Amin Sedaghat 767199fd9b [flex_attention] replace sliced BlockMask noop with helpful error (#164702) ...

Fixes part of #163314

After slicing BlockMask with `[]`, mask_mod was silently replaced with noop_mask. This caused silent incorrect results when users applied transformations to `sliced_mask.mask_mod`.

Replace noop with `_sliced_mask_mod_error` that raises RuntimeError with guidance to use `base_mask.mask_mod` instead.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/164702
Approved by: https://github.com/drisspg, https://github.com/BoyuanFeng

2025-10-20 03:46:16 +00:00

..

attention

[flex_attention] replace sliced BlockMask noop with helpful error (#164702)

2025-10-20 03:46:16 +00:00

backends

[BE][Easy] enable UFMT for torch/nn/parallel (#128596)

2024-06-17 16:29:22 +00:00

intrinsic

Remove unnecessary noqa suppressions (#164106)

2025-10-18 04:52:41 +00:00

modules

Remove unnecessary noqa suppressions (#164106)

2025-10-18 04:52:41 +00:00

parallel

Remove unnecessary noqa suppressions (#164106)

2025-10-18 04:52:41 +00:00

qat

[BE][PYFMT] migrate PYFMT for {torch,test}/{nn,optim}/** to ruff format (#144548)

2025-06-14 11:27:04 +00:00

quantizable

[BE][PYFMT] migrate PYFMT for {torch,test}/{nn,optim}/** to ruff format (#144548)

2025-06-14 11:27:04 +00:00

quantized

[BE][PYFMT] migrate PYFMT for {torch,test}/{nn,optim}/** to ruff format (#144548)

2025-06-14 11:27:04 +00:00

utils

Fix self assignment (#165816)

2025-10-18 18:51:52 +00:00

__init__.py

[BE][Easy][17/19] enforce style for empty lines in import segments in torch/[a-c]*/ and torch/[e-n]*/ (#129769)

2024-08-04 10:24:09 +00:00

_reduction.py

[BE] enable UFMT for torch/nn/*.py (#128593)

2024-06-23 16:05:13 +00:00

common_types.py

[2/N] Apply ruff UP035 check in torch files (#164054)

2025-09-29 03:35:32 +00:00

cpp.py

Pyrefly suppressions 7/n (#164913)

2025-10-08 07:27:17 +00:00

functional.py

Enable all SIM rules except disabled ones (#164645)

2025-10-17 07:27:11 +00:00

functional.pyi.in

Revert "Add label_smoothing param in nn.BCELoss and nn.BCEWithLogitsLoss (#150282)"

2025-08-13 09:01:52 +00:00

grad.py

[BE] enable UFMT for torch/nn/*.py (#128593)

2024-06-23 16:05:13 +00:00

init.py

Pyrefly suppressions 7/n (#164913)

2025-10-08 07:27:17 +00:00

parameter.py

Pyrefly suppressions 7/n (#164913)

2025-10-08 07:27:17 +00:00

parameter.pyi

remove allow-untyped-defs from torch/nn/parameter.pyi (#144654)

2025-01-13 19:02:31 +00:00