strisunshinewentingwang/pytorch

Pull Requests Commits

Merge branch 'master' into ci-all/peterbell10/batch-norm-cuda

ngimel committed 3 years ago

Verified 60e4ba69

Remove _save_data() and _load_data() from mobile (#57879)

iseeyuan committed 3 years ago

737f48df

Add EMA to DecayAdagrad (#57866)

Zhijing Li committed 3 years ago

88a1e8eb

[torch] handle embedding bag with empty bag (#57446)

jspark1105 committed 3 years ago

a46e927b

[TensorExpr] Fix UB in LoopNest::distribute. (#57883)

Mikhail Zolotukhin committed 3 years ago

f51798d0

[profiler][kineto] Support for memory allocs/deallocs in the traces (#57835)

Ilia Cherniavskii committed 3 years ago

8639fd10

[clang-tidy] Exclude cppcoreguidelines-avoid-magic-numbers (#57841)

malfet committed 3 years ago

3a66a1cb

benchmark rpc ps (#57454)

gcramer23 committed 3 years ago

bc2540f0

[RPC Framework] Update rpc.rst (#57876)

Yi Wang committed 3 years ago

94080f45

[RPC Framework] Add a link to the tutorial in RemoteModule docstring (#57875)

Yi Wang committed 3 years ago

4db88307

[RPC Framework] Support passing RemoteModule as an arg (#57695)

Yi Wang committed 3 years ago

74d493cc

Restore var_mean_out

peterbell10 committed 3 years ago

0fffd927

[PyTorch Edge] Add backport to export old bytecode models (#56802)

cccclai committed 3 years ago

8c04593c

Fix flaky test_barrier_timeout_global. (#57523)

pritamdamania committed 3 years ago

e9c3ce30

[fx ir] Handle cases in GraphDrawer when shape, type or stride are not present (#57845)

khabinov committed 3 years ago

73f22bcb

Remove __has_attribute for C10_FALLTHROUGH

peterbell10 committed 3 years ago

05ddd6ad

Improve BatchNorm1d performance (CUDA)

peterbell10 committed 3 years ago

46087cd2

Fix lint in test_tensorexpr_pybind (#57869)

davidriazati@fb.com committed 3 years ago

ee4be532

Update the default detach semantic for forward mode AD (#57820)

albanD committed 3 years ago

4fad8d1a

[Vulkan] Use more optimal command buffer submission rate (#57196)

SS-JIA committed 3 years ago

bc0965ac

Enable backward/forward compatibility for TS runtime (#57498)

tugsbayasgalan committed 3 years ago

b0c27b44

[nnc] Added NNC lowerings for t/transpose/permute/expand + other cleaning (#57426)

Chillee committed 3 years ago

b38f153d

Respect .ini for flake8 and mypy (#57752)

davidriazati@fb.com committed 3 years ago

c88167d2

Change name for namedtuple return of torch.linalg.svd (#57181)

IvanYashchuk committed 3 years ago

18fed3df

Remove compute_uv flag from torch.linalg.svd (#57180)

IvanYashchuk committed 3 years ago

58f32fa5

Avoid 2 extra copies when reducing sparse tensors and fix result() vs inplace output discrepancy (#57822)

agolynski committed 3 years ago

db412a68

Add correction parameter to std/var (#50903)

peterbell10 committed 3 years ago

20430932

[PyTorch] Remove dead get/setTLSCallbacks APIs (#56492)

swolchok committed 3 years ago

3d2ce605

[PyTorch] Use MaybeOwned and avoid resize in bmm_cuda (#56115)

swolchok committed 3 years ago

9234d7fc

Add Gloo TCP_TLS transport (#56442)

pbelevich committed 3 years ago

96e1a83f

Older