-
Notifications
You must be signed in to change notification settings - Fork 27k
Pull requests: pytorch/pytorch
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[dynamo] Preserve Python try/except semantics for AttributeError in tensor getattr
module: dynamo
#176149
opened Mar 2, 2026 by
Gouri174
Loading…
update previous version 2.10 installation in get start xpu
open source
topic: not user facing
topic category
#176141
opened Mar 2, 2026 by
ZhaoqiongZ
Loading…
[Dist][CI] share PGs to shorten distributed unit test
release notes: distributed (c10d)
release notes category
Fix flaky test_extra_collectives by disabling shape padding
topic: not user facing
topic category
#176137
opened Mar 2, 2026 by
aorenste
Loading…
Fix ROCm build in CUDACachingAllocator
ciflow/rocm-mi300
Trigger "default" config CI on ROCm MI300
module: rocm
AMD GPU support for Pytorch
topic: not user facing
topic category
#176136
opened Mar 2, 2026 by
jeanschmidt
Loading…
Add TORCH_VERSION_2_12_0 guard to torch_from_blob_v2 implementation
topic: not user facing
topic category
#176135
opened Mar 2, 2026 by
cluePrints
Loading…
Align input types of lambda function in logical reduce kernels with i…
ciflow/rocm-mi355
Trigger "default" config CI on ROCm MI355 runners
open source
release notes: cuda
release notes category
#176132
opened Mar 2, 2026 by
anatoliylitv
•
Draft
Fix test interaction: clean up DTensorSpec pytree registration
topic: not user facing
topic category
#176128
opened Mar 2, 2026 by
aorenste
Loading…
Fix flaky DTensor sharding prop cache logging test
ciflow/inductor
release notes: distributed (dtensor)
release notes category
#176127
opened Mar 2, 2026 by
aorenste
Loading…
[dynamo] Track has_side_effect on SubgraphTracer
ciflow/inductor
module: dynamo
#176124
opened Mar 2, 2026 by
anijain2305
Loading…
[xpu][draft] Upgrade OneDNN XPU to v3.11 in 2.11-rc3
ciflow/binaries_wheel
Trigger binary build and upload jobs for wheel on the PR
ciflow/xpu
Run XPU CI tasks
module: mkldnn
Related to Intel IDEEP or oneDNN (a.k.a. mkldnn) integration
open source
#176122
opened Mar 2, 2026 by
LuFinch
Loading…
[dynamo] Disallow side effects in invoke_subgraph
ciflow/inductor
ciflow/trunk
Trigger trunk jobs on your pull request
module: dynamo
topic: not user facing
topic category
#176121
opened Mar 2, 2026 by
anijain2305
Loading…
[dynamo] Add auto guard cache for invoke_subgraph
ciflow/inductor
module: dynamo
#176120
opened Mar 2, 2026 by
anijain2305
•
Draft
Fix flaky DTensor sharding prop cache logging test
ciflow/inductor
release notes: distributed (dtensor)
release notes category
#176119
opened Mar 2, 2026 by
aorenste
Loading…
[inductor] Test coverage for custom op out-variant lowering
ciflow/inductor
module: inductor
topic: not user facing
topic category
#176118
opened Mar 2, 2026 by
tianrengao
Loading…
[inductor] Multi-output custom op ExternKernelOut lowering
ciflow/inductor
module: inductor
#176117
opened Mar 2, 2026 by
tianrengao
Loading…
[dynamo] Add auto guard cache for invoke_subgraph
ciflow/inductor
module: dynamo
#176106
opened Mar 2, 2026 by
anijain2305
Loading…
[dynamo] Add is_auto_cacheable check for invoke_subgraph
ciflow/inductor
module: dynamo
#176105
opened Mar 2, 2026 by
anijain2305
•
Draft
[MPS] Migrate fill_ to native Metal kernels
ciflow/mps
Run MPS tests (subset of trunk)
release notes: mps
Release notes category
topic: improvements
topic category
Fix user-defined Triton kernel name mangling in Inductor codegen
module: inductor
open source
topic: not user facing
topic category
#176100
opened Mar 1, 2026 by
universeplayer
Loading…
[MPS] Add regression test for Run MPS tests (subset of trunk)
open source
topic: not user facing
topic category
F.conv2d with large out_channels
ciflow/mps
#176099
opened Mar 1, 2026 by
hvaara
Loading…
[Inductor] allow_32 to fp32_precision
ciflow/inductor
ciflow/trunk
Trigger trunk jobs on your pull request
module: inductor
open source
topic: not user facing
topic category
#176098
opened Mar 1, 2026 by
khushali9
Loading…
Fix SDPA mem-efficient attention CUDA grid overflow for large batch*num_heads (#142228)
open source
#176096
opened Mar 1, 2026 by
Bias92
Loading…
fix cuda torch.topk index bug for super long input which are over 32-bit INT_MAX length
open source
release notes: cuda
release notes category
#176095
opened Mar 1, 2026 by
wangtianxia-sjtu
Loading…
Previous Next
ProTip!
Exclude everything labeled
bug with -label:bug.