import Builder
2022-03-24T19:31:51.3294444Z   File "/opt/conda/lib/python3.7/site-packages/sphinx/builders/__init__.py", line 26, in <module>
2022-03-24T19:31:51.3294778Z     from sphinx.util import import_object, logging, progress_message, rst, status_iterator
2022-03-24T19:31:51.3295180Z   File "/opt/conda/lib/python3.7/site-packages/sphinx/util/rst.py", line 21, in <module>
2022-03-24T19:31:51.3295474Z     from jinja2 import Environment, environmentfilter
2022-03-24T19:31:51.3295900Z ImportError: cannot import name 'environmentfilter' from 'jinja2' (/opt/conda/lib/python3.7/site-packages/jinja2/__init__.py)
2022-03-24T19:31:51.3580869Z Makefile:38: recipe for target 'doctest' failed
2022-03-24T19:31:51.3581188Z make: *** [doctest] Error 1
2022-03-24T19:31:51.3581843Z + cleanup
2022-03-24T19:31:51.3582064Z + retcode=2
2022-03-24T19:31:51.3582361Z + set +x
2022-03-24T19:31:51.3585284Z + cleanup
2022-03-24T19:31:51.3585499Z + retcode=2
2022-03-24T19:31:51.3585755Z + set +x
2022-03-24T19:31:51.3627727Z ##[error]Process completed with exit code 2.
2022-03-24T19:31:51.3662511Z ##[group]Run pytorch/pytorch/.github/actions/chown-workspace@master

pull / win-vs2019-cuda11.3-py3 / test (default, 1, 2, windows.8xlarge.nvidia.gpu) (2/3)

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

2022-03-24T22:24:00.9771434Z RuntimeError: test_torch failed!

2022-03-24T22:23:46.6762666Z Generated XML report: test-reports\python-unittest\test_torch\TEST-TestVitalSignsCudaCUDA-20220324221417.xml
2022-03-24T22:23:46.6764557Z Generated XML report: test-reports\python-unittest\test_torch\TEST-TestVitalSignsCudaCPU-20220324221417.xml
2022-03-24T22:24:00.4197010Z [TORCH_VITAL] CUDA.used		 true
2022-03-24T22:24:00.4197808Z [TORCH_VITAL] Dataloader.basic_unit_test		 TEST_VALUE_STRING
2022-03-24T22:24:00.4198635Z [TORCH_VITAL] Dataloader.enabled		 True
2022-03-24T22:24:00.9750760Z Traceback (most recent call last):
2022-03-24T22:24:00.9752044Z   File "run_test.py", line 1054, in <module>
2022-03-24T22:24:00.9752593Z     main()
2022-03-24T22:24:00.9753272Z   File "run_test.py", line 1032, in main
2022-03-24T22:24:00.9770768Z     raise RuntimeError(err_message)
2022-03-24T22:24:00.9771434Z RuntimeError: test_torch failed!
2022-03-24T22:24:03.7102856Z 
2022-03-24T22:24:03.7120974Z (base) C:\actions-runner\_work\pytorch\pytorch\test>if ERRORLEVEL 1 goto fail 
2022-03-24T22:24:03.7138570Z 
2022-03-24T22:24:03.7139366Z (base) C:\actions-runner\_work\pytorch\pytorch\test>exit /b 1 
2022-03-24T22:24:03.8911507Z + cleanup
2022-03-24T22:24:03.8922094Z + retcode=1
2022-03-24T22:24:03.8922560Z + set +x
2022-03-24T22:24:03.9875178Z ##[error]Process completed with exit code 1.
2022-03-24T22:24:04.7882898Z ##[group]Run pytorch/pytorch/.github/actions/upload-test-artifacts@master
2022-03-24T22:24:04.7883745Z with:

pull / linux-docs / build-docs (python) (3/3)

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

2022-03-24T19:31:16.6945825Z ImportError: canno...da/lib/python3.7/site-packages/jinja2/__init__.py)

