strisunshinewentingwang/pytorch

Pull Requests Commits

[JIT] namedtuple constants (#32873)

Elias Ellison committed 4 years ago

fe01376f

Quantized sigmoid function

z-a-f committed 4 years ago

fbe121e3

Solves Issue #32750 - torch.prod now works fine with FP16 Input Tensor and FP32 Output Tensor (#32831)

krshrimali committed 4 years ago

7b65acdf

Don't serialize None values in observer (#32733)

jerryzh168 committed 4 years ago

8ddd5bb0

Remove wrap_dim from codegen layer. (#32738)

gchanan committed 4 years ago

1760d5b8

Code cleaning: Some iterating variables in builtin_functions.cpp can be const (#32852)

xuhdev committed 4 years ago

660a93c5

[pytorch] avoid `thread_local std::vector<Call>` for mobile build (#32849)

ljk53 committed 4 years ago

ada966b7

Loops.cuh legacy code cleanup -- gpu_kernel_with_index (#32777)

zasdfgbnm committed 4 years ago

d9e99ab5

Updating submodules

svcscm committed 4 years ago

fd3bd777

Coding header is better specified in lowercase letters (#32850)

xuhdev committed 4 years ago

b16dab8a

Updating submodules

svcscm committed 4 years ago

22466552

Updating submodules

svcscm committed 4 years ago

ed10408c

Make save_for_lite_interpreter private (#32771)

Martin Yuan committed 4 years ago

03557a98

Add knobs to set the number of profiling runs and bailout depth (#32735)

Krovatkin committed 4 years ago

c3b4bfcf

Remove Python dependency (toPyTuple/fromPyTuple, jitCompilationUnit, deserialize) in rref_impl.h/cpp (#32753)

xush6528 committed 4 years ago

12bcfa7c

make tests for empty inputs check zero parameter grads (#32820)

ngimel committed 4 years ago

29fabb1f

Update Docs for building PyTorch for Android.

hovhannesgithub committed 4 years ago

bc2e05a3

Remove needs_dynamic_casting from TensorIterator and move it to Loops.cuh (#32755)

zasdfgbnm committed 4 years ago

fcf9fced

Cudnn bn size fix (#32763)

ptrblck committed 4 years ago

0f097205

[PyTorch BC] Fix the ci (#32843)

houseroad committed 4 years ago

bcb7c226

Updating submodules

svcscm committed 4 years ago

5380e16d

[torch] fd error check

gaurav1086 committed 4 years ago

765904f1

Resubmit more code fakefp16 mapping unification (#32798)

Yinghai Lu committed 4 years ago

94ddc2c4

Centralize addition of "always on" dispatch keys. (#32734)

ezyang committed 4 years ago

690d41f2

Make DispatchKeyGuards accept DispatchKey::Undefined (#32729)

ezyang committed 4 years ago

5ddd2cd9

Rename DispatchKey::UndefinedTensorId to Undefined (#32728)

ezyang committed 4 years ago

3d0a470d

Remove GIL from RRefContext (#32807)

mrshenli committed 4 years ago

a40a19cc

Fixes moving after weight norm application (#32563)

mruberry committed 4 years ago

413c0f6c

Make python version a parameterizable option for Windows CI.

peterjc123 committed 4 years ago

9bab617b

Fix backcompat for linear_relu_dynamic_fp16 (#32803)

James Reed committed 4 years ago

cc35c876

Newer Older