Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Oct 3, 2024

• Merge branch 'pytorch:main' into main

  jakeharmon8

  authored
  db63884

  Copy full SHA for db63884

• [cuDNN] Check shapes during graph capture in cuDNN CTCLoss (pytorch#130071)

  Show description for 46f158b
  

  2 peopleauthored and

  pytorchmergebot

  committed
  46f158b

  Copy full SHA for 46f158b

• [PyTorch] Port ExecuTorch bfdot improvement back to ATen BlasKernel (pytorch#136331)

  Show description for 592e3a3
  

  swolchok

  authored and

  pytorchmergebot

  committed
  592e3a3

  Copy full SHA for 592e3a3

• [PyTorch] Add attribute version of C10_ALWAYS_INLINE (pytorch#136445)

  Show description for c8a7da3
  

  swolchok

  authored and

  pytorchmergebot

  committed
  c8a7da3

  Copy full SHA for c8a7da3

• Revert "Fix unlift to unblock training IR + run_decomp on aliasing constants (pytorch#137162)"

  Show description for 525f671
  pytorchmergebot

  committed
  525f671

  Copy full SHA for 525f671

• [FR] Fix duplicate output for the case when not all ranks join on collective (pytorch#137256)

  Show description for c7714b8
  

  fduwjj

  authored and

  pytorchmergebot

  committed
  c7714b8

  Copy full SHA for c7714b8

• Python CAPI cleanup (pytorch#137266)

  Show description for adc48a5
  

  albanD

  authored and

  pytorchmergebot

  committed
  adc48a5

  Copy full SHA for adc48a5

• [inductor] parallel compile: add import of thread_safe_fork for internal (pytorch#137155)

  Show description for 8bb8c39
  

  masnesral

  authored and

  pytorchmergebot

  committed
  8bb8c39

  Copy full SHA for 8bb8c39

• Fix unlift to unblock training IR + run_decomp on aliasing constants (pytorch#137162)

  Show description for f96020c
  

  tugsbayasgalan

  authored and

  pytorchmergebot

  committed
  f96020c

  Copy full SHA for f96020c

• [AOTAutogradCache] add config for AOTAutograd remote cache (pytorch#137011)

  Show description for 4d3c0fc
  

  jamesjwu

  authored and

  pytorchmergebot

  committed
  4d3c0fc

  Copy full SHA for 4d3c0fc

• type _dynamo/external_utils.py (pytorch#137185)

  Show description for a569a8a
  

  bobrenjc93

  authored and

  pytorchmergebot

  committed
  a569a8a

  Copy full SHA for a569a8a

• Fix serialization for torch.uint16, torch.uint32, torch.uint64 (pytorch#137184)

  Show description for b6cb174
  

  mikaylagawarecki

  authored and

  pytorchmergebot

  committed
  b6cb174

  Copy full SHA for b6cb174

• Implement AcceleratorHooksInterface's virtual functions deviceCount() and getCurrentDevice() for CUDA and XPU (pytorch#136752)

  Show description for 89b7a5d
  

  shink

  authored and

  pytorchmergebot

  committed
  89b7a5d

  Copy full SHA for 89b7a5d

• Add py3.13t linux wheel build (pytorch#137127)

  Show description for 63bbf71
  

  atalman

  authored and

  pytorchmergebot

  committed
  63bbf71

  Copy full SHA for 63bbf71

• [async-tp] fix a race condition that can cause silent correctness issue (pytorch#137199)

  Show description for 38114ec
  

  yifuwang

  authored and

  pytorchmergebot

  committed
  38114ec

  Copy full SHA for 38114ec

• Avoid __ne__ weakref comparison and use identity instead in cache_size.py (pytorch#135000)

  Show description for f166d62
  

  vmoens

  authored and

  pytorchmergebot

  committed
  f166d62

  Copy full SHA for f166d62

• cond_batch_rule with boolean pred (pytorch#135009)

  Show description for bd9517c
  

  vmoens

  authored and

  pytorchmergebot

  committed
  bd9517c

  Copy full SHA for bd9517c

• Revert "raw_alloc ignores PYTORCH_NO_CUDA_MEMORY_CACHING (pytorch#131114)"

