strisunshinewentingwang/pytorch

Pull Requests Commits

[JIT] Disable conv-add-relu fusion for cuDNN7 when model uses fp16 (#56579)

desertfire committed 3 years ago

c9714016

refactor autocast to be extensible for devices (#57104)

leslie-fang-intel committed 3 years ago

731cc472

Add supported backward_dtype to OpInfo (#56156)

Rong Rong (AI Infra) committed 3 years ago

095c328d

Revert D27582224: [pytorch][PR] Automated submodule update: FBGEMM

malfet committed 3 years ago

e08303c7

Revert D28114231: [pytorch][PR] Automated submodule update: FBGEMM

malfet committed 3 years ago

0dddfbf3

Remove unused forward AD flag (#57058)

albanD committed 3 years ago

95dc2b6e

Improve perf for forward AD view handling (#57057)

albanD committed 3 years ago

83f18671

fix InplaceOrView implementation for manual functions (#57152)

albanD committed 3 years ago

b016bc1c

Fix use of allow_tensor_metadata in view variable creation (#57069)

albanD committed 3 years ago

c91bd25e

make external codegen aware of autogen'd composite kernels (#56960)

bdhirsh committed 3 years ago

6fa1d880

[special] Add `xlog1py` (#55138)

kshitij12345 committed 3 years ago

d4ddb477

Expand Kineto profiler support (part 1) (#57333)

Ilia Cherniavskii committed 3 years ago

5b3e7638

quote str kwarg values in `test_ops.py::TestCommon::test_jit_alias_remapping` (#57120)

pmeier committed 3 years ago

db32b695

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

CodemodService FBSourceClangFormatLinterBot committed 3 years ago

df69b0d0

Automated submodule update: FBGEMM (#57342)

facebook-github-bot committed 3 years ago

264db195

Automated submodule update: FBGEMM (#55347)

facebook-github-bot committed 3 years ago

54469e15

Static runtime support for fb::expand_dims (#57282)

Edvard Ghazaryan committed 3 years ago

b3e18024

[torch/deploy] opt torch/csrc/depoy into autofromatting

suo committed 3 years ago

e31b67f5

Fix a numerical issue of CUDA channels-last SyncBatchNorm (#57077)

xwang233 committed 3 years ago

ac72881f

Ignore more compiler warnings, unify WERROR options (#56630)

davidriazati@fb.com committed 3 years ago

c44cbc63

Revert "Remove sync for randperm on small tensors. (#54113)" (#57299)

ngimel committed 3 years ago

65968ab8

[kineto] Deprecate ClientTraceActivity and merge it with GenericTraceActivity (#56743)

chaekit committed 3 years ago

49dbe179

port neg to structure kernel (#57212)

Rong Rong (AI Infra) committed 3 years ago

16fc18bf

Fix sort for slow gradcheck (#57192)

soulitzer committed 3 years ago

99516120

[torch] Add backward support for segment reduce (CPU only)

Serhat Yilmaz committed 3 years ago

e27740b3

[torch/debuggability] use log.info() in addition to print() in timeoutguard (#57296)

Aapo Kyrola committed 3 years ago

d1def931

[RPC Framework] Expose a Python API for device map getter (#57179)

Yi Wang committed 3 years ago

c2fbd967

[package] PackageExporter `__import__` logic to not parse dynamic cases (#57283)

Lilyjjo committed 3 years ago

2c6f5e8a

Added OpInfos for sub & mul (#56227)

imaginary-person committed 3 years ago

6ed90ed1

Update compare_set docs (#57203)

H-Huang committed 3 years ago

149000c3

Newer Older