strisunshinewentingwang/pytorch

Pull Requests Commits

[quant][graphmode][fx] Enable fuse handler for sequence of 3 ops (#69658)

jerryzh168 committed 2 years ago

bf089840

[SR] Fix stack/concat bug (#68777)

Mike Iovine committed 2 years ago

102684b2

[JIT] Enable freezing for sparse COO tensors (#69614)

davidberard98 committed 2 years ago

ebc35a7e

Revert D32498572: allow external backend codegen to be used without autograd kernels

bdhirsh committed 2 years ago

33363cea

Revert D32498569: allow external backend codegen to toggle whether to generate out= and inplace kernels

bdhirsh committed 2 years ago

f6cad534

Revert D32498570: make codegen'd device guards not cuda-specific. Allow them to be used in external codegen

bdhirsh committed 2 years ago

0ef52363

Another attempt to fix version comparison check (#69939)

malfet committed 2 years ago

24ee1d13

Add low level torch.profiler.kineto_profile base class (#63302)

guotuofeng committed 2 years ago

d4f83134

[nn] mha : no-batch-dim support (python) (#67176)

kshitij12345 committed 2 years ago

e8d5c7cf

Open source trt lowering workflow (#69381)

Shirong Wu committed 2 years ago

37ec99c0

Build clang builds with -Werror (#69712)

malfet committed 2 years ago

930067d1

[ONNX] Add BFloat16 type support when export to ONNX (#66788)

hwangdeyu committed 2 years ago

c76c6e9b

[shard] add ShardedOptimizer (#68607)

wanchaol committed 2 years ago

800a457b

Porting index_add to structured kernels, add an out variant (#65993)

bdhirsh committed 2 years ago

457ba1dd

fix CompositeImplicitAutograd ops improperly labeled (#69863)

bdhirsh committed 2 years ago

9594a94d

[c2] Remove unused private fields (#69709)

malfet committed 2 years ago

269e9266

Update run_test.py (#69920)

malfet committed 2 years ago

fef99819

[Quant][fx] Lower reference conv[1-3]d module (#69228)

andrewor14 committed 2 years ago

3e43c478

[DateLoader] more clearly expose 'default_collate' and 'default_convert' to users (#69862)

NivekT committed 2 years ago

b67eaec8

TestMathBits: Call functions with original sample input values (#68947)

peterbell10 committed 2 years ago

1188d89a

Add support for transformer layout of masked_softmax (#69272)

zrphercule2 committed 2 years ago

1a299d8f

make codegen'd device guards not cuda-specific. Allow them to be used in external codegen (#68531)

bdhirsh committed 2 years ago

2e7a91c4

allow external backend codegen to toggle whether to generate out= and inplace kernels (#68530)

bdhirsh committed 2 years ago

aa0cf68c

allow external backend codegen to be used without autograd kernels (#68529)

bdhirsh committed 2 years ago

b83b6f74

Allow row sizes to support int64/size_t. (#69303)

Rick Weyrauch committed 2 years ago

8acd0a8b

Modify torch.movedim to handle scalar as no-op (#69537)

francescocastelli committed 2 years ago

2c9dd886

[nnc][fix] xnnpack ifdef (#69870)

IvanKobzarev committed 2 years ago

7503ec58

[Static Runtime] Skip ReplaceWithCopy when inputs have writters (#69819)

donaldong committed 2 years ago

f7294cd8

Properly import LooseVersion (#69904)

malfet committed 2 years ago

07767569

`print` fix in `lr_scheduler` (#68338)

OverLordGoldDragon committed 2 years ago

fdcb78df

Newer Older