strisunshinewentingwang/pytorch

Pull Requests Commits

Don't Require contiguous For Extern Kernels (#87650)

eellison committed 2 years ago

e6ea0a4a

Fix nvFuser Fusion Definition printing of Squeeze and Permute (#88041)

kevinstephano committed 2 years ago

8ef9bda1

[reland][fx][subgraph_rewriter] Change match_filter to be a List in replace_pattern_with_filters (#87998)

jerryzh168 committed 2 years ago

68f9f256

Add meta implementation for aten.max.dim (#88005)

tugsbayasgalan committed 2 years ago

2c7de4a1

remove assert on tensor inputs to FusionGroup (#88018)

jjsjann123 committed 2 years ago

97b3eeac

Add UBSAN to ASAN (#88055)

malfet committed 2 years ago

e1c123d2

[11/N] Update all_to_all with CPU/CUDA implementations (#86407)

H-Huang committed 2 years ago

81f74eed

Rename 'nvfuser' to 'ts_nvfuser' indicating TorchScript usage (#88188)

IvanYashchuk committed 2 years ago

90fa2570

[10/N] Update barrier with CPU/CUDA implementations (#86368)

H-Huang committed 2 years ago

bed81027

[FSDP()][16/N] Refactor post-forward/pre-backward (#87929)

awgu committed 2 years ago

1f34067e

[FSDP()][15/N] Refactor `_init_streams()` (#87928)

awgu committed 2 years ago

5a53f024

[FSDP()][14/N] Refactor pre-forward/post-backward (#87927)

awgu committed 2 years ago

90c5f856

torchdynamo support modules() for nn_module (#88023)

ydwu4 committed 2 years ago

eb91e8a5

Fix meta function for aten.addmm (#88068)

SherlockNoMad committed 2 years ago

de1f641f

Add unit test for torch_geometric library (#85937)

Thiago Crepaldi committed 2 years ago

fdc41978

[codev] Make backport work with flatbuffer models (#88127)

qihqi committed 2 years ago

5c3666cb

Add ability to freeze storages inside functionalization (#88141)

ezyang committed 2 years ago

bb7e6254

Inline Alias into FunctionalStorageImpl (#88140)

ezyang committed 2 years ago

61f955dd

always realize output regardless of the number of reads (#88046)

Natalia Gimelshein committed 2 years ago

73c9911f

Fix meta for aten.fill, constant_pad_nd, _adaptive_avg_pool2d (#88069)

SherlockNoMad committed 2 years ago

c368c0fa

Change dynamo/distributed tests to use cuda/nccl (#88133)

wconstab committed 2 years ago

82a9de16

[BE]fix DDP when the number of output features is zero (#87793)

zhaojuanmao committed 2 years ago

44f8efd5

[9/N] [Dispatchable Collectives] Update reduce_scatter with CPU / CUDA implementations (#86166)

H-Huang committed 2 years ago

20d849b9

Reenable assert sanity testing with ADInplaceOrView reenable (#88102)

ezyang committed 2 years ago

1e5d33b6

[Reland][ONNX] Move all torch.onnx.export related tests to test/onnx (#87292)

titaiwangms committed 2 years ago

bdb14238

Update _distributed_c10d.pyi (#88088)

yhcharles committed 2 years ago

62988e4f

[FSDP()][13/N] Refactor unshard/reshard/grads (#87926)

awgu committed 2 years ago

b1750d04

[FSDP()][12/N] Easy cleanup (#87925)

awgu committed 2 years ago

8039317c

[FSDP()][10/N][11/N] Introduce composable (ctor only) (#87924)

awgu committed 2 years ago

c1e28731

[FSDP()][9/N] Refactor ctor (continued) (#87923)

awgu committed 2 years ago

78170701

Newer Older