strisunshinewentingwang/pytorch

Pull Requests Commits

Updating submodules

svcscm committed 4 years ago

0be2f12e

Fix tracing for dynamic quantized LSTM (#29331)

James Reed committed 4 years ago

821f8bfc

Updating submodules

svcscm committed 4 years ago

5bb35fe9

Skip flaky test

James Reed committed 4 years ago

1dd3c8e5

Use cuDNN's handle pool mechanism to manage cublas handles (#29233)

zasdfgbnm committed 4 years ago

02921e79

explicitly provide memory format when calling to clone() at SparseTensor.cpp

ifedan committed 4 years ago

b008b34b

explicitly provide memory format when calling to clone() at SparseTensorMath.cpp

ifedan committed 4 years ago

09822a1d

Automatic update of fbcode/onnx to fea8568cac61a482ed208748fdc0e1a8e47f62f5 (#29363)

houseroad committed 4 years ago

564384fe

Updating submodules

svcscm committed 4 years ago

255505f2

explicitly provide memory format when calling to clone() at TensorShape.cpp

ifedan committed 4 years ago

d5d524da

NHWC support in cuDNN BatchNorm & Conv2d (#29361)

jjsjann123 committed 4 years ago

fdab1cf0

Add unsafeRemoveAttr and unsafeRemoveSlot to ivalue::Object (#29048)

jerryzh168 committed 4 years ago

0aba5ba1

CPU-strided-complex support for ComplexFloat (#29294)

dylanbespalko committed 4 years ago

abbe6347

Make PyTorch Python 3.8 compatible (#29302)

vishwakftw committed 4 years ago

86c64440

Move unboxed dispatch decision into dispatcher (#29200)

smessmer committed 4 years ago

ca20b569

explicitly provide memory format when calling to clone() at rprop.py

ifedan committed 4 years ago

43d4d019

AsyncIf op implementation

Yavuz Yetim committed 4 years ago

2704af09

Handle warning in torchscript (#27154)

albanD committed 4 years ago

b14c5943

add pybind version of HANDLE_TH_ERRORS

albanD committed 4 years ago

0ff1696c

Buffer python warning to avoid deadlocks

albanD committed 4 years ago

9b875e12

Fix clang-tidy errors in csrc/Module.cpp

albanD committed 4 years ago

cb3232fd

Allow setting tolerations in testing math functions. (#29297)

xuhdev committed 4 years ago

528a0cfb

explicitly provide memory format when calling to clone() at lbfgs.py

ifedan committed 4 years ago

b05e9d45

Fix the issue when NHWC Tensor has height or width larger then max cuda grid (#28931)

ifedan committed 4 years ago

5d70b11d

explicitly provide memory format when calling to clone() at parameter.py

ifedan committed 4 years ago

4926a510

Add some non-contiguous tests

nairbv committed 4 years ago

8498a155

explicitly provide memory format when calling to clone() at batchnorm.py

ifedan committed 4 years ago

9dcf5191

explicitly provide memory format when calling to clone() at Indexing.cpp

ifedan committed 4 years ago

75309b45

Revert D18350353: dump operator names of a module and its sub-modules.

pietern committed 4 years ago

78a34d32

fix @property (#28395)

suo committed 4 years ago

58005382

Newer Older