strisunshinewentingwang/pytorch

Pull Requests Commits

Fixing a bug where allocating a 4GB block results in using 8GB of memory (#95827)

akamali committed 1 year ago

065de430

[MPS] Fall back multi-layer LSTM on macOS 12 (#90909)

qqaatw committed 1 year ago

a87f3f61

[ONNX] Export logical_not (#96315)

BowenBao committed 1 year ago

b0a580a2

[ONNX] STFT Support (#92087)

urinieto committed 1 year ago

5f89d147

[PyTorch] Use c10::FastMap for memoizing in Pickler (#96360)

swolchok committed 1 year ago

69d3fa2e

[PyTorch] add c10/util/FbcodeMaps.h (#96359)

swolchok committed 1 year ago

cc798f1a

reland #96248 [inductor] show performance for each autotune config for a kernel (#96458)

shunting314 committed 1 year ago

cc699c56

Add env PYTORCH_TEST_DO_NOT_USE_PYTEST as an option to not use pytest in unit testing (#96444)

xwang233 committed 1 year ago

cf3d3a58

Skip rexnet_100 in dynamic CI (#96474)

ezyang committed 1 year ago

ff2e14f2

Fix missing items() typo (#96417)

ezyang committed 1 year ago

79313345

[CI] Use different subdirectories for amp and float32 nightly perf run (#96470)

desertfire committed 1 year ago

8ef8bd02

Extra CR comments from #95621 (#96043)

ezyang committed 1 year ago

384d3ec2

Revert "Optimize nn.Module __call__ fast path for dynamo (#95931)" (#96242)

wconstab committed 1 year ago

2f6a371a

[ONNX] Fix circular padding to support dynamic axes (#95647)

ilyasher committed 1 year ago

6154be1d

[Quant][fx] Create new FX-based LSTM reference module (#96343)

andrewor14 committed 1 year ago

faa4cb29

[inductor] don't match indirect indexing in fusion (#96273)

Natalia Gimelshein committed 1 year ago

05b679ce

test only smaller block_k for mm_plus_mm (#96385)

Natalia Gimelshein committed 1 year ago

1bde36ba

[PyTorch] Add Vulkan support and tests for at::select.int operator, 4 dim/rank tensor case (#96228)

liuk22 committed 1 year ago

090b3b95

Correctly resolve dispatch keys for PyOperator (#96306)

ezyang committed 1 year ago

6a675f7c

Assert that there are no None arguments to backwards (#96300)

ezyang committed 1 year ago

30c4ea13

Fix https://github.com/pytorch/pytorch/issues/96278 (#96299)

ezyang committed 1 year ago

bbe1b9bb

[MPS] Allow `float16` input to float32 `LayerNorm` (#96430)

malfet committed 1 year ago

075a4944

[Autograd] `expand_as` instead of `clone` to get `AccumulateGrad` (#96356)

awgu committed 1 year ago

457396fc

[FSDP] Add unsafe setattr gated by env var (#96326)

awgu committed 1 year ago

cb42bc2c

Revert "[reland][inductor] Add an AOT compilation mode for Inductor CPP backend (#95985)"

pytorchmergebot committed 1 year ago

fe05266f

Retry CI Android emulator test (#96163)

huydhn committed 1 year ago

44d8e6c2

[ONNX] Bump onnx submodule to release 1.13.1 from rc2 (#96325)

BowenBao committed 1 year ago

df0ff34b

Rewrite fallthrough to more closely match how C++ works (#96304)

ezyang committed 1 year ago

32ffd706

Refactor to reduce duplicate logic in torch._ops (#96302)

ezyang committed 1 year ago

67c329bc

Add missing types to inductor IR assert (#96221)

wconstab committed 1 year ago

4662ae5b

Newer Older