strisunshinewentingwang/pytorch

Pull Requests Commits

[ci] shard distributed on cuda 10.2

suo committed 2 years ago

68e1e2d9

[static-runtime] refactor out variant for `aten::embedding_bag` (#76207)

tenpercent committed 2 years ago

a41d4f27

Add mode property to distributions. (#76690)

tillahoffmann committed 2 years ago

40576bce

Apply black lint CI to PrimTorch

ezyang committed 2 years ago

feaa64f7

Revert "Apply black lint CI to PrimTorch"

pytorchmergebot committed 2 years ago

bbb1a55d

Fix prims lint broken on trunk due to land race (#77271)

janeyx99 committed 2 years ago

b0bd5926

Update operator list for AutocastCPU (#68725)

CaoE committed 2 years ago

5993cc0b

[test] try to fix torch.tensor for functionalization

bdhirsh committed 2 years ago

9edee09e

General fixes for ShardedTensor op framework.

pritamdamania87 committed 2 years ago

b91a1490

[complex32] add, sub, neg (#77179)

kshitij12345 committed 2 years ago

420b49c3

Add the Runtime components for MPS backend. (#76725)

kulinseth committed 2 years ago

f348b1b2

[ONNX] Bug fix: opset_version checked before set (#76928)

BowenBao committed 2 years ago

93953a48

[GHF] Improve authorship detection (#77266)

malfet committed 2 years ago

fa2bf5ab

[ci] don't run cpu tests on win-cuda builds

suo committed 2 years ago

eaee4ff9

[FSDP] Fix local_state_dict and state_dict_type bugs

fegin committed 2 years ago

4f4ebc54

Register prims as operators.

ezyang committed 2 years ago

0a14a4c2

Apply black lint CI to PrimTorch

ezyang committed 2 years ago

188854ee

Update Metal for "Consolidate customization contiguous/sizes policy into unified policy"

ezyang committed 2 years ago

0f4b1ebc

linux focal builds install cmake from conda (#76964)

jeffdaily committed 2 years ago

e9c681d2

Speedup segmented sort with large nsort

peterbell10 committed 2 years ago

d9351ed5

move SymInt and SymIntArrayRef to c10/core (#77009)

Krovatkin committed 2 years ago

99339fdd

[NVFuser] Add environment variable to force disable NVFuser

davidberard98 committed 2 years ago

6fd14ba9

[NVFuser] assert that vectors are the same size in translateSingleWelford

davidberard98 committed 2 years ago

3c2e0dc6

Update amp document with CPU Training/Inference Examples (#77244)

leslie-fang-intel committed 2 years ago

f2d9fc18

add channels last support for slow_conv_dilated2d

mingfeima committed 2 years ago

3d0e6f16

[Reland] Mixed precision batchnorm fix (#77234)

rohan-varma committed 2 years ago

94939008

[composite compliance] backward, forward: nn.linear (#77151)

kshitij12345 committed 2 years ago

140c8168

[composite compliance] backward: gather, take_along_dim (#77152)

kshitij12345 committed 2 years ago

4a45b88d

[composite compliance] backward: masked_select, combinations (#76794)

kshitij12345 committed 2 years ago

299ebf1e

expand the coverage of conv folding (#75724)

jiayisunx committed 2 years ago

dc5fe2b3

Newer Older