strisunshinewentingwang/pytorch

Pull Requests Commits

Improve BatchNorm1d training performance (CPU) (#57033)

peterbell10 committed 3 years ago

46a32e07

Add cross OpInfo (#55483)

Mike Ruberry committed 3 years ago

4a872f85

add support for complex input to `torch.testing.assert_(equal|close)` (#57162)

pmeier committed 3 years ago

5c68072e

Skip linalg.qr saved mode check if compiled without LAPACK (#56284)

IvanYashchuk committed 3 years ago

eaf00bf7

Port reverse binary ops to `OpInfo` (#56471)

ysiraichi committed 3 years ago

ce444991

[DDP] Uneven inputs: option to throw early (#56755)

rohan-varma committed 3 years ago

57f72b84

Torchelastic: add multiprocessing tests to ci/cd (#56842)

aivanou committed 3 years ago

7fe4c1d0

ns for fx: add missing add_relu and mul_relu patterns (#56927)

vkuzo committed 3 years ago

bb640efa

s/InplaceOrView/ADInplaceOrView/g (#57372)

Ailing Zhang committed 3 years ago

0ecdbfeb

OpInfo: mvlgamma (#56907)

kshitij12345 committed 3 years ago

41099ef7

Revert D27487549: [TensorExpr] Add `CodeGen::call_raw` method.

Mike Ruberry committed 3 years ago

05b255c5

Add torch.linalg.cholesky_ex without checking for errors by default (#56724)

IvanYashchuk committed 3 years ago

75a2a92b

[NNC] Add logical Operators '&&' and '||' (#56947)

huiguoo committed 3 years ago

afe6b4c8

Fix torch.ormqr for non Fortran-contiguous inputs (#57314)

IvanYashchuk committed 3 years ago

2be11533

grad_fn getter for optional strings (#55225)

peterbell10 committed 3 years ago

7c8d0069

Sparse support for division rounding_mode argument (#51989)

peterbell10 committed 3 years ago

a5288a02

ROCM: Re-enable test_norm_fro_2_equivalence_old (#57170)

arindamroy-eng committed 3 years ago

6d681d06

Immediately mark DLPack capsule as used after stealing the ownership (#56789)

kmaehashi committed 3 years ago

4350d4af

Revert D28110359: [TensorExpr] Add `TensorExprKernel::runFast` method.

Mike Ruberry committed 3 years ago

30180930

Inline hooks in ivalue::Future (#57354)

lw committed 3 years ago

82d245fa

Use Devices instead of DeviceIndexes in Future (#57353)

lw committed 3 years ago

fb7469fb

Use Devices instead of DeviceIndexes in TensorPipe agent (#57294)

lw committed 3 years ago

0422e673

Rename DeviceGuardImplInteface's getStreamFromPool method (#57345)

lw committed 3 years ago

0c3e79b5

Add device() method to c10::Event (#57293)

lw committed 3 years ago

6697ef51

Add pybind type caster for c10::Device (#57292)

lw committed 3 years ago

58bc0034

Fix CUDA Stream synchronization when arguments contains RRefs (#57394)

mrshenli committed 3 years ago

2dffa8cd

Fix variable names in torch.fft examples (#57290)

peterbell10 committed 3 years ago

d536e6c6

Revert D28110358: [StaticRuntime] Use NNC's call_raw API to reduce call overheads.

Mike Ruberry committed 3 years ago

3315f142

Support auto generation of device check (#56872)

wenleix committed 3 years ago

20085f6d

Disable device check for foreach kernels (#56871)

wenleix committed 3 years ago

22ecb888

Newer Older