2022-03-24T19:31:16.6942281Z   File "/opt/conda/lib/python3.7/site-packages/sphinx/cmd/build.py", line 25, in <module>
2022-03-24T19:31:16.6942569Z     from sphinx.application import Sphinx
2022-03-24T19:31:16.6942951Z   File "/opt/conda/lib/python3.7/site-packages/sphinx/application.py", line 43, in <module>
2022-03-24T19:31:16.6943275Z     from sphinx.registry import SphinxComponentRegistry
2022-03-24T19:31:16.6943653Z   File "/opt/conda/lib/python3.7/site-packages/sphinx/registry.py", line 24, in <module>
2022-03-24T19:31:16.6943957Z     from sphinx.builders import Builder
2022-03-24T19:31:16.6944331Z   File "/opt/conda/lib/python3.7/site-packages/sphinx/builders/__init__.py", line 26, in <module>
2022-03-24T19:31:16.6944689Z     from sphinx.util import import_object, logging, progress_message, rst, status_iterator
2022-03-24T19:31:16.6945082Z   File "/opt/conda/lib/python3.7/site-packages/sphinx/util/rst.py", line 21, in <module>
2022-03-24T19:31:16.6945380Z     from jinja2 import Environment, environmentfilter
2022-03-24T19:31:16.6945825Z ImportError: cannot import name 'environmentfilter' from 'jinja2' (/opt/conda/lib/python3.7/site-packages/jinja2/__init__.py)
2022-03-24T19:31:16.7222502Z Makefile:38: recipe for target 'html' failed
2022-03-24T19:31:16.7222899Z make[1]: *** [html] Error 1
2022-03-24T19:31:16.7223238Z make[1]: Leaving directory '/var/lib/jenkins/workspace/docs'
2022-03-24T19:31:16.7223572Z Makefile:27: recipe for target 'html-stable' failed
2022-03-24T19:31:16.7223966Z make: *** [html-stable] Error 2
2022-03-24T19:31:16.7225914Z ++ code=2
2022-03-24T19:31:16.7226415Z ++ '[' 2 -ne 0 ']'
2022-03-24T19:31:16.7226615Z ++ set +x
2022-03-24T19:31:16.7226781Z =========================
2022-03-24T19:31:16.7235806Z =========================

1 failure not recognized by patterns:

Job	Step	Action
^{pull / linux-docs / build-docs (cpp)}	^Unknown	🔁 rerun

This comment was automatically generated by Dr. CI (expand for details).

Please report bugs/suggestions to the (internal) Dr. CI Users group.

Click here to manually regenerate this comment.

jerryzh168 · 2022-03-23T06:11:45Z

@jerryzh168 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

… config" Summary: Following #74210, this PR adds the support for some ops using the DefaultNodeQuantizeHandler in the backend_config_dict defintion for pytorch native backend TODO: There is still a few ops we didn't handle with backend_config_dict path: gelu and softmax, need to discuss if we still need them, if so we can change the test to use backend_config_dict and remove the DefaultNodeQuantizeHandler after that Test Plan: python test/test_quantization.py TestQuantizeFxOps Reviewers: Subscribers: Tasks: Tags: Differential Revision: [D35071437](https://our.internmc.facebook.com/intern/diff/D35071437) [ghstack-poisoned]

jerryzh168 · 2022-03-23T06:20:39Z

@jerryzh168 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

torch/ao/quantization/fx/backend_config/native.py

… config" Summary: Following #74210, this PR adds the support for some ops using the DefaultNodeQuantizeHandler in the backend_config_dict defintion for pytorch native backend TODO: There is still a few ops we didn't handle with backend_config_dict path: gelu and softmax, need to discuss if we still need them, if so we can change the test to use backend_config_dict and remove the DefaultNodeQuantizeHandler after that Test Plan: python test/test_quantization.py TestQuantizeFxOps Reviewers: Subscribers: Tasks: Tags: Differential Revision: [D35071437](https://our.internmc.facebook.com/intern/diff/D35071437) [ghstack-poisoned]

jerryzh168 · 2022-03-23T23:05:22Z

test/quantization/fx/test_numeric_suite_fx.py

                    _op_in_base_sets_of_related_ops(base_op),
                    f"{base_op} not in sets of related ops")
-            else:
+            elif not _op_in_base_sets_of_related_ops(base_op):


@vkuzo this is the change I made to the test, please take a look

basically we don't explicitly check the quantize handler type in this branch, but just directly check if op is in the related ops set

