strisunshinewentingwang/pytorch

Pull Requests Commits

Update on "[WIP] enable static graph training in DDP"

zhaojuanmao committed 3 years ago

4c8183f5

Update base for Update on "[WIP] enable static graph training in DDP"

zhaojuanmao committed 3 years ago

ef7cfbe9

[iOS GPU][Perf][2/n] Prepack linear + Fuse relu/hardtanh (#57665)

xta0 committed 3 years ago

36172f34

Disable test for quicklint (#57968)

davidriazati@fb.com committed 3 years ago

f1d01b94

torch.where is now mentioning Bool rather than Byte when given wrong dtype mask (#57942)

venkateshtata committed 3 years ago

f0f69c5d

Fix json parse error for profiler call stack (#57099)

guyang3532 committed 3 years ago

ebb1b74f

add cuda memory and distributed metadata (#57252)

guotuofeng committed 3 years ago

98fcdb80

Fix typo in warning for spawn method (#57927)

kaushikb11 committed 3 years ago

ba07aaf2

Update on "[WIP] enable static graph training in DDP"

zhaojuanmao committed 3 years ago

338eb592

Update base for Update on "[WIP] enable static graph training in DDP"

zhaojuanmao committed 3 years ago

f925f89b

[vulkan] Add sigmoid activation functions (#57867)

SS-JIA committed 3 years ago

19706d91

Fix creation_meta for multi view outputs in NoGradMode/InferenceMode. (#57842)

Ailing Zhang committed 3 years ago

481806be

[Vulkan] Fix seg fault during descriptor set allocation on some platforms (#57825)

SS-JIA committed 3 years ago

478f6397

Use tools/print_test_stats.py in GHA (#57647)

samestep committed 3 years ago

e1cbc43f

Fix hasattr support type (#57950)

nikithamalgi committed 3 years ago

bf053a12

Ensure torch.save() deterministic output (#57536)

FrancescoCasalegno committed 3 years ago

fea38242

[DDP] fix param to name mapping (#57771)

rohan-varma committed 3 years ago

fe3c63d9

Add Python-3.9 CI testing (#50992)

malfet committed 3 years ago

b587354e

Do not download slow test when on sandcastle (#57953)

Rong Rong (AI Infra) committed 3 years ago

29753339

Remove code and logic for old style custom autograd Function (#57357)

soulitzer committed 3 years ago

710a83d0

Fix document around DDP uneven inputs (#57448)

rohan-varma committed 3 years ago

d115e81a

Port maximum and minimum to structured (#57630)

Freey0 committed 3 years ago

4d181ba5

Remove unnecessary indirection through torch::autograd::impl::pyobj/set_pyobj (#57733)

ezyang committed 3 years ago

727c1d69

[JIT] initial support for PEP-585 types (#57363)

walterddr committed 3 years ago

807bea1c

Add run_master_build workflow (#57899)

malfet committed 3 years ago

bc798cdc

[DataLoader] Change Decoder signature and add MatHandler (#57391)

ejguan committed 3 years ago

ece15f69

[AutoAccept][Codemod][FBSourceClangFormatLinter] Daily `arc lint --take CLANGFORMAT`

CodemodService FBSourceClangFormatLinterBot committed 3 years ago

cbfce376

tweak sync note wording for linalg docs (#57924)

ngimel committed 3 years ago

b84a28b5

Revert D28117714: [pytorch][PR] ATen lu_unpack. Required for making `torch.lu_solve` differentiable.

Mike Ruberry committed 3 years ago

3c87fe9b

[JIT] Adding a concat optimization pass (#55474)

navahgar committed 3 years ago

259d19a7

Older