strisunshinewentingwang/pytorch

Pull Requests Commits

try to make at::cat in mm_tree_reduction operate on contig tensors (#18816)

Natalia Gimelshein committed 5 years ago

3875e1ba

Fix the insert_guard for norm decomposation (#19646)

wanchaol committed 5 years ago

c5719691

add max_pool2d to AD, add tests for both autodiff and inference mode

wanchaol committed 5 years ago

3c81eb3a

Fix the depthwise 3x3x3 fast path criteria for the stride (#19692)

csummersea committed 5 years ago

cbd0a2d3

use relative path to load libthnvrtc (#19690)

Natalia Gimelshein committed 5 years ago

614871d9

Change some comments related to moving copy_ to native (#19618)

Roy Li committed 5 years ago

72b8b6c3

Remove old complex Types (#19616)

Roy Li committed 5 years ago

17e4cd0c

Remove function variant of copy_ (#19622)

Roy Li committed 5 years ago

6fead42e

Restore copy_ overload with async arg (#19641)

Roy Li committed 5 years ago

a6811e17

Add some of nn.init to weak script (#19640)

davidriazati committed 5 years ago

c08f3d06

Support serializing std::vector<torch::Tensor> (#19677)

Will Feng committed 5 years ago

9aa0e607

Fix fuser tests on sandcastle (#19684)

James Reed committed 5 years ago

32174bed

Add LONG_BINPUT to unpickler (#19696)

David Riazati committed 5 years ago

3d6e9564

improve err msg (#19645)

Elias Ellison committed 5 years ago

62447a5a

Enable assignment for QTensor in pytorch frontend (#19676)

jerryzh168 committed 5 years ago

6ec55c13

Make torch.equal work with boolean CPU tensors

asuhan committed 5 years ago

4a65ee95

Add torch.from_file function similar to the Storage.from_file, but returning tensor (#18688)

VitalyFedyunin committed 5 years ago

d14abe3a

Add no-gpu build mode for all of PyTorch and Caffe2

Dmytro Dzhulgakov committed 5 years ago

d247912d

Caffe2 shouldn't fail if CUDA peer access is already enabled

David Goodwin committed 5 years ago

c855e04d

Support exporting squeeze & unsqueeze with negative dim attribute

BowenBao committed 5 years ago

96051300

Remove useless input shape checker in conv (#19608)

gujinghui committed 5 years ago

b675f07b

Make it possible for self.forward to return a ScriptMethod (#19217)

zdevito committed 5 years ago

87a69741

Add if ops support for onnxifi and ssa-rewrite (#19585)

zrphercule2 committed 5 years ago

2f73b3d2

GCC ABI variants for nightly builds (#18888)

kostmo committed 5 years ago

41486306

Fix no SIGCHLD checking in DataLoaderIter._shutdown_workers (#19421)

ssnl committed 5 years ago

5e62ee2b

Revert D15008160: Enable assignment for QTensor in pytorch frontend

ezyang committed 5 years ago

c42f3f90

fix rocm test (#19663)

Elias Ellison committed 5 years ago

84b275b7

Enforce consistent dict iteration order for trace inputs. (#19528)

efaust committed 5 years ago

8273b9b3

Enable assignment for QTensor in pytorch frontend (#19530)

jerryzh168 committed 5 years ago

309c15e2

Dont introduce aliasing in CSE or Constant Pooling (#19576)

eellison committed 5 years ago

d902774c

Older