… config" Summary: Following #74210, this PR adds the support for some ops using the DefaultNodeQuantizeHandler in the backend_config_dict defintion for pytorch native backend TODO: There is still a few ops we didn't handle with backend_config_dict path: gelu and softmax, need to discuss if we still need them, if so we can change the test to use backend_config_dict and remove the DefaultNodeQuantizeHandler after that Test Plan: python test/test_quantization.py TestQuantizeFxOps Reviewers: Subscribers: Tasks: Tags: Differential Revision: [D35071437](https://our.internmc.facebook.com/intern/diff/D35071437) [ghstack-poisoned]

jerryzh168 · 2022-03-23T23:12:42Z

@jerryzh168 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

test/quantization/fx/test_numeric_suite_fx.py

… config" Summary: Following #74210, this PR adds the support for some ops using the DefaultNodeQuantizeHandler in the backend_config_dict defintion for pytorch native backend TODO: There is still a few ops we didn't handle with backend_config_dict path: gelu and softmax, need to discuss if we still need them, if so we can change the test to use backend_config_dict and remove the DefaultNodeQuantizeHandler after that Test Plan: python test/test_quantization.py TestQuantizeFxOps python test/test_quantization.py TestFXNumericSuiteCoreAPIs python test/test_quantization.py TestFXGraphMatcher Reviewers: Subscribers: Tasks: Tags: Differential Revision: [D35071437](https://our.internmc.facebook.com/intern/diff/D35071437) [ghstack-poisoned]

jerryzh168 · 2022-03-24T04:44:42Z

@jerryzh168 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

… config" Summary: Following #74210, this PR adds the support for some ops using the DefaultNodeQuantizeHandler in the backend_config_dict defintion for pytorch native backend TODO: There is still a few ops we didn't handle with backend_config_dict path: gelu and softmax, need to discuss if we still need them, if so we can change the test to use backend_config_dict and remove the DefaultNodeQuantizeHandler after that Test Plan: python test/test_quantization.py TestQuantizeFxOps python test/test_quantization.py TestFXNumericSuiteCoreAPIs python test/test_quantization.py TestFXGraphMatcher Reviewers: Subscribers: Tasks: Tags: Differential Revision: [D35071437](https://our.internmc.facebook.com/intern/diff/D35071437) [ghstack-poisoned]

jerryzh168 · 2022-03-24T18:36:37Z

@jerryzh168 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

… config" Summary: Following #74210, this PR adds the support for some ops using the DefaultNodeQuantizeHandler in the backend_config_dict defintion for pytorch native backend TODO: There is still a few ops we didn't handle with backend_config_dict path: gelu and softmax, need to discuss if we still need them, if so we can change the test to use backend_config_dict and remove the DefaultNodeQuantizeHandler after that Test Plan: python test/test_quantization.py TestQuantizeFxOps python test/test_quantization.py TestFXNumericSuiteCoreAPIs python test/test_quantization.py TestFXGraphMatcher Reviewers: Subscribers: Tasks: Tags: Differential Revision: [D35071437](https://our.internmc.facebook.com/intern/diff/D35071437) [ghstack-poisoned]

jerryzh168 · 2022-03-24T20:14:09Z

@jerryzh168 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

…4600) Summary: Pull Request resolved: #74600 Following #74210, this PR adds the support for some ops using the DefaultNodeQuantizeHandler in the backend_config_dict defintion for pytorch native backend TODO: There is still a few ops we didn't handle with backend_config_dict path: gelu and softmax, need to discuss if we still need them, if so we can change the test to use backend_config_dict and remove the DefaultNodeQuantizeHandler after that Test Plan: python test/test_quantization.py TestQuantizeFxOps Imported from OSS Reviewed By: andrewor14 Differential Revision: D35071437 fbshipit-source-id: 70351d2810ca1ac7dc09d4a9c239f6757ccb51ca

github-actions · 2022-03-25T03:00:07Z

Hey @jerryzh168.
You've committed this PR, but it does not have both a 'release notes: ...' and 'topics: ...' label. Please add one of each to the PR. The 'release notes: ...' label should represent the part of PyTorch that this PR changes (fx, autograd, distributed, etc) and the 'topics: ...' label should represent the kind of PR it is (not user facing, new feature, bug fix, perf improvement, etc). The list of valid labels can be found here for the 'release notes: ...' and here for the 'topics: ...'.
For changes that are 'topic: not user facing' there is no need for a release notes label.

