strisunshinewentingwang/pytorch

Pull Requests Commits

Fix Trace inlining of graphs with optional inputs (#22686)

Elias Ellison committed 5 years ago

00df49c9

Add common_quantized test case utilities (#22694)

Lucaskabela committed 5 years ago

3e3e6ee3

Refactor and improve randperm tests.

xuhdev committed 5 years ago

7750cae7

Swap detection order in randperm_out_cuda to avoid unnecessary conversion from float when the input is small.

xuhdev committed 5 years ago

32709af8

Support Half type in randperm.

xuhdev committed 5 years ago

0f7c3710

Delegate Python ~ (invert operator) to Tensor.bitwise_not().

xuhdev committed 5 years ago

9c4c9c3a

Add a bitwise NOT operator for integer and Boolean types (CUDA).

xuhdev committed 5 years ago

574e8086

Add a bitwise NOT operator for integer and Boolean types (CPU).

xuhdev committed 5 years ago

e2dc1fc7

Refactoring _wrap_outputs to remove python dependence.

mal committed 5 years ago

58e20638

fix dce over loops

suo committed 5 years ago

ec1b669d

skip import nccl and gloo_gpu in cpu machine (#22522)

xw285cornell committed 5 years ago

9b8d7717

fix print (#22689)

jerryzh168 committed 5 years ago

b984b0ab

Enable more passes in ProfilingGraphExecutor

Krovatkin committed 5 years ago

f81395b3

Added Bfloat16 tensor for cpu with very limited support (#21860)

izdeby committed 5 years ago

10c60b60

keep reuqires_grad unchanged after converting bn to syncbn (#22569)

tianzhi0549 committed 5 years ago

6eb3969a

Revert D16161144: [pytorch][PR] Add traces to LowerGradOf and SpecializeAutoGrad

ezyang committed 5 years ago

cbb0b816

Enabled BFloat16 storage (#21523)

izdeby committed 5 years ago

3a8d7463

Updating submodules

svcscm committed 5 years ago

932ec8aa

Intoducing bfloat16 type (#21522)

izdeby committed 5 years ago

e72b617e

add forward declaration in stateful dataset (#22562)

xzhu1900 committed 5 years ago

de5a481c

Enable C2 operators running with {cpu, gpu} * {forward, backward} (#22664)

mingzhe09088 committed 5 years ago

3cf5f22f

change c2 bench to use new tensor creation interface (#22663)

mingzhe09088 committed 5 years ago

95a5da17

Add comments about adding new build options. (#22641)

xuhdev committed 5 years ago

e1fdf8a4

Properly formats errors rising up from C++ extension compilation (#22445)

andyljones committed 5 years ago

e2216ada

Add traces to LowerGradOf and SpecializeAutoGrad

Krovatkin committed 5 years ago

50901be9

Avoid potential extra copy in _lu_with_info_cuda (#22634)

ssnl committed 5 years ago

0c2cd93e

change unary, pool, max ops to use new interface (#22661)

mingzhe09088 committed 5 years ago

45aad2e6

introduce a new interface to add a list of operators (#21209)

mingzhe09088 committed 5 years ago

2b2fe525

fix lint (#22654)

jerryzh168 committed 5 years ago

16438815

Report errors through call stack (#22280)

davidriazati committed 5 years ago

8a233b99

Newer Older