Search results

image diffusion model output changes for the worse after 530ff92655

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

module: cublasProblem related to cublas supportProblem related to cublas support

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: numerical-stabilityProblems related to numerical stability of operationsProblems related to numerical stability of operations

Status: Open.

#181737 In pytorch/pytorch;

· zou3519 opened on Apr 28, 2026

·  

2.12.0

[vLLM][inductor][triton] decompose_triton_kernel_wrapper_functional AssertionError under dynamic shapes

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

high priority

module: crashProblem manifests as a hard crash, as opposed to a RuntimeErrorProblem manifests as a hard crash, as opposed to a RuntimeError

module: dynamic shapes

module: inductor

module: user tritonrelated to ability to directly torch.compile triton kernelsrelated to ability to directly torch.compile triton kernels

module: vllm

oncall: pt2

triage review

Status: Open.

#181735 In pytorch/pytorch;

· ZJY0516 opened on Apr 28, 2026

[MPS] nn.MultiheadAttention is ~9x slower than direct F.scaled_dot_product_attention with bit-identical output (B=1)

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

module: mpsRelated to Apple Metal Performance Shaders frameworkRelated to Apple Metal Performance Shaders framework

module: nnRelated to torch.nnRelated to torch.nn

module: performanceIssues related to performance, either of kernel code or framework glueIssues related to performance, either of kernel code or framework glue

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.

#181725 In pytorch/pytorch;

· nuri-yoo opened on Apr 28, 2026

torch.cuda.make_graphed_callables can silently corrupt parameter gradient accumulation by returning static grad buffers

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

module: autogradRelated to torch.autograd, and the autograd engine in generalRelated to torch.autograd, and the autograd engine in general

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

triage review

Status: Open.

#181723 In pytorch/pytorch;

· buptzyb opened on Apr 28, 2026

bf16 matmul throughput drops on RTX 4090 for shapes where N%16==8

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

matrix multiplication

module: bfloat16

module: cublasProblem related to cublas supportProblem related to cublas support

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: linear algebraIssues related to specialized linear algebra operations in PyTorch; includes matrix multiply matmulIssues related to specialized linear algebra operations in PyTorch; includes matrix multiply matmul

module: performanceIssues related to performance, either of kernel code or framework glueIssues related to performance, either of kernel code or framework glue

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.

#181718 In pytorch/pytorch;

· Arthur-St-06 opened on Apr 28, 2026

DISABLED test_control_deps_orders_void_op_across_nested_calls (__main__.TestControlDeps)

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

module: inductor

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

module: xpuIntel XPU related issuesIntel XPU related issues

oncall: pt2

skippedDenotes a (flaky) test currently skipped in CI.Denotes a (flaky) test currently skipped in CI.

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.

#181701 In pytorch/pytorch;

· guangyey opened on Apr 28, 2026

DISABLED test_max_reads_limits_fusion (__main__.OverFusionTest)

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

module: inductor

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

module: xpuIntel XPU related issuesIntel XPU related issues

oncall: pt2

skippedDenotes a (flaky) test currently skipped in CI.Denotes a (flaky) test currently skipped in CI.

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.

#181699 In pytorch/pytorch;

· guangyey opened on Apr 28, 2026

Eval-mode BatchNorm/GroupNorm produces wrong output under torch.compile(backend='inductor')

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

module: correctness (silent)issue that returns an incorrect result silentlyissue that returns an incorrect result silently

module: inductor

module: norms and normalization

oncall: pt2

triage review

Status: Open.

#181696 In pytorch/pytorch;

· jasminetrail opened on Apr 28, 2026

torch.where with broadcast shapes produces wrong output under torch.compile(backend='inductor')

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

module: correctness (silent)issue that returns an incorrect result silentlyissue that returns an incorrect result silently

module: inductor

oncall: pt2

triage review

Status: Open.

#181695 In pytorch/pytorch;

· jasminetrail opened on Apr 28, 2026

torch.addcdiv produces wrong output under torch.compile(backend='inductor')

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

module: correctness (silent)issue that returns an incorrect result silentlyissue that returns an incorrect result silently

module: inductor

oncall: pt2

triage review

Status: Open.

#181694 In pytorch/pytorch;

· jasminetrail opened on Apr 28, 2026

nn.Hardswish produces wrong output under torch.compile(backend='inductor')

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

module: inductor

oncall: pt2

triage review

Status: Open.

#181693 In pytorch/pytorch;

· jasminetrail opened on Apr 28, 2026

nn.RReLU().eval() produces wrong output under torch.compile(backend='inductor')

bot-triagedThis is a label only to be used by the auto triage botThis is a label only to be used by the auto triage bot

module: correctness (silent)issue that returns an incorrect result silentlyissue that returns an incorrect result silently

module: inductor

module: nnRelated to torch.nnRelated to torch.nn

oncall: pt2

triage review

Status: Open.

#181692 In pytorch/pytorch;

· jasminetrail opened on Apr 28, 2026