Fix dynamo handler str userdefinedobjectvariable #130320

datagero · 2024-07-09T09:03:24Z

Adjusted the call_str method to handle str conversion for UserDefinedObjectVariable.

cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @peterbell10 @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @ColinPeppler @amjames @desertfire @chauhang

…jectvariable

pytorch-bot · 2024-07-09T09:03:27Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/130320

📄 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 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

Too many GPU instances requested - any *.nvidia.gpu and ephemeral instances should experience significant queue times

✅ No Failures

As of commit 2cb35d4 with merge base ceedee2 ():
💚 Looks good so far! There are no failures yet. 💚

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

oulgen

Please add a test case showing that the behavior is same between eager and dynamo

Add one for when class defines __str__ and another when it does not

…thub.com:datagero/pytorch into fix-dynamo-handler-str-userdefinedobjectvariable

…jectvariable

datagero · 2024-07-09T20:26:00Z

@oulgen tests added in tests/dynamo/test_functions.py::DefaultTests

oulgen

Looks good!

datagero · 2024-07-09T22:20:35Z

@oulgen Thanks - Do I need to fix the failed checks? lint would be straigthforward but not so sure of the rest which appear unrelated

oulgen · 2024-07-09T23:12:00Z

You won't be able to land your changes until the CI is completely green. If you don't think that these breakages are due to your change, you can rebase on top of viable/strict so trunk is clean

…jectvariable

) Pull Request resolved: pytorch#129343 Approved by: https://github.com/lezcano

…ch#130306) Fixes pytorch#127649 Pull Request resolved: pytorch#130306 Approved by: https://github.com/janeyx99

Some functions used by Caffe2 were removed along with some outdated checks. Follows pytorch#130006. Pull Request resolved: pytorch#130322 Approved by: https://github.com/r-barnes

…30271) Afaict the previous check was too strict. Removing it passes all the mutation tests (mutation checks happen via the TensorVariable's mutable_local). Test Plan: - tests Pull Request resolved: pytorch#130271 Approved by: https://github.com/Chillee, https://github.com/ydwu4 ghstack dependencies: pytorch#130255, pytorch#130268

- Add AMD support for int4 kernel - Only supports CDNA2 and CDNA3 gpus for now - Uses `mfma_f32_16x16x16bf16` instruction for matrix multiply - Uses `v_and_or_b32` instruction and `__hfma2` instrinsic for unpacking bf16 values - Enable hipify for `__nv_bfloat16` and `__nv_bfloat162` data types - Enable int4 unit tests for CDNA2 and CDNA3 AMD gpus - Fix torchscript issues due to hipify for `__nv_bfloat16` type - TorchScript has its own implementation for bfloat16 type - Implemented in `__nv_bloat16` structure at [resource_strings.h](https://github.com/pytorch/pytorch/blob/main/torch/csrc/jit/codegen/fuser/cuda/resource_strings.h) - So, we shouldn't hipify any reference of `__nv_bfloat16` in the torchscript implementation - Hence moved the `__nv_bfloat16` direct references in `codegen.cpp` and `cuda_codegen.cpp` to `resource_strings.h` which is already exempted from hipify Fixes pytorch#124699 Fixes meta-pytorch/gpt-fast/issues/154 Co-authored-by: Nikita Shulga <2453524+malfet@users.noreply.github.com> Pull Request resolved: pytorch#129710 Approved by: https://github.com/malfet

As title, patch the dtype in torch.fx.graph Pull Request resolved: pytorch#130339 Approved by: https://github.com/angelayi

Fixes pytorch#130241 that is a reopen pr of pytorch#130244, for possibly fixing the failed job Pull Request resolved: pytorch#130305 Approved by: https://github.com/Skylion007

This adds a `_run_handler` method that will invoke a specific handler. Test plan: ``` python test/distributed/elastic/test_control_plane.py ``` Pull Request resolved: pytorch#130149 Approved by: https://github.com/kurman, https://github.com/c-p-i-o

