strisunshinewentingwang/pytorch

Pull Requests Commits

[inductor] improve cpp vec implementation of square

Valentine233 committed 2 years ago

9953038f

Error when jit.trace/script is used with torch.compile (#91681)

msaroufim committed 2 years ago

fa92b6a7

Fix node provenance tracking (#95901)

Chillee committed 2 years ago

e8cd173a

Automated submodule update: kineto (#95798)

facebook-github-bot committed 2 years ago

36a6e2c5

[inductor] Add some simple decomps (#96039)

jansel committed 2 years ago

5dd52e25

[CI] Skip xcit_large_24_p8_224 in TIMM (#96048)

desertfire committed 2 years ago

60cf9561

[CI] Further tighten the checking of two eager runs (#95902)

desertfire committed 2 years ago

1359d16f

[inductor] use triu ref instead of lowering (#96040)

Natalia Gimelshein committed 2 years ago

43e71cdd

[dtensor] refactor shape/offset calculation (#95923)

wanchaol committed 2 years ago

789fc4c2

Fix training enablement in AOTAutograd (#95975)

ezyang committed 2 years ago

af8dbe7e

Add doc test in graph_drawer.py (#95919)

wenzhe-nrv committed 2 years ago

b7a3f331

[inductor] Enable some decomps (#96038)

jansel committed 2 years ago

5da6da65

Transformers: fix src and key padding mask bool regression (#96009)

Michael Gschwind committed 2 years ago

03b6e697

[MPS] Fix bidirectional LSTM & small one-direction LSTM fix (#95563)

alexdremov committed 2 years ago

78da315a

Update dynamic skips (#95966)

ezyang committed 2 years ago

c7c4a203

Fix codegen logic for foreach derivatives (#95263)

crcrpar committed 2 years ago

49f6849f

[Dynamo] Fix number of inputs in onnxrt and tvm backend (#95429)

YJ Shi committed 2 years ago

a10897a3

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

wconstab committed 2 years ago

26045336

[Dynamo] Support call dict with list/tuple as input (#95928)

yanboliang committed 2 years ago

6ca286df

Revert "[inductor] Improve error messages (#95567)" (#96014)

jansel committed 2 years ago

43dd043e

Add various uninterpreted bit tensor data types (try 2) (#95860)

vkuzo committed 2 years ago

dc70e817

[vision hash update] update the pinned vision hash (#95932)

pytorchmergebot committed 2 years ago

5e1067bc

Improve error message for instance norm when channels is incorrect (#94624)

soulitzer committed 2 years ago

7ff9612e

[MPS] Error on unsupported types (#95982)

malfet committed 2 years ago

436993d5

[BE] Remind people there are rsts to update in docs/source (#95914)

janeyx99 committed 2 years ago

f4b33791

Make int unspecialization actually work (#95621)

ezyang committed 2 years ago

d303665d

Fix wrong handling of `grad_scale` & `found_inf` in fused optimizers (#95847)

crcrpar committed 2 years ago

7d765cdc

Prettify assert expr in self.symbol_to_source failure (#95972)

ezyang committed 2 years ago

d214d82a

[SPMD] Introduce the cross-iteration graph optimization framework (#94803)

fegin committed 2 years ago

4d9c499d

Remove fake inputs from control flow (#95988)

angelayi committed 2 years ago

5a07c3d3

Older