Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit History

Commits on Jun 5, 2026

• Revert "[Bugfix] Raise error when cuda graph is captured during compile tracing. (#185410)"

  Show description for 7e04a7a
  pytorchmergebot

  committed
  7e04a7a

  Copy full SHA for 7e04a7a

• Drop inputs_to_check from FxGraphCache key; rely on runtime realignment (#185374)

  Show description for 2f8ac03
  

  oulgen

  authored and

  pytorchmergebot

  committed
  2f8ac03

  Copy full SHA for 2f8ac03

• Add experimental profiler integration for the CUPTI monitor (#186037)

  Show description for a87df7b
  

  ngimel

  authored and

  pytorchmergebot

  committed
  a87df7b

  Copy full SHA for a87df7b

• [reland][Inductor][X86] Remove deprecated fusion patterns (#178466)

  Show description for cefb7af
  

  cyxlily

  authored and

  pytorchmergebot

  committed
  cefb7af

  Copy full SHA for cefb7af

• Bring in gemm kernels form quack and CI testing for patch set (#186284)

  Show description for 8e484e4
  

  drisspg

  authored and

  pytorchmergebot

  committed
  8e484e4

  Copy full SHA for 8e484e4

• [dtensor] single dim strategies auto infra (#185386)

  Show description for 2741b89
  

  anshul-si

  authored and

  pytorchmergebot

  committed
  2741b89

  Copy full SHA for 2741b89

• [xpu][fix] Fix segfault in _scaled_mm_xpu_v2 due to empty swizzle vector access (#186187)

  Show description for 4b1cfae
  

  Stonepia

  authored and

  pytorchmergebot

  committed
  4b1cfae

  Copy full SHA for 4b1cfae

• Normalize device_type in distributed reordering/logger tests (#186169)

  Show description for e98035f
  

  cyyever

  authored and

  pytorchmergebot

  committed
  e98035f

  Copy full SHA for e98035f

• Fix missing stride() call in test_redistribute TensorMeta (#186170)

  Show description for 1c02769
  

  cyyever

  authored and

  pytorchmergebot

  committed
  1c02769

  Copy full SHA for 1c02769

• [Bugfix] Raise error when cuda graph is captured during compile tracing. (#185410)

  Show description for 87cd0be
  

  morrison-turnansky

  authored and

  pytorchmergebot

  committed
  87cd0be

  Copy full SHA for 87cd0be

• [1/N][Test] Migrate TestFakeTensor and TestTags device-agnostic skips to `@skipOps` (#184353)

  Show description for 2091905
  

  KarhouTam

  authored and

  pytorchmergebot

  committed
  2091905

  Copy full SHA for 2091905

• [inductor] Fix missing libtorch_python symbols in FBCODE when using cpp-wrapper (#185551) (#185551)

  Show description for f3f2c23
  

  desertfire

  authored and

  pytorchmergebot

  committed
  f3f2c23

  Copy full SHA for f3f2c23

• [Operators] Implement bitwise xor operator in Dynamo (#184789)

  Show description for fc613e1
  

  guilhermeleobas

  authored and

  pytorchmergebot

  committed
  fc613e1

  Copy full SHA for fc613e1

• [Operators] Implement bitwise and operator in Dynamo (#184788)

  Show description for a2e07f1
  

  guilhermeleobas

  authored and

  pytorchmergebot

  committed
  a2e07f1

  Copy full SHA for a2e07f1

• [inductor] simplify dynamic cat indexing (#184351)

  Show description for 784e50b
  

  jansel

  authored and

  pytorchmergebot

  committed
  784e50b

  Copy full SHA for 784e50b

Commits on Jun 4, 2026

• Rename coalesced collective variants to _single_coalesced (#186144)

  Show description for 22fe6f6
  

  d4l3k

  authored and

  pytorchmergebot

  committed
  22fe6f6

  Copy full SHA for 22fe6f6

• Migrate ProcessGroup method usages to the _single names (#186135)

  Show description for 74063cf
  

  d4l3k

  authored and

  pytorchmergebot

  committed
  74063cf

  Copy full SHA for 74063cf

• Rename functional collectives to _single naming scheme (#186125)

