Do our own call into Python for DTensor dispatch by swolchok · Pull Request #166370 · pytorch/pytorch

swolchok · 2025-10-28T04:14:54Z

Stack from ghstack (oldest at bottom):

Another incremental step: don't deal with custom ops on the critical
path, let the dispatcher do that. Take control of calling into Python
on the critical path as well.

cc @H-Huang @awgu @wanchaol @fegin @fduwjj @wz337 @wconstab @d4l3k @pragupta @msaroufim @dcci

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. [ghstack-poisoned]

pytorch-bot · 2025-10-28T04:14:57Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/166370

📄 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

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:

ROCm failures during provisioning step due to network issues

❌ 15 New Failures

As of commit 73ac204 with merge base 84776e1 ():

NEW FAILURES - The following jobs have failed:

inductor / unit-test / inductor-build / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: ‘pybind11::handle get_dtensor_custom_op_handler()’ defined but not used [-Werror=unused-function]
inductor / unit-test / inductor-cpu-build / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: ‘pybind11::handle get_dtensor_custom_op_handler()’ defined but not used [-Werror=unused-function]
inductor / unit-test / inductor-halide-build / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: ‘pybind11::handle get_dtensor_custom_op_handler()’ defined but not used [-Werror=unused-function]
inductor / unit-test / inductor-triton-cpu-build / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: ‘pybind11::handle get_dtensor_custom_op_handler()’ defined but not used [-Werror=unused-function]
pull / cuda12.8-py3.10-gcc9-sm75 / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: ‘pybind11::handle get_dtensor_custom_op_handler()’ defined but not used [-Werror=unused-function]
pull / linux-jammy-cuda12.8-cudnn9-py3.10-clang12 / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: unused function 'get_dtensor_custom_op_handler' [-Werror,-Wunused-function]
pull / linux-jammy-py3.10-clang12 / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: unused function 'get_dtensor_custom_op_handler' [-Werror,-Wunused-function]
pull / linux-jammy-py3.10-clang12-onnx / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: unused function 'get_dtensor_custom_op_handler' [-Werror,-Wunused-function]
pull / linux-jammy-py3.10-clang18-asan / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: unused function 'get_dtensor_custom_op_handler' [-Werror,-Wunused-function]
pull / linux-jammy-py3.10-gcc11 / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: ‘pybind11::handle get_dtensor_custom_op_handler()’ defined but not used [-Werror=unused-function]
pull / linux-jammy-py3.10-gcc11-mobile-lightweight-dispatch-build / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: ‘pybind11::handle get_dtensor_custom_op_handler()’ defined but not used [-Werror=unused-function]
pull / linux-jammy-py3.10-gcc11-no-ops / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: ‘pybind11::handle get_dtensor_custom_op_handler()’ defined but not used [-Werror=unused-function]
pull / linux-jammy-py3.10-gcc11-pch / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: ‘pybind11::handle get_dtensor_custom_op_handler()’ defined but not used [-Werror=unused-function]
pull / linux-jammy-py3.13-clang12 / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: unused function 'get_dtensor_custom_op_handler' [-Werror,-Wunused-function]
pull / linux-jammy-xpu-n-py3.10 / build (gh)
/var/lib/jenkins/workspace/torch/csrc/autograd/python_variable.cpp:961:5: error: ‘pybind11::handle get_dtensor_custom_op_handler()’ defined but not used [-Werror=unused-function]

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

…spatch" Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. [ghstack-poisoned]

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. [ghstack-poisoned]

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. ghstack-source-id: 1327e9f Pull Request resolved: #166370

swolchok · 2025-10-29T00:00:23Z

torch/csrc/autograd/python_variable.cpp

+      // REVIEW: without the device change here,
+      // test_dtensor_save_load breaks. I'm not at all sure why it was
+      // OK before.


based on test failures, changing the device to meta here clearly isn't correct . will go back.

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. [ghstack-poisoned]

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. ghstack-source-id: c5cfbf4 Pull Request resolved: #166370

swolchok · 2025-10-29T04:16:45Z

running into some weird coupling here. test_dtensor_save_load is broken. (repro: python test/distributed/tensor/test_dtensor.py -k test_dtensor_save_load, you need to patch the test to reduce world_size to 1 if you have only one GPU)
Apparently, the presence of DTensor.__torch_dispatch__ somehow causes DTensors' storage to have a valid data_ptr and thus be compatible with serialization. This PR deletes DTensor.__torch_dispatch__ and breaks it; I know it's the mere presence of DTensor.__torch_dispatch__ because 1) putting it back fixes the test 2) printing inside DTensor.__torch_dispatch__ shows that it's not called before or during serialization.