…4600) Summary: Pull Request resolved: #74600 Following #74210, this PR adds the support for some ops using the DefaultNodeQuantizeHandler in the backend_config_dict defintion for pytorch native backend TODO: There is still a few ops we didn't handle with backend_config_dict path: gelu and softmax, need to discuss if we still need them, if so we can change the test to use backend_config_dict and remove the DefaultNodeQuantizeHandler after that Test Plan: python test/test_quantization.py TestQuantizeFxOps Imported from OSS Reviewed By: andrewor14 Differential Revision: D35071437 fbshipit-source-id: 70351d2810ca1ac7dc09d4a9c239f6757ccb51ca (cherry picked from commit 5e68f75)

facebook-github-bot added the cla signed label Mar 23, 2022

jerryzh168 mentioned this pull request Mar 23, 2022

[quant][fx] Merge is_general_tensor_shape_op into is_general_tensor_value_op in QuantizeHandler #74601

Closed

facebook-github-bot added the module: fx label Mar 23, 2022

jerryzh168 mentioned this pull request Mar 23, 2022

[quant][fx] Removing more unused code #74603

Closed

jerryzh168 requested review from andrewor14 and vkuzo March 23, 2022 06:46

andrewor14 approved these changes Mar 23, 2022

View reviewed changes

torch/ao/quantization/fx/backend_config/native.py Outdated Show resolved Hide resolved

torch/ao/quantization/fx/backend_config/native.py Outdated Show resolved Hide resolved

jerryzh168 added 3 commits March 23, 2022 14:14

jerryzh168 commented Mar 23, 2022

View reviewed changes

jerryzh168 added 2 commits March 23, 2022 16:08

jerryzh168 commented Mar 23, 2022

View reviewed changes

test/quantization/fx/test_numeric_suite_fx.py Show resolved Hide resolved

jerryzh168 added 2 commits March 23, 2022 20:02

pytorchmergebot closed this in b347b8c Mar 25, 2022

jerryzh168 added the topic: not user facing topic category label Mar 25, 2022

facebook-github-bot deleted the gh/jerryzh168/754/head branch March 28, 2022 14:17

WBobby mentioned this pull request Aug 17, 2022

Add ROCm5.2.3/AMDGPU support for PyTorch WBobby/pytorch#2

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[quant][fx] Support some default ops in the native backend config #74600

[quant][fx] Support some default ops in the native backend config #74600

Uh oh!

jerryzh168 commented Mar 23, 2022 •

edited

Loading

Uh oh!

facebook-github-bot commented Mar 23, 2022 •

edited

Loading

Uh oh!

jerryzh168 commented Mar 23, 2022

Uh oh!

jerryzh168 commented Mar 23, 2022

Uh oh!

Uh oh!

Uh oh!

jerryzh168 Mar 23, 2022

Uh oh!

jerryzh168 Mar 23, 2022

Uh oh!

jerryzh168 commented Mar 23, 2022

Uh oh!

Uh oh!

jerryzh168 commented Mar 24, 2022

Uh oh!

jerryzh168 commented Mar 24, 2022

Uh oh!

jerryzh168 commented Mar 24, 2022

Uh oh!

github-actions bot commented Mar 25, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

[quant][fx] Support some default ops in the native backend config #74600

[quant][fx] Support some default ops in the native backend config #74600

Uh oh!

Conversation

jerryzh168 commented Mar 23, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

facebook-github-bot commented Mar 23, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful links

💊 CI failures summary and remediations

🕵️ 3 new failures recognized by patterns

pull / linux-xenial-py3.7-gcc5.4 / test (docs_test, 1, 1, linux.2xlarge) (1/3)

pull / win-vs2019-cuda11.3-py3 / test (default, 1, 2, windows.8xlarge.nvidia.gpu) (2/3)

pull / linux-docs / build-docs (python) (3/3)

1 failure not recognized by patterns:

Uh oh!

jerryzh168 commented Mar 23, 2022

Uh oh!

jerryzh168 commented Mar 23, 2022

Uh oh!

Uh oh!

Uh oh!

jerryzh168 Mar 23, 2022

Choose a reason for hiding this comment

Uh oh!

jerryzh168 Mar 23, 2022

Choose a reason for hiding this comment

Uh oh!

jerryzh168 commented Mar 23, 2022

Uh oh!

Uh oh!

jerryzh168 commented Mar 24, 2022

Uh oh!

jerryzh168 commented Mar 24, 2022

Uh oh!

jerryzh168 commented Mar 24, 2022

Uh oh!

github-actions bot commented Mar 25, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

jerryzh168 commented Mar 23, 2022 •

edited

Loading

facebook-github-bot commented Mar 23, 2022 •

edited

Loading