…128599)" This reverts commit 940e447. Reverted pytorch#128599 on behalf of https://github.com/izaitsevfb due to breaking internal APS tests, see D59498864 ([comment](pytorch#128599 (comment)))

…odes) (pytorch#130064) We add torch.library.Library._register_torch_dispatch_rule. Here, a user can provide us a specific rule to run for a specific (torch_dispatch_class, operator) pair. The motivation is that a user might want to extend a subclass/mode but may not have access to the source code of the subclass/mode. I'll make this public in a follow-up PR if we think the approach and API is good. Keep in mind that many subclasses will likely deliver their own open registration solution (DTensor has register_sharding_prop_rule and NJT has register_jagged_op); _register_torch_dispatch_rule is meant as a catch-all open registration mechanism for when the subclass hasn't provided anything more specific. Test Plan: - new tests Pull Request resolved: pytorch#130064 Approved by: https://github.com/albanD

…0261) This is the API for defining the interaction between a torch_dispatch class and a custom op. Taking API bikeshedding. Test Plan: - new tests Pull Request resolved: pytorch#130261 Approved by: https://github.com/albanD ghstack dependencies: pytorch#130064

…ces (pytorch#130315) Summary: This diff fixes a bug, where all record_annotations will save a TraceEntry to each of the device_traces. Instead, we should only save annotations to the current device_trace that is being called by the thread calling the native allocator's recordAnnotation. Test Plan: CI and ran workloads on MVAI WPR FBR. Reviewed By: zdevito Differential Revision: D59477339 Pulled By: aaronenyeshi Pull Request resolved: pytorch#130315 Approved by: https://github.com/zdevito

…30271) Afaict the previous check was too strict. Removing it passes all the mutation tests (mutation checks happen via the TensorVariable's mutable_local). Test Plan: - tests Pull Request resolved: pytorch#130271 Approved by: https://github.com/Chillee, https://github.com/ydwu4 ghstack dependencies: pytorch#130255, pytorch#130268

…ytorch#130271)" This reverts commit adb6568. Reverted pytorch#130271 on behalf of https://github.com/clee2000 due to broke inductor/test_flex_attention https://github.com/pytorch/pytorch/actions/runs/9863205414/job/27236960046 https://hud.pytorch.org/pytorch/pytorch/commit/adb65682affdfc37f724c02ea8c8930d3925fc07 Test not run on PR due to bad TD ([comment](pytorch#130271 (comment)))

…orch#130227)" This reverts commit 6413998. Reverted pytorch#130227 on behalf of https://github.com/izaitsevfb due to breaks internal builds, please see D59498662 ([comment](pytorch#130227 (comment)))

…ytorch#130216) Follows pytorch#120353 Pull Request resolved: pytorch#130216 Approved by: https://github.com/albanD

Due to the removal of Caffe2 code. Pull Request resolved: pytorch#130089 Approved by: https://github.com/r-barnes, https://github.com/albanD

Fixes pytorch#129719 Pull Request resolved: pytorch#129824 Approved by: https://github.com/angelayi, https://github.com/yushangdi

…ytorch#130329) Follows pytorch#130300 Pull Request resolved: pytorch#130329 Approved by: https://github.com/ezyang

Pull Request resolved: pytorch#128416 Approved by: https://github.com/amjames, https://github.com/lezcano

…thub.com:datagero/pytorch into fix-dynamo-handler-str-userdefinedobjectvariable

anijain2305 · 2024-07-10T16:23:16Z

@datagero Can you take a look at the failures?

