strisunshinewentingwang/pytorch

Pull Requests Commits

switch back to azure pipelines

suo committed 4 years ago

6de1016f

Updating submodules

svcscm committed 4 years ago

73a926fd

Add option to clean up allocated activations between c2 runs (#29619)

Yinghai Lu committed 4 years ago

f0dd7517

Allow unrelated histories when rebasing to master (#29699)

houseroad committed 4 years ago

03d021dd

Revert D18451046: CPU-Strided-Complex Fixes for real and imag ops

ezyang committed 4 years ago

5635a720

Missing host device (#29547)

iotamudelta committed 4 years ago

6d54c5dd

CPU-Strided-Complex Fixes for real and imag ops (#29607)

dylanbespalko committed 4 years ago

9b1ff809

Delete all trivial uses of make_variable. (#29213)

ezyang committed 4 years ago

0c91ebb6

Miscellaneous follow up for code review comments (#29204)

ezyang committed 4 years ago

89e187a2

Rename getNonVariableDeprecatedTypeProperties to getDeprecatedTypeProperties (#29203)

ezyang committed 4 years ago

30092df1

Revert D18455666: Gradle build with offline dependencies

ezyang committed 4 years ago

7da9ac5a

Revert D18458751: use new overload mechanism for rnns

ezyang committed 4 years ago

715e951e

More checks on MSVC (#29709)

peterjc123 committed 4 years ago

e870a9a8

Switch XLA to only override abstract functions (#29636)

Ailing Zhang committed 4 years ago

7b86199f

Restructure comparison ops so as to better support XLA dispatch (#29591)

xuhdev committed 4 years ago

3a72662d

Changed default args in quantization observers

z-a-f committed 4 years ago

09d359df

observer benchmarking

z-a-f committed 4 years ago

d2aa4c61

Gradle build with offline dependencies (#29262)

IvanKobzarev committed 4 years ago

d8732b3b

PackedSequence support for quantized LSTM

James Reed committed 4 years ago

20fb8a81

Revert D18466043: Pin Linux image and modules version to 4.4.0-166

Will Feng committed 4 years ago

87363a81

Do not show cuda stats in autograd profiler when `use_cuda=False` (#29666)

bddppq committed 4 years ago

5a8ad663

Turn on named tensors for all builds (#29603)

zou3519 committed 4 years ago

95cad573

Pin Linux image and modules version to 4.4.0-166 (#29690)

Will Feng committed 4 years ago

907a29de

Fix a missing comma in quantized benchmark

z-a-f committed 4 years ago

29e509ff

Make dropout condition on training.

Ailing Zhang committed 4 years ago

8875120b

Fix some issues for lite interpreter internal build. (#29620)

xcheng16 committed 4 years ago

422fbfb1

Add op bitwise_xor to replace __xor__ and __ixor__ (#25665)

xuhdev committed 4 years ago

bd0394d4

use new overload mechanism for rnns (#29614)

Elias Ellison committed 4 years ago

8e7b4067

Change arg dtype from float to double in LPPool and nn/utils/clip_grad.h (#29584)

Will Feng committed 4 years ago

433baf1b

Use c10::variant-based enums for SmoothL1Loss module and functional

Will Feng committed 4 years ago

65bfcde0

Newer Older