strisunshinewentingwang/pytorch

Pull Requests Commits

[ONNX] Test and address autograd.func (FusedLayerNorm) shape inference (#81931)

titaiwangms committed 2 years ago

6cb8f170

[torch] Add more functions to __init__.pyi.in for torch._C for Node and Value (#79654)

dulinriley committed 2 years ago

852c9b22

[ONNX] Quantization support for quantized::cat (#79826)

justinchuby committed 2 years ago

63133d3f

[ONNX] Add quantization support to _avg_pool opset 9 and clean up (#79793)

justinchuby committed 2 years ago

8d14d6c1

[ONNX] Quantization support for five ops (#78103)

justinchuby committed 2 years ago

98f81df2

[ONNX] Variable length argument support for quantized_args (#78775)

justinchuby committed 2 years ago

cb2ca482

Fix invalid read in masked softmax (#82272) (#82272) (#82405)

atalman committed 2 years ago

Verified 617c4fe5

[ci] Release only change: bump macos worker instance type (#82113)

atalman committed 2 years ago

Verified f469bc1f

Fix deserialization of TransformerEncoderLayer (#81832) (#81832) (#82094)

atalman committed 2 years ago

Verified 66f6e793

[CI] Disable ios-12-5-1-x86-64 (#81612) (#81612) (#82096)

atalman committed 2 years ago

Verified 35eb4884

1.12.1/bt fix (#81952)

erichan1 committed 2 years ago

Verified e65e4ac1

MPS cherry picks for 1.12.1 (#81976)

atalman committed 2 years ago

Verified e8534b92

Disable XLA builds (#80099) (#80099) (#81977)

atalman committed 2 years ago

Verified 03b82bdd

Add check for cuda lazy init (#80912) (#80912) (#81970)

atalman committed 2 years ago

Verified 48947f73

Raise proper timeout when sharing the distributed shared seed (#81666) (#81666) (#81892)

atalman committed 2 years ago

Verified 787b469b

Cudnn conv cache key patch (#81418) (#81418) (#81888)

atalman committed 2 years ago

Verified 37b49cf9

[DataLoader] Locking lower ranks seed recepients (#81071) (#81071) (#81886)

atalman committed 2 years ago

Verified 91605088

Change cudnn incompatibility message wording (#80877) (#80877) (#81881)

atalman committed 2 years ago

Verified 60f9724e

Make nn.stateless correctly reset parameters if the forward pass fails (#81262) (#81262) (#81880)

atalman committed 2 years ago

Verified 23ec48ce

Add 3.10 stdlib to torch.package (#81261) (#81261) (#81879)

atalman committed 2 years ago

Verified 1d8ea836

[forward ad] Fix codegen to ignore undefined outputs (#81114) (#81114) (#81878)

atalman committed 2 years ago

Verified 5525230f

Don't error if _warned_capturable_if_run_uncaptured not set (#80345) (#80345) (#81877)

atalman committed 2 years ago

Verified 12954c72

fix weight norm backward bug on CPU when OMP_NUM_THREADS <= 2 (#80930) (#80930) (#81872)

atalman committed 2 years ago

Verified a93c9014

[Prims] Unbreak CUDA lazy init (#80899) (#80899) (#81870)

atalman committed 2 years ago

Verified 9d9bba4c

Use fabi-version=11 to ensure compatibility between gcc7 and gcc9 binaries (#81058) (#81058) (#81884)

atalman committed 2 years ago

Verified 0e43325a

Allow register float16 weight_norm on cpu and speed up test (#80600) (#80600) (#81866)

atalman committed 2 years ago

Verified b556fb30

Fix `Module.share_memory` error (#80843) (#80843) (#81867)

atalman committed 2 years ago

Verified 1680cd0e

Don't error if _warned_capturable_if_run_uncaptured not set (#80345) (#80345) (#81865)

atalman committed 2 years ago

Verified 86864674

remove overly restrictive checks for cudagraph (#80881) (#81858)

atalman committed 2 years ago

Verified cd6ec073

Fix distributed store to use add for the counter of DL shared seed (#80348) (#80348) (#81860)

atalman committed 2 years ago

Verified 430416fc

Older