strisunshinewentingwang/pytorch

Pull Requests Commits

Revert "Reland "Add torch.utils.device_mode" (#91796)"

pytorchmergebot committed 1 year ago

9b415240

Fix dynamo handling for tensor attributes: T, H, mT, mH (#90463)

anjali411 committed 1 year ago

9945a78a

fix sort crash when the input is expanded scalar (#91752)

mingfeima committed 1 year ago

3643b4ee

fix norrow_copy correctness issue for non-contiguous input for cpu path (#91789)

XiaobingSuper committed 1 year ago

136dadd6

Apply clang-tidy fixes to api/csrc/api/include/torch/nn (#91766)

Skylion007 committed 1 year ago

8cec433c

Symintify pytorch slicing logic (#91340)

tugsbayasgalan committed 1 year ago

f59845db

Reland "Add torch.utils.device_mode" (#91796)

ezyang committed 1 year ago

81b5eff3

Torch package support in dynamo (#91821)

voznesenskym committed 1 year ago

eeb3e49e

Apply clang-tidy perf fixes to aten (#91772)

Skylion007 committed 1 year ago

73e5379f

remove unnecessary decomps (#91828)

Natalia Gimelshein committed 1 year ago

2c000641

[ONNX] Add aten::zero support (#91731)

titaiwangms committed 1 year ago

e3ed55d4

Dynamo benchmark: add CPU specific changes (#88477)

blzheng committed 1 year ago

0c1777ac

Assert valid base source for derivative sources (#91711)

wconstab committed 1 year ago

75c65282

[primTorch] Use torch.fill to implement prims.fill (#91747)

peterbell10 committed 1 year ago

edaba335

[CI][ROCm] prune all stopped containers (#91815)

jeffdaily committed 1 year ago

faed4db4

Make torch.split take symint as arg (#91724)

tugsbayasgalan committed 1 year ago

b32b81a0

Revert "[ONNX] Add aten::zero support (#91731)"

pytorchmergebot committed 1 year ago

08a378a2

Un fold squeeze permute (#91656)

MDK8888 committed 1 year ago

a2c5efaf

[PT-D][3/N] Add FSDP hook with Named Optimizer (#91321)

fduwjj committed 1 year ago

5fabd96f

[ROCm] fix hipify mapping for cuDeviceGet (#90726)

jeffdaily committed 1 year ago

acab0edf

[MPS] Add support for randperm (#91708)

DenisVieriu97 committed 1 year ago

53ef96fa

[ONNX] Add aten::zero support (#91731)

titaiwangms committed 1 year ago

ff23508c

Remove hard numpy dependency introduced by inductor (#90796)

amjames committed 1 year ago

766ebf44

Properly guard all numpy usage within dynamo and remove UnspecializedNumpyVariable (#90795)

amjames committed 1 year ago

7cd951c2

[CI][ROCm] fix device visibility, again (#91813)

jeffdaily committed 1 year ago

f4494628

[JIT] Skip builtins while enumerating class methods (#91805)

malfet committed 1 year ago

4f1f14e3

Automatically convert real tensors to fake in dynamo export (#91742)

tugsbayasgalan committed 1 year ago

69acc340

make sure mutated args are iterated in the same order (#91792)

Natalia Gimelshein committed 1 year ago

ef495b7d

Revert "Deduplicate c10 error and PyTorchError hierarchy (#87855)"

pytorchmergebot committed 1 year ago

b3603f81

Return empty attention weights when need_atten_weights = False (#91782)

drisspg committed 1 year ago

f2199709

Newer Older