  Show description for 1e228b6
  

  d4l3k

  authored and

  pytorchmergebot

  committed
  1e228b6

  Copy full SHA for 1e228b6

• [DyanmicSpec] Add STATIC sentinel for shapes_spec dim slots (#185165)

  Show description for 8bd77e4
  

  laithsakka

  authored and

  pytorchmergebot

  committed
  8bd77e4

  Copy full SHA for 8bd77e4

• [TEST][B200] Bump tolerances in `test_conv2d_backward_parametrized` (#185994)

  Show description for a55a4c2
  

  eqy

  authored and

  pytorchmergebot

  committed
  a55a4c2

  Copy full SHA for a55a4c2

• Handle wrap_triton under Triton interpreter mode (#185597)

  Show description for 52e0deb
  

  jansel

  authored and

  pytorchmergebot

  committed
  52e0deb

  Copy full SHA for 52e0deb

• [inductor] Avoid Python fallback for cpp custom op lists (#184099)

  Show description for cc80040
  

  jansel

  authored and

  pytorchmergebot

  committed
  cc80040

  Copy full SHA for cc80040

• Add hidden experimental CUPTI monitor API (#186034)

  Show description for 8054e1b
  

  ngimel

  authored and

  pytorchmergebot

  committed
  8054e1b

  Copy full SHA for 8054e1b

• [MPS] Support captured buffers in flex attention (#186215)

  Show description for 4d0ea4b
  

  Isalia20

  authored and

  pytorchmergebot

  committed
  4d0ea4b

  Copy full SHA for 4d0ea4b

• Fix standalone_compile training artifact save (#185635)

  Show description for f5d8a56
  

  jansel

  authored and

  pytorchmergebot

  committed
  f5d8a56

  Copy full SHA for f5d8a56

• [inductor] Resolve relative TORCHINDUCTOR_CACHE_DIR (#185723)

  Show description for 289bda0
  

  jansel

  authored and

  pytorchmergebot

  committed
  289bda0

  Copy full SHA for 289bda0

• Fix Dynamo .grad reads for new in-graph parameters (#184972)

  Show description for 16b8bf1
  

  jansel

  authored and

  pytorchmergebot

  committed
  16b8bf1

  Copy full SHA for 16b8bf1

• Fix private convolution fake symint handling (#185081)

  Show description for 363a8d9
  

  jansel

  authored and

  pytorchmergebot

  committed
  363a8d9

  Copy full SHA for 363a8d9

• Inline decomposed quantization helpers in Dynamo (#185628)

  Show description for 3d4fbc2
  

  jansel

  authored and

  pytorchmergebot

  committed
  3d4fbc2

  Copy full SHA for 3d4fbc2

• [dynamo] Fast path guardless cache hits (#184683)

  Show description for 3f4bfa6
  

  jansel

  authored and

  pytorchmergebot

  committed
  3f4bfa6

  Copy full SHA for 3f4bfa6

• [inductor] Keep C++ loop split groups aligned (#183715)

  Show description for cdc73d0
  

  jansel

  authored and

  pytorchmergebot

  committed
  cdc73d0

  Copy full SHA for cdc73d0

• Clean up export buffer hooks on trace failures (#184956)

  Show description for 3c49b74
  

  jansel

  authored and

  pytorchmergebot

  committed
  3c49b74

  Copy full SHA for 3c49b74

• Simplify Min/Max of scaled symbolic terms (#185092)

  Show description for e73863f
  

  jansel

  authored and

  pytorchmergebot

  committed
  e73863f

  Copy full SHA for e73863f

• Fix lp_pool infinity norm handling (#183997)

  Show description for e081eff
  

  jansel

  authored and

  pytorchmergebot

  committed
  e081eff

  Copy full SHA for e081eff

• Fix ONNX type promotion overload selection (#185005)

  Show description for 5ed7697
  

  jansel

  authored and

  pytorchmergebot

  committed
  5ed7697

  Copy full SHA for 5ed7697