strisunshinewentingwang/pytorch

Pull Requests Commits

[dynamo] Save global torch state to restore on graph break

anijain2305 committed 2 years ago

925c2e7a

Revert "PyTorch -> C++17 (#98209) (#100557)"

pytorchmergebot committed 2 years ago

da02ccc6

Turn on anomaly detection for AOTAutograd backward tracing (#101047)

ezyang committed 2 years ago

2621fbda

simplify sdpa backward meta registration (#101128)

Natalia Gimelshein committed 2 years ago

15a51e20

[vision hash update] update the pinned vision hash (#101142)

pytorchmergebot committed 2 years ago

5f89d89a

[Dynamo] Fix nested function resume execution (#100426)

yanboliang committed 2 years ago

075d36d3

benchmarks: make --amp works for cpu path (#101057)

XiaobingSuper committed 2 years ago

c84627c2

[dtensor] tensor ops to use strategy based sharding prop (#100607)

wanchaol committed 2 years ago

a1aa32e2

Run C++ test_api binary directly in CI slow jobs (#101088)

huydhn committed 2 years ago

d1f0c8e2

[c10d] Figure out device to use for object collectives (#100954)

kwen2501 committed 2 years ago

0848ed21

Restore Vulkan tests to periodic (#101026)

huydhn committed 2 years ago

a0e6ae2c

Move periodic dynamo benchmarks to inductor workflow (#100915)

clee2000 committed 2 years ago

13d445c2

inductor(CPU): fix masked_fill issue when filled value is nan (#101058)

XiaobingSuper committed 2 years ago

b1a8a10a

Revert "Fix header inclusions in c10 by iwyu (#100304)"

pytorchmergebot committed 2 years ago

3271413e

[efficiency_camp] Vector Realloc Optimize caffe2::BinaryElementwiseWithArgsOp::DoRunWithType (#100631)

annwesh committed 2 years ago

bb7d9886

Enable reordering pass (#100747)

eellison committed 2 years ago

7110060c

Revert "Enable reordering pass (#100747)"

pytorchmergebot committed 2 years ago

91ca9a27

Revert "inductor(CPU): fix issue when padding/stride/dilation size is one for cpu weight packing pass (#100951)"

pytorchmergebot committed 2 years ago

6c3af6a9

[MPS] Export check for 13.3 to `is_macos13_or_newer` (#101119)

malfet committed 2 years ago

176dabf8

[pt2] Add some helper function for SymIntVector (#101056)

xw285cornell committed 2 years ago

c650b12e

[Dynamo] Fix torch.{cuda/cpu}.amp.autocast arguments binding bug (#101052)

yanboliang committed 2 years ago

8a20ea0a

Delete Python-2 checks from setup.py (#101112)

malfet committed 2 years ago

08ef92e7

Preserve PyTest Cache across job runs (#100522)

ZainRizvi committed 2 years ago

96f46316

[ROCm]Fixed ut test_memory_timeline (#96752)

BLOrange-AMD committed 2 years ago

2dc93c20

Allow cmake vars in docker build (#100867)

kcarlson committed 2 years ago

e762cce6

[reland][quant][pt2e] Change input act annotation to a map and allow dynamic quantization for non zeroth argument (#101005) (#101041)

jerryzh168 committed 2 years ago

058d740f

CUDA Graph Trees - error on deallocated access (#100927)

eellison committed 2 years ago

3941bbc5

[inductor] Make codecache file permissions less restrictive (#100870)

peterbell10 committed 2 years ago

3b2a93a3

[CI] Run test_multi_gpu in test_inductor_distributed (#100135)

desertfire committed 2 years ago

32c9e7d3

Allow C++ custom class to define __repr__ and use it from Python (#100724)

Luthaf committed 2 years ago

000368b0

Older