strisunshinewentingwang/pytorch

Pull Requests Commits

Catherine Lee committed 2 years ago

910ed04d

Catherine Lee committed 2 years ago

a974fe72

Stop runner service when its GPU crashes (#97585)

huydhn committed 2 years ago

099b2801

Use the latest NVIDIA driver from setup-nvidia (#97840)

huydhn committed 2 years ago

2806fa44

[dynamo, benchmarks] Add inductor-mode (for max-autotune) and warm start options to dynamo benchmarks (#97719)

williamwen42 committed 2 years ago

b93e1f37

[SPMD] Make compile cache the compilation result and add option to perform transformation (#97836)

fegin committed 2 years ago

942e587d

Fix typo under torch/csrc/jit/runtime directory (#97243)

kiszk committed 2 years ago

d70f9c78

[RFC][inductor][index_put] fallback to aten in torch deterministic mode (#96898)

yuguo68 committed 2 years ago

1f71ac78

[Dynamo] Fix for tuple construction from tuple iterators (#97862)

mlazos committed 2 years ago

e6909f6c

Simplify by using yield from (#97831)

kit1980 committed 2 years ago

477f3f55

Update ufmt to v2.1.0 (#97900)

Skylion007 committed 2 years ago

22b72313

Add config setting to error on recompile (#97829)

mlazos committed 2 years ago

e626be79

Delete fusions_possible counter (#97881)

ezyang committed 2 years ago

bb40b625

[BE][autograd Function] Raise an error if input is returned as-is and saved for forward or backward in setup_context (#97212)

soulitzer committed 2 years ago

313db584

Revert "[dtensor] remove typing hack of DeviceMesh (#94526)"

pytorchmergebot committed 2 years ago

4114c1ea

Refactor dynamic dims api, stateless internals, higher level export API (#96699)

ezyang committed 2 years ago

8372c5dc

Remove comma from parametrized test name (#97844)

malfet committed 2 years ago

2c16b73a

[2/n] Consolidate `replicate` and `DDP`: split `forward` function (#96658)

yhcharles committed 2 years ago

44e73db3

[ONNX] Fix typo "scipt" -> "script" (#97850)

justinchuby committed 2 years ago

170a1c3a

[inductor] make `run_and_get_cpp_code` signature match `run_and_get_triton_code` (#97826)

nkaretnikov committed 2 years ago

2ce6ad9a

Fix typo when setting FSDP state dict config (#97110)

VikParuchuri committed 2 years ago

4ae4c6f6

inductor: fix vision_maskrcnn dynamic_shapes error on CPU (#97312)

chunyuan-w committed 2 years ago

004bb34f

prepare doc preview s3-prefix for future change (#97433)

pmeier committed 2 years ago

2f06fc24

[NNC] Fix the issue that the void** could not store a scalar if the bit width of the scalar is greater than 32bit on a 32bit platform (#97669)

EikanWang committed 2 years ago

faccd876

Avoid copies in matmul (no ghstack) (#97355)

lezcano committed 2 years ago

6871665a

Simplify by using yield from in torch/utils/data (#97839)

kit1980 committed 2 years ago

46faa79e

[Dynamo] torch.Generator state should have a source and be reconstructed properly (#97403)

yanboliang committed 2 years ago

f388bec9

Disable dynamo tracing torchrec.distributed (#97824)

yanboliang committed 2 years ago

9d1d9509

Add dynamo config skip_nnmodule_hook_guards (#97830)

wconstab committed 2 years ago

f4ac8e00

Remove rocm python 3.11 restriction (#97818)

atalman committed 2 years ago

91166ef7

Older