…jectvariable

…lexAttention numerics to be as accurate as FA2) (pytorch#130250)" This reverts commit 3e48d92. Reverted pytorch#130250 on behalf of https://github.com/izaitsevfb due to depends on pytorch#130227 which needs to be reverted ([comment](pytorch#130250 (comment)))

…orch#130227)" This reverts commit 6413998. Reverted pytorch#130227 on behalf of https://github.com/izaitsevfb due to breaks internal builds, please see D59498662 ([comment](pytorch#130227 (comment)))

datagero · 2024-07-10T23:42:09Z

Attempting to fix tests, I think #130506 will have a better chance as starting clean.

datagero and others added 2 commits July 9, 2024 02:50

[Dynamo] Fix - adding Handler for UserDefinedObjectVariable

10af808

Merge branch 'pytorch:main' into fix-dynamo-handler-str-userdefinedob…

1924ead

…jectvariable

pytorch-bot bot added the module: dynamo label Jul 9, 2024

pytorchbot added the open source label Jul 9, 2024

oulgen reviewed Jul 9, 2024

View reviewed changes

datagero and others added 3 commits July 9, 2024 14:20

[Dynamo] Fix - tests Handler for UserDefinedObjectVariable

9ddabdf

Merge branch 'fix-dynamo-handler-str-userdefinedobjectvariable' of gi…

cef2760

…thub.com:datagero/pytorch into fix-dynamo-handler-str-userdefinedobjectvariable

Merge branch 'pytorch:main' into fix-dynamo-handler-str-userdefinedob…

1b9363f

…jectvariable

oulgen approved these changes Jul 9, 2024

View reviewed changes

datagero and others added 17 commits July 9, 2024 20:06

Merge branch 'pytorch:main' into fix-dynamo-handler-str-userdefinedob…

6419ab5

…jectvariable

[Dynamo] Fix - adding Handler for UserDefinedObjectVariable

e996813

[Dynamo] Fix - tests Handler for UserDefinedObjectVariable

363ed46

[inductor] Add FileCheck to flex attention epilogue test (pytorch#129343

51f8f0d

) Pull Request resolved: pytorch#129343 Approved by: https://github.com/lezcano

fix the use of initial learning rate in the OneCycleLR example (pytor…

e157f82

…ch#130306) Fixes pytorch#127649 Pull Request resolved: pytorch#130306 Approved by: https://github.com/janeyx99

[CMake] [3/N] Remove unused code (pytorch#130322)

6fafe10

Some functions used by Caffe2 were removed along with some outdated checks. Follows pytorch#130006. Pull Request resolved: pytorch#130322 Approved by: https://github.com/r-barnes

add bits16 to graph dtype_abbrs (pytorch#130339)

771df7f

As title, patch the dtype in torch.fx.graph Pull Request resolved: pytorch#130339 Approved by: https://github.com/angelayi

Fix typo (pytorch#130305)

7476ae1

Fixes pytorch#130241 that is a reopen pr of pytorch#130244, for possibly fixing the failed job Pull Request resolved: pytorch#130305 Approved by: https://github.com/Skylion007

[Dynamo] Fix - tests Handler for UserDefinedObjectVariable - lint adj

edca493

[Dynamo] Fix - tests Handler for UserDefinedObjectVariable

b3e831e

Revert "[runtime asserts] deduplicate runtime asserts & CSE (pytorch#…

9b4d9fb

…128599)" This reverts commit 940e447. Reverted pytorch#128599 on behalf of https://github.com/izaitsevfb due to breaking internal APS tests, see D59498864 ([comment](pytorch#128599 (comment)))

datagero and others added 16 commits July 9, 2024 22:24

[Dynamo] Fix - adding Handler for UserDefinedObjectVariable

09b3256

[Dynamo] Fix - tests Handler for UserDefinedObjectVariable

5a01d71

[Dynamo] Fix - tests Handler for UserDefinedObjectVariable - lint adj

bbd6160

[Dynamo] Fix - tests Handler for UserDefinedObjectVariable

81dbf6f

rebase viable

dafb0ee

Revert "Add block mask utility support for batches and heads > 1 (pyt…

698826c

…orch#130227)" This reverts commit 6413998. Reverted pytorch#130227 on behalf of https://github.com/izaitsevfb due to breaks internal builds, please see D59498662 ([comment](pytorch#130227 (comment)))

[structural binding][7/N] Replace std::tie with structural binding (p…

d34b0eb

…ytorch#130216) Follows pytorch#120353 Pull Request resolved: pytorch#130216 Approved by: https://github.com/albanD

