strisunshinewentingwang/pytorch

Pull Requests Commits

[GPU] Introduce USE_PYTORCH_METAL (#46383)

xta0 committed 3 years ago

04e5fcc0

Add flatten loops transformation (#46365)

navahgar committed 3 years ago

fa108bd2

[GPU] Add metal to DispatchKeySet (#46455)

xta0 committed 3 years ago

5da4a086

[WIP] Move catchAll to Math (#45939)

Ailing Zhang committed 3 years ago

8c629ecc

[Gradient Compression] Rename the first arg of pybinding of _register_comm_hook: ddp_model -> reducer. (#46498)

Yi Wang committed 3 years ago

d1ca7ef3

caffe2: use at::mt19937 instead of std::mt19937 (10x speedup) (#43987)

d4l3k committed 3 years ago

0c9787c7

[hotfix] remove test.pypi.org (#46492)

Rong Rong committed 3 years ago

e6e83bf8

Run __setstate__ when cloning modules (#45858)

bzinodev committed 3 years ago

11cc7f14

Removing hack so that NCCL is not removed in base Docker commands (#46495)

janeyx99 committed 3 years ago

478fa180

type check for torch.quantization.stubs (#46475)

Rong Rong committed 3 years ago

89108ba6

Move NCCL installation for xenial-cuda10.1 to Docker image instead of for every job (#46476)

janeyx99 committed 3 years ago

997e672a

Avoid mat1 references in mm_mat1_backward (#45777)

kurtamohler committed 3 years ago

28f8372b

caffe2/plan_executor: propagate exceptions from reporter substeps (#46424)

d4l3k committed 3 years ago

dd169ca1

[fx] allow custom behavior for args, kwargs, and bool (#45193)

zdevito committed 3 years ago

24ca2763

[NCCL] Modularize ncclCommWatchdog (#46051)

osalpekar committed 3 years ago

2e2fe8cf

Fix implicit cast in custom_function (#46445)

r-barnes committed 3 years ago

be0c4318

Make `torch.lu` support complex input on CUDA. (#45898)

nikitaved committed 3 years ago

9300a277

[Flaky tests] Fix test_all_gather_timeout test (#45989)

rohan-varma committed 3 years ago

5c5484c8

[caffe2] add concat benchmark (#46457)

jspark1105 committed 3 years ago

c37baa91

Fix out-of-bounds access for caching allocator calls (#46439)

malfet committed 3 years ago

b5702e23

install lcov in Docker image if coverage is specified (#46404)

janeyx99 committed 3 years ago

d6e60730

Fix deprecated warnings for nan_to_num (#46309)

RockingJavaBean committed 3 years ago

7b788d11

[mlf][efficiency] modify equalization scale operator to return single output (#46449)

hwangjeff committed 3 years ago

ecf63351

[TensorExpr] Correctly handle negative dimensions in aten::cat when lowering to tensor expressions. (#46446)

Mikhail Zolotukhin committed 3 years ago

4359c5e0

Remove variable_excluded_from_dispatch() check for factory functions. (#46371)

Ailing Zhang committed 3 years ago

ec5f81f9

fix type check for torch.quantization._numeric_suite (#46330)

Rong Rong committed 3 years ago

d1745c36

Add named tuple's error message and workaround for RET failure (#46347)

jwpark1985 committed 3 years ago

92921c82

Update VariableTypeManual.cpp to not use catchAllKernel. (#46353)

Ailing Zhang committed 3 years ago

d278e83e

Fixes bug in sspaddmm (#45113) (#45963)

vfdev-5 committed 3 years ago

dc7cd974

More Timer refinement (#46023)

Taylor Robie committed 3 years ago

dda95e69

Newer Older