…l into Python for DTensor dispatch" Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. ghstack-source-id: d19b694 Pull Request resolved: #166370

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. ghstack-source-id: 45022fc Pull Request resolved: #166370

…o our own call into Python for DTensor dispatch" Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. ghstack-source-id: 8a4b52c Pull Request resolved: #166370

swolchok · 2025-10-30T15:08:21Z

Current failure looks like Dynamo needs to know about the DTensor dispatch key somehow; we're hitting __torch_dispatch__ path for a DTensor (AIUI it's actually a FakeTensor wrapping a DTensor somehow, so perhaps the way the FakeTensor is created is wrong? maybe we can make that impossible?)

swolchok · 2025-10-30T16:56:13Z

torch/_tensor.py

+            and (
+                type(self).__torch_dispatch__ is not torch.Tensor.__torch_dispatch__
+                or _is_dtensor(self)
+            )


not sure how to structurally avoid is_dtensor special cases in N things that currently just care about __torch_dispatch__

thinking about whether we put DTensor.torch_dispatch back and have it just upcall to the DTensor dispatch key stuff

…call into Python for DTensor dispatch" Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. ghstack-source-id: 0937067 Pull Request resolved: #167051

…dd C++ fast path for `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. ghstack-source-id: 2f0ea67 Pull Request resolved: #167051

This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. ghstack-source-id: b3451db Pull Request resolved: #167051

…r `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…ch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…n "Add C++ fast path for `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…path for `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. ghstack-source-id: 672fa5d Pull Request resolved: #167051

…or.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…tch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

swolchok · 2025-11-06T21:37:12Z

superseded by #167051

… path for `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…or.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…patch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…th for `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…nd add a couple comments on "Add C++ fast path for `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…comments on "Add C++ fast path for `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

… gating in is_dtensor on "Add C++ fast path for `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…ensor on "Add C++ fast path for `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…C++ fast path for `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

…r `DTensor.__torch_dispatch__`" This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. cc H-Huang awgu wanchaol fegin fduwjj wz337 wconstab d4l3k pragupta msaroufim dcci [ghstack-poisoned]

This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike #166370 and #166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of #166370. Pull Request resolved: #167051 Approved by: https://github.com/ezyang

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. ghstack-source-id: 453ddca Pull Request resolved: pytorch/pytorch#166370

This patches the `__torch_dispatch__` machinery to detect DTensor and hand over control to a C++ fast path. Unlike pytorch#166370 and pytorch#166369 (which added a DTensor dispatch key and are intended to be replaced by this PR), this approach fundamentally *is* `__torch_dispatch__`, hopefully sidestepping all manner of thorny "does it work just like `__torch_dispatch__`?" that came up during development and review of pytorch#166370. Pull Request resolved: pytorch#167051 Approved by: https://github.com/ezyang

Do our own call into Python for DTensor dispatch

9161c3b

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. [ghstack-poisoned]

swolchok requested review from albanD and soulitzer as code owners October 28, 2025 04:14

pytorch-bot bot added ciflow/inductor oncall: distributed Add this issue/PR to distributed oncall triage queue labels Oct 28, 2025

fix bottom-of-stack PR on "Do our own call into Python for DTensor di…

0f0e55e

…spatch" Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. [ghstack-poisoned]

swolchok mentioned this pull request Oct 28, 2025

Avoid creating Python OpSchema in the DTensor dispatch fast path #166372

Closed

Update on "Do our own call into Python for DTensor dispatch"

4a032a1

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. [ghstack-poisoned]

swolchok commented Oct 29, 2025

View reviewed changes

put it back on "Do our own call into Python for DTensor dispatch"

51a660f

Another incremental step: don't deal with custom ops on the critical path, let the dispatcher do that. Take control of calling into Python on the critical path as well. [ghstack-poisoned]

swolchok added the topic: not user facing topic category label Oct 29, 2025

swolchok commented Oct 30, 2025

View reviewed changes

swolchok closed this Nov 6, 2025

github-actions bot deleted the gh/swolchok/862/head branch December 7, 2025 02:20

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Do our own call into Python for DTensor dispatch#166370

Do our own call into Python for DTensor dispatch#166370
swolchok wants to merge 17 commits intogh/swolchok/862/basefrom
gh/swolchok/862/head

swolchok commented Oct 28, 2025 •

edited

Loading

Uh oh!

pytorch-bot bot commented Oct 28, 2025 •

edited

Loading

Uh oh!

swolchok Oct 29, 2025

Uh oh!

swolchok commented Oct 29, 2025

Uh oh!

swolchok commented Oct 30, 2025

Uh oh!

swolchok Oct 30, 2025

Uh oh!

swolchok Oct 30, 2025

Uh oh!

swolchok commented Nov 6, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Conversation

swolchok commented Oct 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot bot commented Oct 28, 2025 • 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/166370

❗ 1 Active SEVs

❌ 15 New Failures

Uh oh!

swolchok Oct 29, 2025

Choose a reason for hiding this comment

Uh oh!

swolchok commented Oct 29, 2025

Uh oh!

swolchok commented Oct 30, 2025

Uh oh!

swolchok Oct 30, 2025

Choose a reason for hiding this comment

Uh oh!

swolchok Oct 30, 2025

Choose a reason for hiding this comment

Uh oh!

swolchok commented Nov 6, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

swolchok commented Oct 28, 2025 •

edited

Loading

pytorch-bot bot commented Oct 28, 2025 •

edited

Loading