strisunshinewentingwang/pytorch

Pull Requests Commits

[NNC] Read/Write Dependency analysis (#46952)

nickgg committed 3 years ago

0edc6a39

Fix pickling for Tensor subclasses. (#47115)

hameerabbasi committed 3 years ago

c4209f11

Add torch.overrides checks for submodules. (#47285)

hameerabbasi committed 3 years ago

60ae8475

Benchmark combining Distributed Data Parallel and Distributed RPC (#46993)

BorisValkov5 committed 3 years ago

6c5a1c50

[TensorExpr] Run constant pooling in fusion groups to dedupe constants. (#47402)

Mikhail Zolotukhin committed 3 years ago

ca293ec4

add partition_by_partition_cost (#47280)

scottxu0730 committed 3 years ago

5107a411

[ONNX] Add export of prim::data (#45747)

KsenijaS committed 3 years ago

878032d3

[quant][graphmode][fx][test] Add test for nn.Sequential (#47411)

jerryzh168 committed 3 years ago

192b2967

Validate number of GPUs in distributed_test. (#47259)

pritamdamania committed 3 years ago

c8872051

Make `torch.det()` support complex input. (#45980)

nikitaved committed 3 years ago

8a3728c8

Expand the test of torch.bmm on CUDA (#47124)

zasdfgbnm committed 3 years ago

030caa19

Split IGamma cuda kernel into it's own file to speed up compilation times. (#47401)

gchanan committed 3 years ago

32c76dbe

[DI] Allow explicit taskLauncher for torchscript interpreter (#46865)

Gaoxiang Liu committed 3 years ago

735f8cc6

[FX] Speed up non-parameter tensor lookup (#47325)

James Reed committed 3 years ago

b704cbef

Clean up some imports in cuda kernel code. (#47400)

gchanan committed 3 years ago

ff3e1de6

Fix collect_env when pytorch is not installed (#47398)

malfet committed 3 years ago

848901f2

Split up BinaryMiscOpKernels.cu because it's slow to compile. (#47362)

gchanan committed 3 years ago

da491d75

[NNC] Fix an issue in Cuda fusion with fp16 scalar vars coerced to float (#47229)

nickgg committed 3 years ago

33cf7fdd

Add tests for DDP control flow models. (#47206)

rohan-varma committed 3 years ago

31c9d2ef

Add `input` argument to `autograd.backward()` cpp api (#47214)

soulitzer committed 3 years ago

2e5bfa98

Skip iomp5 emebedding if torch_cpu could not be found (#47390)

malfet committed 3 years ago

6f602518

[Pytorch] Add basic c10::optional tests (#47014)

swolchok committed 3 years ago

ae706378

[pytorch] Remove c10::nullopt_t::init (#47013)

swolchok committed 3 years ago

17be8ae1

[JIT] add freeze to docs (#47120)

Elias Ellison committed 3 years ago

7ab843e7

Expand GRADIENT_IMPLEMENTED_FOR_COMPLEX to allow named tensors (#47289)

jonasteuwen committed 3 years ago

a11bc049

Add vulkan reshape op (hack) (#47252)

SS-JIA committed 3 years ago

5d82311f

[Gradient Compression] Export sizes, along with length and offset of each variable to GradBucket for PowerSGD (#47203)

Yi Wang committed 3 years ago

6b3802a7

Renamed a TensorListMetaData property. Cleaned up a test (#46662)

Iurii Zdebskyi committed 3 years ago

2c554266

[quant][graphmode][fx] Test to make sure dequantize node are placed properly (#47332)

jerryzh168 committed 3 years ago

f588ad6a

Revert D24481801: Optimize backward for torch.repeat

ejguan committed 3 years ago

bba5a311

Newer Older