strisunshinewentingwang/pytorch

Pull Requests Commits

Small fix in SDPA docstring codeblock (#109086)

mikaylagawarecki committed 1 year ago

abd83ce1

[MPS] Adding lgamma, digamma, and polygamma implementations (#106292)

igm503 committed 1 year ago

1b9b3a2d

[inductor] Lower masked_scatter on CUDA (#108803)

peterbell10 committed 1 year ago

c8e577bf

[meta] Add meta implementation for aten.masked_scatter (#108802)

peterbell10 committed 1 year ago

464f9c37

Update HF version to commit hash (6c26faa) (#107400)

angelayi committed 1 year ago

c3945b5f

[export] Lift constant tensors as buffes (reland) (#109040)

angelayi committed 1 year ago

58391aea

Revert "Force synced KJT to trace unbacked SymInt (#108960)"

pytorchmergebot committed 1 year ago

1d32c9c7

[ONNX] bump submodule to onnx==1.14.1 (#108895)

abock committed 1 year ago

8c981c8c

Revert "[ROCm] Add ROCm AMDGPU support for inductor cpp codegen (#105141)"

pytorchmergebot committed 1 year ago

5a7c008b

Don't set requires_grad inside meta function (#108988)

ezyang committed 1 year ago

5531a23b

LazyBatchNorm{1-3}d support dict&set (#109015)

FFFrog committed 1 year ago

bc3f0d34

Revert "Remove fixed skips (#108674)"

pytorchmergebot committed 1 year ago

41bd0fde

[Pytorch][Vulkan] Call binary_op_scalar when 'other' is a 0-dim tensor (#109035)

lucylq committed 1 year ago

65a3d398

Revert "Reland 2: Add PyObject preservation for UntypedStorage (#109039)"

pytorchmergebot committed 1 year ago

59f605be

untracked inputs in constraints (#109037)

avikchaudhuri committed 1 year ago

47be61e1

Force synced KJT to trace unbacked SymInt (#108960)

ezyang committed 1 year ago

f9a250c3

[export] Add a private interface for customizing decomp. (#109058)

zhxchen17 committed 1 year ago

6c8b0dfb

[caffe2] Remove cxx override to c++17 (#108687)

thevinster committed 1 year ago

15202cc8

Prerequisite of ATen/native/utils header for C++ extension (#109013)

guangyey committed 1 year ago

b1f21399

Fix 0-sized views of tensors in cudagraphs (#109055)

eellison committed 1 year ago

85428f5e

Reland 2: Add PyObject preservation for UntypedStorage (#109039)

kurtamohler committed 1 year ago

419e4e17

Revert "[inductor] Parallelize Max Autotune step 1: Use Popen (#107982)"

pytorchmergebot committed 1 year ago

2039f30c

Revert "[inductor] Parallelize Max Autotune step 2: Use all GPUs (#107983)"

pytorchmergebot committed 1 year ago

c36c2bfc

[Reland] increase clang-tidy coverage in torch/csrc (#108875)

cyyever committed 1 year ago

f150f962

[DCP] Enable nD device_mesh resharding DTensor in DCP and add associated tests (#106230)

wz337 committed 1 year ago

b6f9d4db

Re-enable some Windows tests (#108930)

cyyever committed 1 year ago

8025b193

Disable compile for massive data pipe test (#109063)

voznesenskym committed 1 year ago

4691cb26

[Easy] log graphs in compiled_autograd if TORCH_LOGS=compiled_autograd (#108991)

voznesenskym committed 1 year ago

55a204eb

Added limit on number of warps for coordesc autotuner (#108997)

Chillee committed 1 year ago

33c1136f

[quant][be] Rewrite xnnpack_quantizer_utils.py to use decorators (#108920)

jerryzh168 committed 1 year ago

241e84bf

Newer Older