strisunshinewentingwang/pytorch

Pull Requests Commits

Update on "[c10d] Introduce ProcessGroupWrapper"

rohan-varma committed 3 years ago

38cb7e96

Update base for Update on "[c10d] Introduce ProcessGroupWrapper"

rohan-varma committed 3 years ago

ad40aebc

add channels last support for AvgPool2d on CPU (#48918)

mingfeima committed 3 years ago

8ac0917c

[RPC Framework] Add _script_module_reducer unconditionally for RecursiveScriptModule in RPC pickler (#58020)

Yi Wang committed 3 years ago

fd3d3ef9

[Static Runtime] Support clamp.Tensor (#58191)

Hao Lu committed 3 years ago

993a35a8

Update on "[c10d] Introduce ProcessGroupWrapper"

rohan-varma committed 3 years ago

0cba84eb

More stable and faster implementation of the gradient of torch.linalg.eigh (#55049)

lezcano committed 3 years ago

1f3807ce

Update on "[c10d] Introduce ProcessGroupWrapper"

rohan-varma committed 3 years ago

7900a0fe

Update internal code for torch.linalg.solve (#56613)

IvanYashchuk committed 3 years ago

b0833533

Gelu Backward, Contribution from Kevin Stephano (#58249)

Krovatkin committed 3 years ago

d304bb07

Print stderrs in tools/mypy_wrapper.py (#58265)

samestep committed 3 years ago

3a898c26

Migrate pytorch_python_doc_build to github action (#57371)

zhouzhuojie committed 3 years ago

7756cb6a

Only quicklint files that exist (#58261)

samestep committed 3 years ago

3f9126f3

Make norm and vector_norm use the same kernels. (#58214)

ngimel committed 3 years ago

f6532468

Disable more of quicklint test (#58257)

samestep committed 3 years ago

26aeec35

[PyTorch Mobile][Forward/backward compatibility] Number of arguments for operators (#56845)

iseeyuan committed 3 years ago

d833caaf

Reimplement spectral_norm using new parametrization functionality (#57784)

soulitzer committed 3 years ago

e1bb9d2d

[ONNX] Handle mixed mask, index input for index_put (#57604)

BowenBao committed 3 years ago

51cd89ec

[ONNX] ListUnpack on dynamic tensor list (#56592) (#57603)

BowenBao committed 3 years ago

01374d69

[ONNX] Handle NoneType in Assign Output Shapes (#54623) (#57602)

BowenBao committed 3 years ago

8e29863a

[ONNX] Process const folding progressively when converts to ONNX (#54569) (#57601)

BowenBao committed 3 years ago

bfe7728f

[ONNX] Support registering custom export for prim::PythonOp from torch.autograd.Function (#55630) (#57600)

BowenBao committed 3 years ago

346dc88b

Add support to to(device) op. (#56857) (#57599)

BowenBao committed 3 years ago

2b0f481d

onnx.symbolic_helper.parse_args: document and clean up (#56956) (#57598)

BowenBao committed 3 years ago

9e56314d

[ONNX] Special post process for onnx::Cast and onnx::ConstantOfShape shape type inference (#55962) (#57597)

BowenBao committed 3 years ago

dc0071df

Add a new operator for fill_() function. (#56859) (#57596)

BowenBao committed 3 years ago

ac9e79e5

[ONNX] Warning when using __len__ to calculate tensor shape (#55151) (#57595)

BowenBao committed 3 years ago

6d7fe763

[ONNX] Support .item() export & NumberType to tensor conversion (#55697) (#57594)

BowenBao committed 3 years ago

3bc8a226

Overwrite with `ln` if libc10.so already exists (#58243)

samestep committed 3 years ago

061c7a1e

update abs forward ad formula (#58235)

albanD committed 3 years ago

9b95568d

Newer Older