strisunshinewentingwang/pytorch

Pull Requests Commits

Add ONNX export support for torch.rand. (#20559)

Spandan Tiwari committed 5 years ago

22865d4c

clean code to unify the logic to use fp16 by the optimizer engine (#20915)

Xing Wang committed 5 years ago

7d84ca6e

change test_name to be globally unique value across tests (#21206)

mingzhe09088 committed 5 years ago

3004b397

introduce a new intrace to add op [PT changes] (#21149)

mingzhe09088 committed 5 years ago

ca80ec7c

don't materialize constants (#21229)

Elias Ellison committed 5 years ago

88d033f8

Improve ONNX Loop export (#20445)

BowenBao committed 5 years ago

9a41f447

Renaming member variables in engine.cpp/h (#21283)

mal committed 5 years ago

4980b8b9

Rename FC to Linear for the function name (#21268)

jianyuh committed 5 years ago

37fed9b2

Replace AT_ASSERTM with TORCH_CHECK (#21267)

jianyuh committed 5 years ago

63b3c5a6

Improve performance of advanced indexing backward (#20557)

Natalia Gimelshein committed 5 years ago

ad971a37

file:line for tracing (#21247)

James Reed committed 5 years ago

4ac732ed

Export ONNX Dropout for opset 10 (#20710)

lara-hdr committed 5 years ago

27d1daab

math module support: isnan, asinh, atanh, cosh, sinh, and tanh (#19337)

lhendre committed 5 years ago

770089c2

Remove onnx export expects (#21238)

Elias Ellison committed 5 years ago

fb726252

add contiguous function type hint for tensor (#21285)

shihongzhi committed 5 years ago

2e59a0a6

Write add_scalars data in the same file (#21100)

nataliakliushkina committed 5 years ago

96667dfe

Fix build error in c10 on Windows (#21005)

peterjc123 committed 5 years ago

5b336987

Move THCTensor_{normal, normal_means, normal_stddevs, normal_means_stddevs} to ATen (#21287)

syed-ahmed committed 5 years ago

155f7673

EliminateGuards

Krovatkin committed 5 years ago

21113c2d

Make is_contiguous checks generic in number of arguments (#21106)

colesbury committed 5 years ago

c8539be9

Significantly simplify the spawning of pytorch libs building process. (#21105)

xuhdev committed 5 years ago

b159e0ce

Retry Fix Python DataParallel RNN in no_grad mode (#21262)

mrshenli committed 5 years ago

f62a0060

Fix gelu documents (#21265)

xiaomengy committed 5 years ago

0c6efbd4

Add autograd for layer_norm on CPU (#20883)

xiaomengy committed 5 years ago

eaa3ba65

Add gelu gradient for pytorch (#21237)

xiaomengy committed 5 years ago

31c79b71

Add gelu activation in pytorch (#20665)

xiaomengy committed 5 years ago

93ae040f

Revert D15577342: [pytorch][PR] Fix Python DataParallel RNN in no_grad mode

Karl Ostmo committed 5 years ago

aac424a6

Fixes a bug in the test (#21146)

Zafar Takhirov committed 5 years ago

360e6d1b

Exclude file:line from graphs used for fuser kernel cache (#21252)

James Reed committed 5 years ago

62ae348d

Save the weight shape info the first time we have chance to extract it (#21233)

Yinghai Lu committed 5 years ago

7c40576c

Newer Older