strisunshinewentingwang/pytorch

Pull Requests Commits

use explicitly non-returning GPU atomics (#60607)

jeffdaily committed 3 years ago

d36ce61a

[skip ci] Add GitHub Actions label for g3.16xlarge (#60888)

samestep committed 3 years ago

d62c3ea3

Use expecttest from PyPI (#60658)

samestep committed 3 years ago

d5a44f9f

Fix the NNC-disabled path in static runtime for perf comparisons

bertmaher committed 3 years ago

ddb1f293

[quant][fx][fix] Fused modules with object_type in qconfig (#60779)

angelayi committed 3 years ago

9b94aa53

don't return in __init__ functions (#60830)

cyyever committed 3 years ago

cadce14e

[caffe2/libtorch] Remove already-owned source

andrewjcg committed 3 years ago

9af8aecd

[caffe2] Fix include of corresponding header

andrewjcg committed 3 years ago

eeea696c

[caffe2/utils] Add some fine-grained rules to avoid package boundary violations

andrewjcg committed 3 years ago

c3977bf3

[caffe2/utils] Add explicit rule to avoid package boundary violation (#60677)

andrewjcg committed 3 years ago

03de807d

Enable bionic-cuda10.2-cudnn7-py3.9-gcc7 in GHA (#60204)

samestep committed 3 years ago

41c380e6

Upgrade benchmark to v1.5.5 (#60750)

dreiss committed 3 years ago

971cdafd

[pruning] Speedup activation reconstruction (#60683)

kazhou committed 3 years ago

007ba37c

[pruning] Additional pruning tests (#60681)

kazhou committed 3 years ago

f302e0c7

[pruning] Activation reconstruction (#60292)

kazhou committed 3 years ago

8d4a6ef9

Allow resizing of parametrized tensors (#60418)

kazhou committed 3 years ago

965dad25

[fix] cauchy sampling inf on cuda (#60186)

kshitij12345 committed 3 years ago

956faea5

Use the new URL for docs preview link (#60893)

driazati committed 3 years ago

70e205a2

Enable parallel clang-tidy on ec2 runner (#60870)

1ntEgr8 committed 3 years ago

f5e5ced2

Print stdout and stderr to console on parallel runs (#60869)

1ntEgr8 committed 3 years ago

c8fb7858

docs: add `permute` in torch docs (#60821)

ydcjeff committed 3 years ago

a8057e7e

[vulkan] Implement tanh activation function (#60695)

hanhsienhuang committed 3 years ago

d7c58e5a

[quant] Input-Weight Equalization - tests (#60378)

angelayi committed 3 years ago

da70dd19

[quant] Input-Weight Equalization - support for connected F.linear layer (#60272)

angelayi committed 3 years ago

dfb9c0ba

[quant] Input-Weight Equalization - support for connected linear layers (#60034)

angelayi committed 3 years ago

ddf2ce03

[quant] Input-Weight Equalization - support for F.linear layers (#59964)

angelayi committed 3 years ago

79173189

support non-contiguous tensor in bilinear (#38409)

joerg-de committed 3 years ago

387289d4

Make requires grad check run only when grad mode is enabled (#60740)

albanD committed 3 years ago

f118d20b

Add an optional Device parameter to pin_memory/is_pinned that does nothing (#60201)

ezyang committed 3 years ago

3ad3f20b

Remove some unnecessary functions from CUDAHooks (#59655)

ezyang committed 3 years ago

85af24f5

Newer Older