strisunshinewentingwang/pytorch

Pull Requests Commits

Catherine Lee committed 2 years ago

f1381d6a

Modify repeat_interleave docs to highlight potential overloading (#99650)

tfsingh committed 2 years ago

2a6a159c

Add bertmaher to triton pin CODEOWNERS (#100390)

ezyang committed 2 years ago

73dac484

Use correct standard when compiling NVCC on Windows (#100031)

r-barnes committed 2 years ago

5f92909f

Add CUDA 12.1 CI workflows (#98832)

ptrblck committed 2 years ago

73645a84

Improve detection of workspace/non-output allocations in cudagraphs (#99985)

eellison committed 2 years ago

3edff6b6

Report timeout/infra_error instead of 0.0000 on infra error (#100372)

ezyang committed 2 years ago

5d93265c

[Easy][FSDP] Clarify `_use_unsharded_grad_views` comment (#100359)

Andrew Gu committed 2 years ago

a014d1b1

Refactor repro/minifier into CLI; add analyze (#100226)

ezyang committed 2 years ago

2d8deffc

Revert "Produce constant variables in cases where a SymNode is created with a constant (#100144)"

pytorchmergebot committed 2 years ago

89c43f41

[FSDP] Fix `use_orig_params=True`, CPU offload, `no_sync()` (#100180)

Andrew Gu committed 2 years ago

83b803c2

[BE] Fix SIM109 `compare-with-tuple` (#100337)

justinchuby committed 2 years ago

e779a30d

[BE] Fix all B022 `useless-contextlib-suppress` (#100335)

justinchuby committed 2 years ago

01abbfba

Produce constant variables in cases where a SymNode is created with a constant (#100144)

voznesenskym committed 2 years ago

d7bdfd34

[inductor] avoid zero division error for dropout (#100222)

Valentine233 committed 2 years ago

cc3ed8ae

Fix intermediate hooks on inplace buffers, enable it in testing (#100322)

ezyang committed 2 years ago

beb7f795

Use sympy.And instead of bitwise operator, for better promotion (#100328)

ezyang committed 2 years ago

155fa4e7

Skip invalid grads in outplace foreachs' backward (#100256)

crcrpar committed 2 years ago

6c934a89

fix TIMM mobilevit_s complier issue for dynamic CPU path (#100230)

XiaobingSuper committed 2 years ago

76bcc872

[decomp] Bad accuracy for elu_backward (#100284)

anijain2305 committed 2 years ago

e1021ec5

Revert "Move win-vs2019 build and test to unstable (#100281)"

pytorchmergebot committed 2 years ago

3d55bce3

[MPS] Fix `layer_norm_backward_mps` key (#100295)

malfet committed 2 years ago

2442858f

[dynamo] Compile torchvision augmentations (#100292)

anijain2305 committed 2 years ago

03806edd

Update docstring for dynamo.export tracing_mode (#100205)

BowenBao committed 2 years ago

6647e61a

Revert "Run test_fx_to_onnx_with_onnxruntime serially (#100298)"

pytorchmergebot committed 2 years ago

9075e3c2

[ONNX] Skip flaky dynamic test in CI (#100297)

BowenBao committed 2 years ago

1267897d

Run test_fx_to_onnx_with_onnxruntime serially (#100298)

huydhn committed 2 years ago

3a3f781f

Add size check before calling .back() in rpc/script_call.cpp (#94297)

m4drat committed 2 years ago

7684044b

fix(docs): torch.autograd.graph.Node.register_hook can override grad_inputs, not grad_outputs (#100272)

olokshyn committed 2 years ago

35991df5

Update testing aggregate data (#100070)

PaliC committed 2 years ago

2b79d6c4

Older