strisunshinewentingwang/pytorch

Pull Requests Commits

CUDA BFloat16 batchnorm (non-cuDNN) (#44994)

zasdfgbnm committed 3 years ago

59605811

[StaticRuntime] Add aten::narrow (#48991)

Hao Lu committed 3 years ago

e8ec8486

Put Flake8 requirements into their own file (#49032)

samestep committed 3 years ago

d1fb4b4f

[TensorExpr] Enable inlining for output tensors too. (#48967)

Mikhail Zolotukhin committed 3 years ago

2b70bcd0

Delete ATen mirror stuff (#49028)

samestep committed 3 years ago

0fb9d366

Add LKJCholesky distribution (#48798)

neerajprad committed 3 years ago

dee82ef3

[FX][2/2] Make docstrings pretty when rendered (#48871)

James Reed committed 3 years ago

c92c8598

Add fftw3 cmake as alternative for FFT/DFT (#48808)

Rong Rong committed 3 years ago

b89c3284

Avoid initializing gradInput twice in the backward phase of replication (#48890)

Jokeren committed 3 years ago

b0e919cf

[static runtime] Add Internal Ops to the registry (#48616)

bwasti committed 3 years ago

274ce26f

[BE] Fix signed-unsigned warnings (#48848)

malfet committed 3 years ago

ad3fed8b

Modify NEON check for ARM64 on OS X (#48982)

malfet committed 3 years ago

c29f5164

Back out "Revert D25375885: [pytorch][PR] Reenable some BF16 tests on CUDA"

Rong Rong committed 3 years ago

58c13cf6

minor README change to fix #25464 (#48970)

janeyx99 committed 3 years ago

e2befb84

Revert D25375885: [pytorch][PR] Reenable some BF16 tests on CUDA

Rong Rong committed 3 years ago

39445f71

[static runtime] fuse inference ops (1) (#48948)

ajyu committed 3 years ago

07978bd6

VariableType calls faithful C++ API for c10-full out ops (#47792)

smessmer committed 3 years ago

b643dbb8

Faithful out arguments (#47712)

smessmer committed 3 years ago

3ef36dca

Enable faithful API for all ops (#47711)

smessmer committed 3 years ago

046ea669

Add and adjust kernel launch checks (#46727)

r-barnes committed 3 years ago

32b098ba

Fix some convoluted(?) code (#48893)

r-barnes committed 3 years ago

cb6233aa

Move aten::__contains__.int_list for lite jit (#48950)

jiatongzhou committed 3 years ago

c3a90bed

docker: Add make variable to add docker build args (#48942)

seemethere committed 3 years ago

881e9583

CUDA BF16 backwards (#48809)

zasdfgbnm committed 3 years ago

5533be51

[DOCS]Correct docs for torch.lu_solve (#47762)

bharatr21 committed 3 years ago

3aeb9cc8

Added entry for torch.linalg.cond to linalg.rst (#48941)

IvanYashchuk committed 3 years ago

bea88ee1

[Gradient Compression] Let the dtype of created low-rank tensors P and Q be the same type as the input tensor (#48902)

Yi Wang committed 3 years ago

c876d4f4

Register OpInfos for torch.fft transforms (#48427)

peterbell10 committed 3 years ago

533c8378

[package] pre-emptively install submodules (#48799)

zdevito committed 3 years ago

adbb74de

Reenable some BF16 tests on CUDA (#48805)

zasdfgbnm committed 3 years ago

e3893b86

Newer Older