strisunshinewentingwang/pytorch

Pull Requests Commits

[NVFuser] Upstream push 1026 (#87779)

jjsjann123 committed 1 year ago

7b419e85

[MPS] Fix embedding backward with scalar index (#82809)

qqaatw committed 1 year ago

15e54293

Modified roundup_power2_divisions to specify the number of divisions for each power of two interval (#87290)

c-odrin committed 1 year ago

5b767d40

[vulkan] enable prepacking for Batchnorm op (#88433)

SS-JIA committed 1 year ago

b78b8727

Revert "Revert "Put Python Dispatcher cache in dict, clear it on new registrations. (#88329)"" (#88489)

ezyang committed 1 year ago

53eac1d4

nvprim python runtime dtype correctness patch (#88452)

jjsjann123 committed 1 year ago

79abea56

Revert "remove assert_allclose from torch.testing (#87974)"

pytorchmergebot committed 1 year ago

8c1c6759

Fix typo in clones (#88501)

ezyang committed 1 year ago

bda688c1

[torch package] Treat builtins as default extern module (#88385)

842974287 committed 1 year ago

633f0d62

Add dep on Accelerate framework to torch podspecs (#88422)

jmdetloff committed 1 year ago

ead36e5a

[Vulkan][TCC] Add tests for conv2d prepack context (#88316)

manuelcandales committed 1 year ago

dc00bb51

Add use_lazy_shape flag to GenLazyIr class (#88444)

wonjoolee95 committed 1 year ago

a171b063

TorchDynamo: enable convolution and batchnorm folding for inference path (#87435)

XiaobingSuper committed 1 year ago

b3206268

Introduce TORCH_DISABLE_GPU_ASSERTS (#84190)

pruthvistony committed 1 year ago

fbd08fb3

Add hf_bert + DDP multigpu test (#88435)

wconstab committed 1 year ago

70b00b13

TorchDynamo: Add linear binary fusion for cpu in BF16 inference mode (#87066)

XiaobingSuper committed 1 year ago

71f793d3

Run all fallback kernels with FakeTensor (#88248)

eellison committed 1 year ago

7d95b1e3

TorchDynamo: Add linear unary fusion for cpu in BF16 inference mode (#87065)

XiaobingSuper committed 1 year ago

e4efea4f

[MPS] Add native `cumsum` implementation (#88319)

malfet committed 1 year ago

657f2e12

TorchDynamo: Add convolution binary fusion for cpu in inference mode (#87064)

XiaobingSuper committed 1 year ago

52173188

Disable Current Modes when printing Tensor (#88344)

eellison committed 1 year ago

2ce2fc13

[LTC] Update merge_rules.yaml (#88291)

alanwaketan committed 1 year ago

e804c722

[FSDP][Docs] Reword `sharding_strategy` docs and other minor doc changes (#88431)

awgu committed 1 year ago

a84d68cd

[FSDP][Docs] Simplify CPU offload docs (#88430)

awgu committed 1 year ago

ff23e07b

[FSDP] Allow to use TorchDispatch with FSDP (#88014)

fegin committed 1 year ago

4de50b25

Reset NVIDIA devices stuck in failed mode (#88459)

huydhn committed 1 year ago

31ebd3cc

[FSDP][Docs] Simplify `mixed_precision` ctor docs (#88429)

awgu committed 1 year ago

ab8f3333

[dynamo] Skip mutation detection for inference mode (#88406)

anijain2305 committed 1 year ago

36582574

Revert "Put Python Dispatcher cache in dict, clear it on new registrations. (#88329)"

pytorchmergebot committed 1 year ago

410ce96a

[functorch] make hessian docs actually use hessian function (#88451)

samdow committed 1 year ago

9946041a

Older