strisunshinewentingwang/pytorch

Pull Requests Commits

fix compilation error with gcc 5.5 (#38112)

cloudhan committed 4 years ago

6e1e2a60

`unfold_backward` gets its own kernel (#36612)

nikitaved committed 4 years ago

a7c29dbf

[quant][graphmode][refactor] Split quantization.cpp (#37975)

jerryzh168 committed 4 years ago

0ed7fc58

[quant][graphmode][refactor] Remove unused code in quantization.cpp (#37974)

jerryzh168 committed 4 years ago

ff9a809c

Back out "Revert D20229168: [quantization] Use torchbind for Linear PackedParams" (#38101)

James Reed committed 4 years ago

c1e7758b

[TensorPipe] Do not require user to provide worker name-to-rank map (#38052)

lw committed 4 years ago

91f451a5

Don't use Profiler key in lite interpreter (#37962)

Ilia Cherniavskii committed 4 years ago

b4946b96

Replace torch.is_tensor usages with isinstance checks. (#38062)

rgommers committed 4 years ago

726aa713

remove uses of type() and type_as() part 1. (#38029)

Ailing Zhang committed 4 years ago

9232356e

Revert D21449612: [pytorch][PR] Migrate AT_DISPATCH_ALL_TYPES_AND_COMPLEX_AND3 to c10::complex

Mike Ruberry committed 4 years ago

0c936f94

[TensorExpr] Correctly print 'bool' dtype in Cuda printer. (#38077)

Mikhail Zolotukhin committed 4 years ago

0f60c8d8

[TensorExpr] Don't include prim::Constant nodes with Tensor type into TE fusion groups - we can't handle them. (#38105)

Mikhail Zolotukhin committed 4 years ago

ff1a627b

[TensorExpr] Properly handle Bool dtype in several other places. (#38104)

Mikhail Zolotukhin committed 4 years ago

a253ea92

[TensorExpr] Correctly print dtypes in Cast and Allocate. (#38091)

Mikhail Zolotukhin committed 4 years ago

459f14e9

[tensorboard] Let hparam render values correctly (#31544)

lanpa committed 4 years ago

609d5a44

Run QEMU to test that default dispatch doesn't use AVX (#38094)

malfet committed 4 years ago

4c358b8b

Add option to skip tests after retries (#38079)

malfet committed 4 years ago

53aa7d8b

Fix CUDA memory management issues caused by not using PinnedCPUAllocator (#38066)

Allan Di Wu committed 4 years ago

d35ab0b7

Migrate AT_DISPATCH_ALL_TYPES_AND_COMPLEX_AND3 to c10::complex (#37977)

zasdfgbnm committed 4 years ago

f4d9713d

Check the _geev input matrix for NaNs and infs (#37642)

Chris Paulse committed 4 years ago

deeef504

Fix typo: TupleUnpack. (#38043)

ezyang committed 4 years ago

16e3df3a

Fix ldflags string for HIPExtensions (#38047)

ashish committed 4 years ago

5a386a0a

Give _VariableFunctions class a different name, so pickling works (#38033)

ezyang committed 4 years ago

c2f787ce

Fix minor issue in type stub for Optimizer (#38067)

rgommers committed 4 years ago

9fe82435

Automatic update of fbcode/onnx to 807c62cf7e4c96ce49040bcf073b7e4a054f28a5 (#37983)

houseroad committed 4 years ago

3cade9cd

Remove static initalizers from Vec256 (#38088)

malfet committed 4 years ago

12bbda05

[c2][opt] nomnigraph transform for ClipRangesGatherSigridHashV2 fusion (#38004)

ajyu committed 4 years ago

25413635

[nomni] fix outputs check to replaceSubgraph (#38005)

ajyu committed 4 years ago

32329c33

Get rid of javasphinx dependency. (#38042)

ezyang committed 4 years ago

f8c93c5d

Revert D20229168: [quantization] Use torchbind for Linear PackedParams

malfet committed 4 years ago

4bc0a7f8

Newer Older