strisunshinewentingwang/pytorch

Pull Requests Commits

Remove hard numpy dependency introduced by inductor (#90796)

amjames committed 1 year ago

766ebf44

Properly guard all numpy usage within dynamo and remove UnspecializedNumpyVariable (#90795)

amjames committed 1 year ago

7cd951c2

[CI][ROCm] fix device visibility, again (#91813)

jeffdaily committed 1 year ago

f4494628

[JIT] Skip builtins while enumerating class methods (#91805)

malfet committed 1 year ago

4f1f14e3

Automatically convert real tensors to fake in dynamo export (#91742)

tugsbayasgalan committed 1 year ago

69acc340

make sure mutated args are iterated in the same order (#91792)

ngimel committed 1 year ago

ef495b7d

Revert "Deduplicate c10 error and PyTorchError hierarchy (#87855)"

pytorchmergebot committed 1 year ago

b3603f81

Return empty attention weights when need_atten_weights = False (#91782)

drisspg committed 1 year ago

f2199709

Add shape function for movedim op (#91696)

vivekkhandelwal1 committed 1 year ago

f77a9a58

Revert "[dynamo] Support dynamic slicing (#91341)"

pytorchmergebot committed 1 year ago

f556c5b9

Docs push fix .netrc sometimes a directory (#91745)

clee2000 committed 1 year ago

f4b3b577

[MPS] Fix the ChannelsLast memory format in cat_out_mps() (#91786)

razarmehr committed 1 year ago

87164ace

Preserve node's meta during fx.transformation (#90737)

SherlockNoMad committed 1 year ago

eeba9d5a

[dynamo] Support dynamic slicing (#91341)

tugsbayasgalan committed 1 year ago

8e7dcd14

[MPS] Remap the view ops to exisiting graph APIs. (#89436)

kulinseth committed 1 year ago

de99bc39

[functorch] test: try using reference_inputs in vmap tests (#91355)

kshitij12345 committed 1 year ago

2354ff5f

Add a NestedTensor Readme (#91472)

drisspg committed 1 year ago

eb8547e9

[Inductor] Support loop split at given depth in CPP codegen (#91397)

jgong5 committed 1 year ago

859ac58c

[inductor] fix output_stride of cat (#91233)

chunyuan-w committed 1 year ago

2555971b

[inductor] decompose tanh in CPP backend (#91687)

chunyuan-w committed 1 year ago

c99a2a43

Revert "[functorch] test: try using reference_inputs in vmap tests (#91355)"

pytorchmergebot committed 1 year ago

ad70a701

[functorch] test: try using reference_inputs in vmap tests (#91355)

kshitij12345 committed 1 year ago

a51090d4

Don't remove files across the whole OS on clean (#91503)

aostrowski-hbn committed 1 year ago

d0a4e2e7

[DTensor] fix test_device_mesh failure on GPU (#91783)

XilunWu committed 1 year ago

e3bd38d2

[ONNX] Support constant 'aten::__contains__' (#91660)

BowenBao committed 1 year ago

66745831

[MPS] Fix correctness issues with Pooling ops (#91519)

razarmehr committed 1 year ago

2f0e4839

inductor: Move graph.lint() in Intel's FX Passes to the End of Loop to Reduce Compile Time(part 2) (#91677)

XiaobingSuper committed 1 year ago

33547bb5

inductor:enable conv+unary fusion for torch unary function (#91609)

XiaobingSuper committed 1 year ago

25ff10ca

[cpu] implement erf based on oneDNN algorithm for aten::Vec (#91613)

Valentine233 committed 1 year ago

2175c941

[inductor] optimize lowering for empty-related operators (#91350)

Valentine233 committed 1 year ago

745dc3a1

Newer Older