strisunshinewentingwang/pytorch

Pull Requests Commits

DO NOT MERGE, testing manywheel build consolidation

seemethere committed 2 years ago

618c0f52

[PyTorch][Distributed] Enable Reduce Scatter and modify all_to_all for sharded linear with more test cases. (#68786)

fduwjj committed 2 years ago

7c2489bd

[Autograd/Checkpoint] Checkpoint implementation without reentrant autograd (#69027)

rohan-varma committed 2 years ago

e032dae3

add VSX dispatch for fft_fill_with_conjugate_symmetry_stub (#68914)

cdeepali committed 2 years ago

4d81175a

.github: Volume mount local netrc for docs push (#69472)

seemethere committed 2 years ago

f87faf3c

[FSDP] Enforce wrapper_cls as a mandatory kwarg in enable_wrap. (#69358)

rohan-varma committed 2 years ago

1859e5f0

[FSDP] Kill config_auto_wrap_policy, remove policy from enable_wrap, (#69357)

rohan-varma committed 2 years ago

00245fed

[FSDP] Remove auto_wrap() (#69356)

rohan-varma committed 2 years ago

c95277e9

[FSDP] Make recursive_wrap, wrap APIs independent of ConfigAutoWrap. (#68776)

rohan-varma committed 2 years ago

f333cde1

FX pass: fuse_sparse_matmul_add (#69340)

Marat Subkhankulov committed 2 years ago

456139d0

[Vulkan] Implement slice operator (#69382)

beback4u committed 2 years ago

68b5c86e

unify compare kernels (#69111)

ngimel committed 2 years ago

a84ed8be

Clean up CODEOWNERS for .github/ (#69395)

seemethere committed 2 years ago

38c576cf

Move THC_sleep to ATen (#69038)

peterbell10 committed 2 years ago

bf01cd52

Skips failing ROCm test (#69456)

Mike Ruberry committed 2 years ago

a9746996

expose return_types in Python (#66614)

kshitij12345 committed 2 years ago

b737e09f

Enable native_dropout/backward for lazy (#69374)

wconstab committed 2 years ago

78b7a419

The Jiterator (#69439)

Mike Ruberry committed 2 years ago

b6f41bb8

[Core ML] Avoid recompiling models when the OS version is not changed (#69438)

xta0 committed 2 years ago

3202028e

Revert D32780415: [Static Runtime] Move implementation details from impl.h into internal.h

d1jang committed 2 years ago

c97dc928

Revert D32743881: [Core ML] Avoid recompiling models when the OS version is not changed

suo committed 2 years ago

29a45f00

[Static Runtime] Move implementation details from impl.h into internal.h (#69274)

d1jang committed 2 years ago

999e93e6

[Core ML] Avoid recompiling models when the OS version is not changed (#69234)

xta0 committed 2 years ago

b97903ab

[LT] Upstream LazyView and view ops IR Nodes (#69277)

desertfire committed 2 years ago

e8f4c9cc

[LT] Upstream more util functions (#69098)

desertfire committed 2 years ago

0bbe21b1

[Linalg] Add a runtime switch to let pytorch prefer a backend impl in linalg functions on GPU (#67980)

xwang233 committed 2 years ago

bfe5ad28

[Static Runtime] Fix a bug that aten::embedding_bag keeps cannot handle resized input tensors (#69219)

d1jang committed 2 years ago

9663e086

Add OpInfos for misc nn.functional operators (#68922)

saketh-are committed 2 years ago

6a4fa860

[CUDA graphs] Fixes make_graphed_callables example typos (#69379)

mcarilli committed 2 years ago

da023611

Update CUDA version to 11.3 and setup proper environment variables. (#69383)

xuzhao9 committed 2 years ago

e92b14bf

Older