strisunshinewentingwang/pytorch

Pull Requests Commits

Add support for max_pool2d_with_indices +backward (#69729)

wconstab committed 2 years ago

Verified 9a87ac8e

Add sync after each overhead computation step to try to flush GPU ops; should also try fixing CPU/GPU clocks. not sure this helps..

wconstab committed 2 years ago

e6abf446

Add codegen support for avg_pool2d + backward (#69739)

wconstab committed 2 years ago

Verified 98d2dbc1

Adds a `maximize` flag to Adam (#68164)

oliver committed 2 years ago

3d358a76

Hook up general convolution to convolution_backward (#69584)

jbschlosser committed 2 years ago

fc37e5b3

[SR] Log SR options (#69809)

Hao Lu committed 2 years ago

0420de35

General convolution_backward function (#69044)

jbschlosser committed 2 years ago

f0e98dcb

Fix typo in aten::full in version_map (#69807)

tugsbayasgalan committed 2 years ago

a5b5152d

Forward AD for inplace comparison operators (#69597)

soulitzer committed 2 years ago

af7ee9fc

Add some forward AD formulas (#69384)

soulitzer committed 2 years ago

0dcbd73e

Fix copy_ forward AD to handle broadcasting (#69592)

soulitzer committed 2 years ago

baf92f9d

Do not test batched forward grad for inplace ops (#69558)

soulitzer committed 2 years ago

db32daf4

Revert D32606547: torch/monitor: add C++ events and handlers

suo committed 2 years ago

f565167f

[quant][fx][graphmode] Move more patterns to use ModuleReLU fuse handler (#69644)

jerryzh168 committed 2 years ago

f5751799

torch/monitor: add C++ events and handlers (#68783)

d4l3k committed 2 years ago

e61fc1c0

Populate runtime with upgrader graph (#68773)

tugsbayasgalan committed 2 years ago

20f7c893

Back out "[pytorch][PR] Add ability for a mobile::Module to save as flatbuffer" (#69796)

gmagogsfm committed 2 years ago

17f3179d

clear predict_net field from PredictorExporterMeta stored in the exporter to save memory (#68485)

smacke committed 2 years ago

3906f824

[PyTorch][kineto] Remove heap-allocated vectors in saveExtraArgs (#69737)

swolchok committed 2 years ago

19fecc63

Fix the TorchBench CI when running with a benchmark branch. (#69795)

xuzhao9 committed 2 years ago

731c8255

Make c10 tests compilable with -Werror (#69711)

malfet committed 2 years ago

59deee83

Suppress common warnings when building by clang (#69710)

malfet committed 2 years ago

e305e4d4

Revert D32739976: fix CompositeImplicitAutograd ops improperly labeled

malfet committed 2 years ago

41c344d4

Fix docker builds for Python-3.6 (#69785)

malfet committed 2 years ago

77213fa4

[DataPipe] Adding interface for MapDataPipes (#69648)

NivekT committed 2 years ago

a5a7e309

[DataPipe] Adding output types to DataPipe interface file (#69647)

NivekT committed 2 years ago

81a60b98

[PyTorch] Update SmallVector from LLVM (#69110)

swolchok committed 2 years ago

d026057b

Ensure different output filenames for test/train results (#69784)

wconstab committed 2 years ago

Verified 7ec7a37f

[PyTorch] Simple refcount bump fixes in standardizeVectorForUnion & callees (#66695)

swolchok committed 2 years ago

1d269e8c

Move internal ops under ts_backend

desertfire committed 2 years ago

f3229d64

Newer Older