strisunshinewentingwang/pytorch

Pull Requests Commits

per channel quantization support (#25134)

dskhudia committed 5 years ago

d8062575

hyperparameter plugin (#23134)

lanpa committed 5 years ago

cd14518e

Add Python/C++ torch.nn API parity test harness (#23852)

Will Feng committed 5 years ago

1bf1970f

prevent generating caffe2::mkl for multiple times (#25167)

jasjuang committed 5 years ago

573b1cd2

Ensure tests get passed on Windows (#25145)

peterjc123 committed 5 years ago

c24314bf

torch.from_numpy fix for np.int (#25139)

pbelevich committed 5 years ago

30bc6527

Support focal loss in MTML

Yu Shi committed 5 years ago

43a2fd0e

Fix ios_crash:backtrace=FBCameraFramework:caffe2::getClockTimeMilliseconds() (perf_observer.cc (#24813)

geof90 committed 5 years ago

b7b80c6b

Add logging to JIT CSE pass.

Mikhail Zolotukhin committed 5 years ago

f2bcad5d

Switch hub to use `requests` because of SSL (#25083)

James Reed committed 5 years ago

f71ddd42

SubgraphMatcher: matching modules support. (#25075)

Mikhail Zolotukhin committed 5 years ago

85bca16a

SubgraphMatcher: add logging.

Mikhail Zolotukhin committed 5 years ago

16289c2f

SubgraphMatcher: Factor out matchAttributes.

Mikhail Zolotukhin committed 5 years ago

b5096b68

use avx2 for Add without broadcast and when inputs are uint8_t (#25098)

jspark1105 committed 5 years ago

a54f8f0f

Use the EmbeddingLookup API which takes the offsets instead of lengths (#24945)

jianyuh committed 5 years ago

363655dc

print padding_mode for Conv modules if not zeros (#23996)

Ailing Zhang committed 5 years ago

35a00155

Support lowering of fp16 weights

Yinghai Lu committed 5 years ago

add57fd2

Revert "per channel quantization support (#24936)" (#25131)

zrphercule committed 5 years ago

bc83ed10

move some methods into function.cpp (#25119)

suo committed 5 years ago

98544355

Add a skip_override option to should_run_job.py (#25118)

suo committed 5 years ago

65beee58

add some sparse tensor ops support in TorchScript (#24967)

wanchaol committed 5 years ago

5fd3251c

Add missing functions and methods for channelwise quantization (#24934)

dskhudia committed 5 years ago

12ea1d74

per channel quantization support (#24936)

dskhudia committed 5 years ago

9e996503

Per Channel quantization APIs (#24935)

dskhudia committed 5 years ago

aba15ce9

Make EmbeddingLookup APIs take offsets rather than lengths to match the PyTorch's EmbeddingBag (#24944)

jianyuh committed 5 years ago

ad7250d3

Update QNNPACK submodule to 901e9d4 (#25044)

dreiss committed 5 years ago

6981b4e5

Add helper function Tensor::names() (#24914)

zou3519 committed 5 years ago

c013c066

Rename Tensor::names() to Tensor::opt_names() (#24907)

zou3519 committed 5 years ago

530db2c7

Fix `FIXME_default_names` by storing static list of 64 none names (#24885)

zou3519 committed 5 years ago

867d8af2

Optimize performance for unboxed-only kernels (#25055)

smessmer committed 5 years ago

6c834246

Newer Older