  Show description for 0d1701f
  pytorchmergebot

  committed
  0d1701f

  Copy full SHA for 0d1701f

• [compiled autograd] initialize cudagraph tls from context manager (pytorch#136735)

  Show description for 87bf2a8
  

  xmfan

  authored and

  pytorchmergebot

  committed
  87bf2a8

  Copy full SHA for 87bf2a8

• Unify cpp_extension build directory removal (pytorch#136059)

  Show description for b86269f
  

  xmfan

  authored and

  pytorchmergebot

  committed
  b86269f

  Copy full SHA for b86269f

• [DTensor] Register replication strategy for a few upsampling interpolate ops (pytorch#137201)

  Show description for 55c343f
  

  wz337

  authored and

  pytorchmergebot

  committed
  55c343f

  Copy full SHA for 55c343f

• Move _is_static_problem to mm_common (pytorch#137150)

  Show description for 84cac35
  

  drisspg

  authored and

  pytorchmergebot

  committed
  84cac35

  Copy full SHA for 84cac35

• Skip Flaky Test: for pytorch#134602 (pytorch#137226)

  Show description for 5c0ce8d
  

  drisspg

  authored and

  pytorchmergebot

  committed
  5c0ce8d

  Copy full SHA for 5c0ce8d

• [inductor] Reduce block sizes when using Triton CPU backend (pytorch#136612)

  Show description for b3953ff
  

  int3

  authored and

  pytorchmergebot

  committed
  b3953ff

  Copy full SHA for b3953ff

• [Inductor] Use parametrize to break down some unit tests (pytorch#137156)

  Show description for 4513fb5
  

  desertfire

  authored and

  pytorchmergebot

  committed
  4513fb5

  Copy full SHA for 4513fb5

• [c10d] Fix the device query story of ProcessGroup (pytorch#136790)

  Show description for 7631a04
  

  kwen2501

  authored and

  pytorchmergebot

  committed
  7631a04

  Copy full SHA for 7631a04

• unflatten with specialized graphs per submodule call (pytorch#137013)

  Show description for cd5d1fe
  

  avikchaudhuri

  authored and

  pytorchmergebot

  committed
  cd5d1fe

  Copy full SHA for cd5d1fe

• Fix dependency on filesystem on Linux (pytorch#137209)

  Show description for 6241006
  

  atalman

  authored and

  pytorchmergebot

  committed
  6241006

  Copy full SHA for 6241006

Commits on Oct 2, 2024

• [CI] upload_metrics function to upload to s3 instead of dynamo (pytorch#136799)

  Show description for 235f7e0
  

  clee2000

  authored and

  pytorchmergebot

  committed
  235f7e0

  Copy full SHA for 235f7e0

• Revert "[FSDP2] support torch._foreach_copy_(float8) for fully_shard(Float8Linear) (pytorch#135955)"

  Show description for 2c9e194
  pytorchmergebot

  committed
  2c9e194

  Copy full SHA for 2c9e194

• [FlexAttention] Fix max-autotune when captured buffers are View nodes (pytorch#137204)

  Show description for bb03ef7
  

  drisspg

  authored and

  pytorchmergebot

  committed
  bb03ef7

  Copy full SHA for bb03ef7

• [Profiler] Update Kineto Submodule (pytorch#137137)

  Show description for 759cd73
  

  sraikund16

  authored and

  pytorchmergebot

  committed
  759cd73

  Copy full SHA for 759cd73

• [inductor] Fix build_paths usage in config.py (pytorch#137187)

  Show description for e9e5d76
  

  danzimm

  authored and

  pytorchmergebot

  committed
  e9e5d76

  Copy full SHA for e9e5d76

• Fix NJT serialization (pytorch#137031)

  Show description for e95b230
  

  jbschlosser

  authored and

  pytorchmergebot

  committed
  e95b230

  Copy full SHA for e95b230

• [SDPA] Bump grad_query fudge factor for Flash Attention (pytorch#135711)

  Show description for be423a8
  

  eqy

  authored and

  pytorchmergebot

  committed
  be423a8

  Copy full SHA for be423a8

Pagination

Previous

Next