strisunshinewentingwang/pytorch

Pull Requests Commits

pin_memory malloc now uses existing context if available. (#22229)

ssnl committed 5 years ago

8482efb2

Update note about tensors on CPU for certain MAGMA functions, elimina… (#22618)

vishwakftw committed 5 years ago

054c7eb0

Fix torch.triu / torch.tril on contiguous tensors with non-default st… (#22730)

vishwakftw committed 5 years ago

f8ad65ad

For consistent CUDA_HOME behavior (#22845)

HaoTang@descartes committed 5 years ago

0ea8e61f

add benchmark for PT fill_ op (#22867)

mingzhe09088 committed 5 years ago

560d847d

remove RTTI check for TensorImpl shadow copy (#22773)

ljk53 committed 5 years ago

3b1c3996

Revert D16197605: [jit] Make traced fns also go into the global python CU

suo committed 5 years ago

c5afdd0b

Revert D16197608: [jit] Make classtypes hold a weak_ptr to their CU

Will Feng committed 5 years ago

a326aad8

Updating submodules

svcscm committed 5 years ago

5f05037d

add num_runs flag to the benchmark (#22892)

mingzhe09088 committed 5 years ago

94d99f25

Use original module's class name for ScriptModules (#22873)

davidriazati committed 5 years ago

6ffacd5f

remove stray print

Krovatkin committed 5 years ago

24833694

Add FakeQuantize Module (#21767)

jerryzh168 committed 5 years ago

f7de9be3

update README (#21312)

mingzhe09088 committed 5 years ago

0cddd3e7

Port SVD to ATen, enable batching for matrix inputs (#21588)

vishwakftw committed 5 years ago

7d055c21

Make classtypes hold a weak_ptr to their CU

suo committed 5 years ago

260b0e84

Make traced fns also go into the global python CU

suo committed 5 years ago

5fc1260e

_script_compile and _script_class_compile add to the python CU

suo committed 5 years ago

16aa235f

Close loophole to create untyped tuples (#22518)

smessmer committed 5 years ago

f2f80744

Deprecate untyped Lists (#22517)

smessmer committed 5 years ago

800f4936

Added .bfloat16() (#22852)

izdeby committed 5 years ago

bd88fd07

Set up CI with Azure Pipelines (#22839)

ethomson committed 5 years ago

8399197d

Back out "Back out "[pytorch][PR] Move thnvrtc and DynamicLibrary to ATen"" (#22794)

ezyang committed 5 years ago

535c5540

Remove tensor_data() call in Python Variable() and nn.Parameter() constructors (#22821)

Will Feng committed 5 years ago

317cf7c8

Make the signature of fill_out consistent with fill_.

xuhdev committed 5 years ago

14e8fb70

Move the body of fill_kernel_impl into fill_kernel_cuda

xuhdev committed 5 years ago

1c266c27

Move fill and fill_diagonal to Fill.cpp, Fill.h, and FillKernel.{cpp,cu}

xuhdev committed 5 years ago

fc297b8e

make_variable consumes the Tensor if it only has one reference

James Reed committed 5 years ago

815e73bc

Temporarily skip mypy-0.720 to unbreak master type checks

bddppq committed 5 years ago

b5fa9a34

lly-zero-one committed 5 years ago

1a93b968

Newer Older