[Caffe2] Remove Caffe2 documentation (pytorch#130089)

201f9f1

Due to the removal of Caffe2 code. Pull Request resolved: pytorch#130089 Approved by: https://github.com/r-barnes, https://github.com/albanD

check if the input fx graph of aot_compile return tuple (pytorch#129824)

c7fe101

Fixes pytorch#129719 Pull Request resolved: pytorch#129824 Approved by: https://github.com/angelayi, https://github.com/yushangdi

[4/N] Change #include <c10/util/Optional.h> to #include <optional> (p…

af70deb

…ytorch#130329) Follows pytorch#130300 Pull Request resolved: pytorch#130329 Approved by: https://github.com/ezyang

Add decompositions for copy variants of view ops (pytorch#128416)

c4ef362

Pull Request resolved: pytorch#128416 Approved by: https://github.com/amjames, https://github.com/lezcano

[Dynamo] Fix - tests Handler for UserDefinedObjectVariable

011fae2

[Dynamo] Fix - tests Handler for UserDefinedObjectVariable - dedup

090dcf2

Merge branch 'fix-dynamo-handler-str-userdefinedobjectvariable' of gi…

3b55c68

…thub.com:datagero/pytorch into fix-dynamo-handler-str-userdefinedobjectvariable

anijain2305 approved these changes Jul 10, 2024

View reviewed changes

datagero and others added 3 commits July 10, 2024 16:54

Merge branch 'pytorch:main' into fix-dynamo-handler-str-userdefinedob…

6d826ed

…jectvariable

Revert "Add block mask utility support for batches and heads > 1 (pyt…

2cb35d4

…orch#130227)" This reverts commit 6413998. Reverted pytorch#130227 on behalf of https://github.com/izaitsevfb due to breaks internal builds, please see D59498662 ([comment](pytorch#130227 (comment)))

datagero requested review from albanD, jbschlosser and mikaylagawarecki as code owners July 10, 2024 23:40

pytorch-bot bot added the module: inductor label Jul 10, 2024

datagero closed this Jul 10, 2024

datagero mentioned this pull request Jul 11, 2024

[Dynamo] Fix - str handler for UserDefinedObjectVariable #130506

Closed

datagero deleted the fix-dynamo-handler-str-userdefinedobjectvariable branch July 16, 2024 20:11

datagero restored the fix-dynamo-handler-str-userdefinedobjectvariable branch July 19, 2024 04:12

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Fix dynamo handler str userdefinedobjectvariable #130320

Fix dynamo handler str userdefinedobjectvariable #130320

Uh oh!

datagero commented Jul 9, 2024 •

edited by pytorch-bot bot

Loading

Uh oh!

pytorch-bot bot commented Jul 9, 2024 •

edited

Loading

Uh oh!

oulgen left a comment

Uh oh!

datagero commented Jul 9, 2024

Uh oh!

oulgen left a comment

Uh oh!

datagero commented Jul 9, 2024

Uh oh!

oulgen commented Jul 9, 2024

Uh oh!

anijain2305 commented Jul 10, 2024

Uh oh!

datagero commented Jul 10, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

19 participants

Fix dynamo handler str userdefinedobjectvariable #130320

Fix dynamo handler str userdefinedobjectvariable #130320

Uh oh!

Conversation

datagero commented Jul 9, 2024 • edited by pytorch-bot bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot bot commented Jul 9, 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/130320

❗ 1 Active SEVs

✅ No Failures

Uh oh!

oulgen left a comment

Choose a reason for hiding this comment

Uh oh!

datagero commented Jul 9, 2024

Uh oh!

oulgen left a comment

Choose a reason for hiding this comment

Uh oh!

datagero commented Jul 9, 2024

Uh oh!

oulgen commented Jul 9, 2024

Uh oh!

anijain2305 commented Jul 10, 2024

Uh oh!

datagero commented Jul 10, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

19 participants

datagero commented Jul 9, 2024 •

edited by pytorch-bot bot

Loading

pytorch-bot bot commented Jul 9, 2024 •

edited

Loading