strisunshinewentingwang/pytorch

Pull Requests Commits

Fix debug dir bugs and minifier output directories (#87682)

mlazos committed 2 years ago

44d7ba7e

Intercept aten._reshape_alias for nvFuser (#87072)

IvanYashchuk committed 2 years ago

ff2569bc

Fix typos under functorch directory (#87663)

kiszk committed 2 years ago

a3d495bd

Fix stride for prims.where (#87563)

SherlockNoMad committed 2 years ago

0b162f5b

[Dynamo] Symbolic shape guards (#87570)

voznesenskym committed 2 years ago

bc194948

[ao] Adding FAQ to docs (#87322)

HDCharles committed 2 years ago

d0e12d1c

Fix _refs for aten.zeros/ones/empty/randn (#87569)

SherlockNoMad committed 2 years ago

ece3758a

[inductor] Revert channels-last support (#87588)

anijain2305 committed 2 years ago

ebe5aad4

Fixed minor typos in torch.flip and torch.rot90 (#87724)

scaomath committed 2 years ago

312628d2

[ONNX] Fix pad Circular Mode (#86984)

titaiwangms committed 2 years ago

52ac8adc

Use setup_instance script to enable conda and load cuda libraries (#87296)

xuzhao9 committed 2 years ago

e532fb9a

build: support DNNL_GRAPH_CPU_RUNTIME=TBB (#87512)

min-jean-cho committed 2 years ago

7a6808c5

Add prepend argument to nn.Module hooks (#87370)

mrshenli committed 2 years ago

82698b89

[ONNX] replace AT_ASSERT with TORCH_INTERTNAL_ASSERT take 2 (#86405)

titaiwangms committed 2 years ago

82dff8ee

[ONNX] Support quantized::conv1d_relu (#85997)

titaiwangms committed 2 years ago

65b4a633

Disable test_inductor_timm_shard (#87710)

desertfire committed 2 years ago

15370d32

Graph-break on FSDP in dynamo (#87420)

wconstab committed 2 years ago

874625e0

[pytorch] Layer norm backward speed gain with warp shuffles (#87445)

valentinandrei committed 2 years ago

b6f28334

Add named_buffers to torchdynamo nn_module (#87644)

tugsbayasgalan committed 2 years ago

7b597825

consider numel args when identifying aligned args (#87394)

stumpOS committed 2 years ago

8a2a4ed4

Add get_guard_expr to symbolic_shapes which returns all guards in a single expression (#87665)

Chillee committed 2 years ago

569eebb4

Prefer python meta function over c++ meta function (#87426)

SherlockNoMad committed 2 years ago

eb99c1ef

[ONNX] Add Support on 0d tensor Broadcast (#87211)

titaiwangms committed 2 years ago

65601f5e

Revert "Intercept aten._reshape_alias for nvFuser (#87072)"

pytorchmergebot committed 2 years ago

5308886e

Performance improvment to cumulative seq len (#87530)

drisspg committed 2 years ago

0cba7888

[inductor] Trivial smoke-test (#87598)

bertmaher committed 2 years ago

87163fe8

[ROCm] [FakeTensorTest] Enable test_fallback_memory_prop (#85760)

jaglinux committed 2 years ago

9efca7c0

Support `signbit` in MPS. (#87214)

dfalbel committed 2 years ago

e818574e

Intercept aten._reshape_alias for nvFuser (#87072)

IvanYashchuk committed 2 years ago

163a829c

[vision hash update] update the pinned vision hash (#87639)

pytorchmergebot committed 2 years ago

9bbdc7ab

Newer Older