strisunshinewentingwang/pytorch

Pull Requests Commits

Port torch.outer method_tests() to OpInfos (#54798)

lithuak committed 3 years ago

9e6877c5

[testing] support op with args/kwargs in test_unary_ufunc (#52194)

kshitij12345 committed 3 years ago

b7c5d575

enable bf16 for cat serial kernel (#54674)

haozhe.zhu committed 3 years ago

07350da3

enable bf16 vec copy (#54671)

haozhe.zhu committed 3 years ago

01b15570

[numpy] Add torch.take_along_dim (#52833)

kshitij12345 committed 3 years ago

0527d142

Make index_add take a scalar argument alpha (#54176)

zasdfgbnm committed 3 years ago

eec48303

optimizer exploration - v1 and v2 + fix position_weighted optimizer + decoupled weight decay (#54042)

lanlanfb committed 3 years ago

695eef05

[DDP] Mark a few variables as const in reducer (#54764)

rohan-varma committed 3 years ago

5c3d80d8

[c10d] s/torch::autograd::variable/at::Tensor/g (#54763)

rohan-varma committed 3 years ago

671f80a3

Implement public API InferenceMode and its error handling (#53343)

Ailing Zhang committed 3 years ago

7caa4646

Compute Tensor::toString() without reference to backend (#54711)

ezyang committed 3 years ago

23091731

Make memory overlap a little less precise so it works with null data ptr (#54710)

ezyang committed 3 years ago

f0679725

Make the fuser raise NotImplementedError when unknown device is hit (#54709)

ezyang committed 3 years ago

c782949e

Avoid testing device in cdist when called in a "Math" context (#54708)

ezyang committed 3 years ago

6445c9a1

Make convolution_overrideable default implementation raise NotImplementedError (#54707)

ezyang committed 3 years ago

c9e0aab2

Disambiguate where 'Doesn't run' error message comes from (#54706)

ezyang committed 3 years ago

ed560cf2

Fix missing format string qualifier (#54705)

ezyang committed 3 years ago

b5ab3482

Rename linalg.det test so that it generates a valid method name (#54704)

ezyang committed 3 years ago

d9a7c758

Add empty_generic, which allocates an empty tensor in a device-generic way (#54703)

ezyang committed 3 years ago

05fa570b

Fix some more native_functions.yaml mistakes (#54597)

ezyang committed 3 years ago

90e70ace

Nasty little hack to preserve NotImplementedError raised in interpreter (#54627)

ezyang committed 3 years ago

e70f3d11

More types for torch (#54037)

r-barnes committed 3 years ago

e5634f5f

Add complex autograd support for `torch.unfold` (#52999)

iramazanli committed 3 years ago

d59fb7a2

[codemod] fix tautological imports

rahulg committed 3 years ago

6eaf9696

Enable faulthandler for distributed tests. (#54531)

pritamdamania committed 3 years ago

65781f94

Move requires_grad_/backward out of VariableTypeManual. (#54543)

Ailing Zhang committed 3 years ago

1d5cc6c5

Add JIT support for cmath unary ops (#54089)

anjali411 committed 3 years ago

d63dd07f

Improve error message while setting error twice. (#54464)

pritamdamania committed 3 years ago

8eb896ce

Add 'remote_parameters' and 'get_module_rref' to RemoteModule docs. (#54645)

pritamdamania committed 3 years ago

f612d4eb

[test_c10d] Add wait in nccl high priority stream test (#54714)

jaglinux committed 3 years ago

316804e3

Older