strisunshinewentingwang/pytorch

Pull Requests Commits

Debug logging around DDP mixed precision copies (#96438)

rohan-varma committed 1 year ago

707d8925

[inductor] Consolidate codegen functions in sizevars.py into wrapper.py (#96654)

desertfire committed 1 year ago

b60d6e24

Update CI skips (#96745)

ezyang committed 1 year ago

037acd5a

[PyTorch][XNNPACK] Update build files for newly added kernels (#95911)

digantdesai committed 1 year ago

a198ce6d

[BE]: Merge startswith calls - rule PIE810 (#96754)

Skylion007 committed 1 year ago

dd5e6e85

Revert "[CI] switch torchbench to a pinned version (#96553)"

pytorchmergebot committed 1 year ago

be4eaa69

Revert "Update perf smoke test threshold in check_hf_bert_perf_csv.py (#96772)"

pytorchmergebot committed 1 year ago

2951a75c

[Inductor] aten.{avg_pool2d/max_pool2d_with_indices} arguments can be 1 element tuple (#96727)

yanboliang committed 1 year ago

e7d795dc

Automatically register/clear dynamo profiler hooks while profiling (#96199)

wconstab committed 1 year ago

784dd583

Update perf smoke test threshold in check_hf_bert_perf_csv.py (#96772)

weiwangmeta committed 1 year ago

2eed4493

Add support for torch.complex in functorch (#96032)

lithomas1 committed 1 year ago

159145a1

Add `torch._check*` functions analogous to C++ `TORCH_CHECK*` (#88725)

kurtamohler committed 1 year ago

06b72851

add amp support for custom backend (#96188)

heidongxianhua committed 1 year ago

cf12edee

Replace non-reentrant checkpoint with a rewrite that can be nested and contain grad (#90105)

soulitzer committed 1 year ago

d30db9a2

[MPS] Add C++ API support for MPS backend (#96668)

razarmehr committed 1 year ago

234df299

Revert "Default specialize_int to False (#96624)"

pytorchmergebot committed 1 year ago

ba4fb9b6

Fix signed-unsigned compare in FlattenIndicesCommon.h (#96765)

malfet committed 1 year ago

da1489e4

One line print for check_graph_breaks (#96750)

wconstab committed 1 year ago

66871d61

Make share_memory_ call thread safe within itself. (#96664)

albanD committed 1 year ago

6ea790c5

Fixes 96676 (#96714)

cyyever committed 1 year ago

799521fa

[CI] switch torchbench to a pinned version (#96553)

desertfire committed 1 year ago

61d6ccd2

Default specialize_int to False (#96624)

ezyang committed 1 year ago

1ac8782d

Only allow SymInt across partitioner boundaries, and fixes (#96653)

ezyang committed 1 year ago

02f6d14b

Mark empty, rand, randn as core aten op (#96661)

SherlockNoMad committed 1 year ago

9cb02b2e

[memory profiling] add a facility to gather combined C++/Python/TorchScript stack traces. (#95541)

zdevito committed 1 year ago

4e1060c6

[inductor][CI] also skip rexnet_100 on non-dynamic shapes (#96691)

davidberard98 committed 1 year ago

6e3d51b0

[FSDP] Relax `sharded_grad` assert to allow IDLE (#96584)

awgu committed 1 year ago

a916d649

bsr_dense_mm Triton kernel: fix out kwarg (#96648)

nikitaved committed 1 year ago

05dda7ff

[AO] Update qLSTM implementation to remove unsupported backend ops (#96436)

Ninja91 committed 1 year ago

40df3b41

Moves SDPA backward helper native function to functionsmanual.cpp (#95821)

drisspg committed 1 year ago

7ec0d6f0

Newer Older