strisunshinewentingwang/pytorch

Pull Requests Commits

Support int32 indices in torch.repeat_interleave (#55102)

Brandon Lin committed 3 years ago

d806b061

Added OpInfos of add & mm (#55915)

imaginary-person committed 3 years ago

b6b2fc7e

Add OpInfo for (nan)quantile (#55548)

heitorschueroff committed 3 years ago

92991d95

Revert D27600457: [pytorch][PR] Support factory kwargs in torch.nn modules

samestep committed 3 years ago

d05e7c16

[special] OpInfo `i0e`: fix missing check (#56232)

kshitij12345 committed 3 years ago

7d175591

Support factory kwargs in torch.nn modules (#54508)

jbschlosser committed 3 years ago

1077f872

Make tensordot resize output tensor's size if out= argument is specified & make it safely cast & copy output (#56286)

imaginary-person committed 3 years ago

7513455c

add tests for torch.testing (#54784)

pmeier committed 3 years ago

0e106fce

Updated internal code for orgqr function (#56247)

IvanYashchuk committed 3 years ago

2219286d

[NNC] Make normalization transformation in-place (#56158)

navahgar committed 3 years ago

b387f7ca

[pytorch][PR] Automated submodule update: tensorpipe (#56348)

Lucas Hosseini committed 3 years ago

22d4d9f4

[CUDA graphs] Allows DeviceCachingAllocator to capture cross-stream memory use (#55860)

mcarilli committed 3 years ago

ffdecc1a

Porting logcumsumexp tests to OpInfo (#56135)

iramazanli committed 3 years ago

3e42da09

[c10d] Remove deprecated use of torch.LongTensor, torch.ByteTensor (#55861)

rohan-varma committed 3 years ago

ce05b7a3

Short circuits DistributedDataParallel._recursive_to's copy and stream syncs if input is already on the right device (#55624)

mcarilli committed 3 years ago

a24b1724

[NNC] Fuse loops that have the same bounds as expressions (#55997)

navahgar committed 3 years ago

29c5cb79

OpInfo porting for logsumexp operator (#55520)

iramazanli committed 3 years ago

b0e0841f

Vectorize copysign on CPU (#51792)

peterbell10 committed 3 years ago

8c74e1b8

Added OpInfos for eq, ne, ge, gt, le, and lt (#55709)

imaginary-person committed 3 years ago

36b476cc

[TensorExpr] Add support for constant tensors in tensorexpr kernel. (#56319)

Mikhail Zolotukhin committed 3 years ago

85126629

[TensorExpr] TensorExprKernel: switch type of tensors_ from Tensor to Buf. (#56318)

Mikhail Zolotukhin committed 3 years ago

dd9ef529

Allow inlining of more Tensor methods (#53905)

peterbell10 committed 3 years ago

50d4c63f

[TensorPipe] Update tensorpipe subodule + remove TP_NEW_API switch. (#56260)

Lucas Hosseini committed 3 years ago

be2a0805

[nnc] Only lower float conv2d's (#56289)

bertmaher committed 3 years ago

928a4733

Add adaptive_avgpool2d to the set of fusible ops (#56180)

Krovatkin committed 3 years ago

04e7891a

clean up unused reduction functions in THC (#56293)

ngimel committed 3 years ago

7636cb6b

A heuristic to avoid perf incompatible MKLDNN formats for binary ops (#56089)

Krovatkin committed 3 years ago

a4348358

[FX] Make shape_prop handle targets with aggregate outputs (#56221)

James Reed committed 3 years ago

d02919dd

Fix warnings in ivalue test (#56303)

davidriazati@fb.com committed 3 years ago

72a93a63

fx quant: fix subtle bug in BinaryOpQuantizeHanlder logic in matching (#56294)

vkuzo committed 3 years ago

48e675ac

Newer Older