strisunshinewentingwang/pytorch

Pull Requests Commits

Minor fixs to make torchbench runable on torch/xla (#107919)

JackCaoG committed 1 year ago

e73ec92a

Remove fixed skips (#108674)

eellison committed 1 year ago

518cfda2

Try to use linux.arm64.2xlarge runners (#107672)

DanilBaibak committed 1 year ago

e6042db0

Use linux.24xlarge for conda linux nightly builds (#108695)

atalman committed 1 year ago

cd6a332b

[export] Change _generate_new_graph_signature (#108571)

angelayi committed 1 year ago

d856f3b4

Fix inductor `sub` with symbolic integers. (#108518)

ysiraichi committed 1 year ago

089950b8

add packaging to requirements.txt (#108554)

FFFrog committed 1 year ago

3f74e57e

Enable mypy checking in torch/_inductor/sizevars.py (#107862)

jennychristopher committed 1 year ago

8a76f8e6

[quant][pt2e] Support int16 quantization (#108453)

jerryzh168 committed 1 year ago

32a16d49

[PT2 Inference] Prototype of Inference Runtime (#108482)

SherlockNoMad committed 1 year ago

bee7e781

Revert "Force synced KJT to trace unbacked SymInt (#107788)" (#108684)

huydhn committed 1 year ago

5a4fe05a

Revert "[export] Fix dict.get() to dict.setdefault() for param lookup. (#108587)"

pytorchmergebot committed 1 year ago

1aacbaed

Revert "Use global variables to register the return_types namedtuples (#107000)"

pytorchmergebot committed 1 year ago

27d5dcf5

Revert "docs: Match open bracket with close bracket in unsqueeze (#95215)"

pytorchmergebot committed 1 year ago

e5e653a6

Fix extension rebuilding on Linux (#108613)

QuarticCat committed 1 year ago

20812d69

Improve triton bsr_dense_mm performance on column-major ordered inputs with float32 dtype (#108512)

pearu committed 1 year ago

4e042cfe

Add TORCH_API to expose RPC module functions for RPC module device extension (#108553)

fakeYan committed 1 year ago

1dabfb68

Enable mypy checking in torch/_inductor/__init__.py (#108336)

jennychristopher committed 1 year ago

e471c12a

Torchbench model tolerance changes (#108598)

eellison committed 1 year ago

738106c1

[Doc] Move Dynamo IPEX backend to training/inference category (#108643)

jingxu10 committed 1 year ago

aa89f0a1

Fix empty vector segfault during version parsing in quantized serialization (#108418)

apach301 committed 1 year ago

79bc4eeb

[sdpa decomp] change sdpa decomp to be consistent with flash attention (#108608)

kimishpatel committed 1 year ago

ebed490c

Fix copy=True behavior for torch.asarray when device is not None/cpu (#108511)

FFFrog committed 1 year ago

6edd0644

Back out "Faster gc_count update for CUDACachingAllocator" (#108632)

Yinghai Lu committed 1 year ago

aebb86fe

Inductor cpp wrapper: fix codegen of positional args with default value (#108552)

chunyuan-w committed 1 year ago

ca9f4222

[inductor] Move AOTInductor runtime headers (#108564)

desertfire committed 1 year ago

60bd30ee

[MPS] Pixel shuffle unshuffle support (#99306)

alexdremov committed 1 year ago

b60273b8

[DeviceMesh] Minor docstring update for init_device_mesh and rename variables (#108391)

wz337 committed 1 year ago

ca2cdb30

[PyTorch] Add the lazy init call for p2p access function (#1991) (#108589)

zdevito committed 1 year ago

3fe84176

[DTensor] Replace usage of compute_local_offset by compute_local_shape_and_global_offset (#108547)

wz337 committed 1 year ago

49aa8d19

Newer Older