strisunshinewentingwang/pytorch

Pull Requests Commits

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

README for the quantized op creation (#22165)

Zafar Takhirov committed 5 years ago

13d58fd9

Cleanup integer sign warnings

bwasti committed 5 years ago

dd4982e2

lu: When not using pivoting, return the identity permutation instead of zeros (#22242)

bamos committed 5 years ago

046c4589

Fix interpreter lines for files with python2-only syntax.

thatch committed 5 years ago

7fcfed19

torch.quantization conversion utilities, observers for eager mode quantization (#22010)

jerryzh168 committed 5 years ago

5040d52a

add GRAPH_UPDATE logging to guard_elimination.cpp

Krovatkin committed 5 years ago

073fa6f4

Performance improvements for depthwise convolutions in FP16 (#22302)

ptrblck committed 5 years ago

a3346e10

Move thnvrtc and DynamicLibrary to ATen (#22362)

ssnl committed 5 years ago

31d821e2

Fix typos in gradcheck error message

ssnl committed 5 years ago

74883d48

Sync worker requirement mismatches

Skory committed 5 years ago

92e8d040

Add a higher-level log traces to DCE

Krovatkin committed 5 years ago

9fb4386c

Typo fixed in documentation

shoaibahmed committed 5 years ago

5395db22

torchvision version changed to the latest one (#22598)

ifedan committed 5 years ago

b5860b55

use caffe2_dnnlowp_force_slow_path in FC (#22143)

jspark1105 committed 5 years ago

738aba17

fix uninitialized variable in BailOutInserter

Krovatkin committed 5 years ago

905b9a89

Adding FC and Relu QNNPACK ops to C10 registry (#22174)

supriyar committed 5 years ago

c97829d7

Convert VariableVersion counter to intrusive_ptr, saving a memory allocation on every Tensor.

resistor committed 5 years ago

0e7b65e4

Simplify the flow control in div_kernel_cuda. (#22555)

xuhdev committed 5 years ago

0c1ecf19

Add Module.requires_grad_ (#22576)

ssnl committed 5 years ago

478d480d

Update module.h (Fix a grammatical error of the comment in line 233) (#22548)

xz-liu committed 5 years ago

456d27df

Pass Variable into Caffe2 ops, by requiring that the Variable doesn't require grad (#22473)

Will Feng committed 5 years ago

3a125208

Enabled masked fill and scatter for bool tensors (#22491)

izdeby committed 5 years ago

304552b6

Fixed RNNImplBase reset and flat_weights methods to handle bidirectional flag correctly (#22493)

pbelevich committed 5 years ago

a48cf8f5

Add type stubs to import 'nn' modules (#22411)

malmaud committed 5 years ago

595e3447

Fix torch.normal with CUDA tensors (#22533)

vishwakftw committed 5 years ago

9bafe5d4

Newer Older