Search results

[flex_attention] replace sliced BlockMask noop with helpful error

module: flex attention

module: inductor

open source

release notes: nnrelease notes categoryrelease notes category

Status: Open (in progress).

pytorch/pytorchnumber 164702

#164702 In pytorch/pytorch;

· Aminsed opened on Oct 5, 2025

[dynamo] Fix flex attention graph break on create_block_mask

module: dynamo

module: flex attention

oncall: pt2

open source

release notes: dynamo

Status: Open (in progress).

pytorch/pytorchnumber 164349

#164349 In pytorch/pytorch;

· jmaczan opened on Oct 1, 2025

fix flex attention eager bwd: more rounding

ciflow/inductor

ciflow/trunkTrigger trunk jobs on your pull requestTrigger trunk jobs on your pull request

module: flex attention

module: inductor

release notes: nnrelease notes categoryrelease notes category

Status: Open (in progress).

pytorch/pytorchnumber 164317

#164317 In pytorch/pytorch;

· v0i0 opened on Oct 1, 2025

flex_attention doesn't work when wrapped with inductor backend

intelThis tag is for PR from IntelThis tag is for PR from Intel

module: flex attention

module: higher order operatorstorch.cond and similartorch.cond and similar

module: pt2-dispatcherPT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,PT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,

oncall: pt2

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#164248 In pytorch/pytorch;

· tugsbayasgalan opened on Sep 30, 2025

Dynamo graph break on flex attention code

module: dynamo

module: flex attention

module: graph breaks

module: higher order operatorstorch.cond and similartorch.cond and similar

module: pt2-dispatcherPT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,PT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,

oncall: pt2

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#164247 In pytorch/pytorch;

· tugsbayasgalan opened on Sep 30, 2025

Flexattention create mask function return a bool tensor when using score mod

module: flex attention

module: pt2-dispatcherPT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,PT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,

oncall: pt2

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#164096 In pytorch/pytorch;

· paoxiaode opened on Sep 29, 2025

Add tensor descriptor implementation to Flex backwards template

module: flex attention

module: pt2-dispatcherPT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,PT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,

module: sdpaAll things related to torch.nn.functional.scaled_dot_product_attentiionAll things related to torch.nn.functional.scaled_dot_product_attentiion

oncall: pt2

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163543 In pytorch/pytorch;

· whitneywhtsang opened on Sep 22, 2025

Rename <code>USE_TMA</code> to better reflect tensor descriptor functionality

module: docsRelated to our documentation, both in docs/ and docblocksRelated to our documentation, both in docs/ and docblocks

module: flex attention

module: higher order operatorstorch.cond and similartorch.cond and similar

module: pt2-dispatcherPT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,PT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,

oncall: pt2

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163536 In pytorch/pytorch;

· whitneywhtsang opened on Sep 22, 2025

[vLLM] Crash during inferece in torch dynamo on 2xRTX2080Ti

module: flex attention

module: higher order operatorstorch.cond and similartorch.cond and similar

module: pt2-dispatcherPT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,PT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,

oncall: pt2

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

vllm-compile

Status: Open.

#163428 In pytorch/pytorch;

· ir1ka opened on Sep 21, 2025

Flex Attention Decoding Silently Incorrect with H=None Broadcasting and Sliced mask_mod

high priority

module: flex attention

module: pt2-dispatcherPT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,PT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163314 In pytorch/pytorch;

· zaptrem opened on Sep 19, 2025

[CI][CUDA][Distributed] test_ring_flex_attention failed on 8xB200 Runner

module: ciRelated to continuous integrationRelated to continuous integration

module: flex attention

module: higher order operatorstorch.cond and similartorch.cond and similar

module: pt2-dispatcherPT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,PT2 dispatcher-related issues (e.g., aotdispatch, functionalization, faketensor, custom-op,

module: testsIssues related to tests (not the torch.testing module)Issues related to tests (not the torch.testing module)

oncall: distributedAdd this issue/PR to distributed oncall triage queueAdd this issue/PR to distributed oncall triage queue

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#162820 In pytorch/pytorch;

· nWEIdia opened on Sep 12, 2025

fp8-kv-cache-flex

ciflow/inductor

module: flex attention

module: inductor

topic: new featurestopic categorytopic category

Status: Open (in progress).

pytorch/pytorchnumber 162679

#162679 In pytorch/pytorch;

· drisspg opened on Sep 11, 2025