strisunshinewentingwang/pytorch

Pull Requests Commits

ROCm support for test_lazy_init (#84333)

jataylo committed 2 years ago

09bcc006

Add synchronize hooks (#84427)

sypneiwski committed 2 years ago

67d6f716

Redo how custom/python_custom methods on TensorImpl work (#84641)

ezyang committed 2 years ago

591b75bf

Add config to PrimTorch's nvFuser executor (#84482)

IvanYashchuk committed 2 years ago

d802fcfc

test mkldnn conv2d channels last when weight is nchw format (#77348)

mingfeima committed 2 years ago

6f72c13f

[FSDP] Ensure that all ranks use the same order to iterate through optimizer states (#84654)

fegin committed 2 years ago

1840f24d

Moving CommTensor from tests to private _spmd folder (#84719)

mrshenli committed 2 years ago

22119495

[torch.fx.wrap] Use callable / function.__name__ instead of function.__code__.co_name (#84373)

kunalb committed 2 years ago

b00a4b7c

[vision hash update] update the pinned vision hash (#84731)

pytorchmergebot committed 2 years ago

1c8cb087

[torchdynamo hash update] update the pinned torchdynamo hash (#84730)

pytorchmergebot committed 2 years ago

1c8f02d4

Return x.alias() when transpose is an nop (#84674)

SherlockNoMad committed 2 years ago

e6ee8e61

[ONNX] Fix node attributes when namespace is aten (#84211)

justinchuby committed 2 years ago

dbdc1cd5

[ONNX] Rename constants for clarity (#84645)

justinchuby committed 2 years ago

2fa8142c

[quant] remove mkldnn headers in OnednnUtils.h (#84195)

xndcn committed 2 years ago

bc3683de

Disable Transformer/MHA fast path when autocast is enabled (#84722)

erichan1 committed 2 years ago

7a5d5a00

Fix conda cmake setup for macos x86-64 (#84682)

huydhn committed 2 years ago

f23a1cf8

Add __all__ for a few distributed modules plus a little typing (#84119)

Rodrigo Kumpera committed 2 years ago

6f216805

[functorch] excise older custom_vjp prototype (#84638)

zou3519 committed 2 years ago

9b8e0a38

[functorch] Add some C++ documentation (#84637)

zou3519 committed 2 years ago

8c91dd26

Revert "Add mkl implementation for exponential on CPU (#69967)"

pytorchmergebot committed 2 years ago

05b778f9

[aarch64] Fix aarch64 build so that quantize_val_arm is defined (#84564)

psaab committed 2 years ago

6bedb7a7

Revert "Moving CommTensor from tests to private _spmd folder (#84655)"

pytorchmergebot committed 2 years ago

a6e6276c

fix [rpc] Wrong usage of RRefContext::handleException #71458 (#83166)

xiaxiaohua1 committed 2 years ago

c5cf8f6b

Added mv, mm, and binary_cross_entropy_with_logits decomps (#84451)

Chillee committed 2 years ago

1459a909

Enable manual test config label selection on Windows (#84669)

huydhn committed 2 years ago

fe353e14

Moving CommTensor from tests to private _spmd folder (#84655)

mrshenli committed 2 years ago

07dad155

Reland: [Profiler] Unify global and thread local profiler lookup. (#83894) (#84668)

Taylor Robie committed 2 years ago

9beb0c0b

Reland: [Profiler][Trivial] Create orchestration folder and move observer management there. (#83893)" (#84667)

Taylor Robie committed 2 years ago

bea01840

Fix linalg.norm sample inputs function and related failures (#84452)

peterbell10 committed 2 years ago

b9793a66

asyncio increase throughput (pytorch change) (#84301)

Junteng Jia committed 2 years ago

335033f7

Newer Older