strisunshinewentingwang/pytorch

Pull Requests Commits

removed half support for cpu impl and updated tensor_dim_apply3

anjali411 committed 4 years ago

8f58d2ea

Revert "removed half support for cpu impl and updated tensor_dim_apply3"

anjali411 committed 4 years ago

779865ed

removed half support for cpu impl and updated tensor_dim_apply3

anjali411 committed 4 years ago

70d3e7e8

cuda function correctly handles nan and passes correct initi vals for floating and integer types, also added tests

anjali411 committed 4 years ago

bb0c15a7

fixed cpu function to handle nan

anjali411 committed 4 years ago

9061383c

anjali411 committed 4 years ago

bb558504

templated indices type for cpu function

anjali411 committed 4 years ago

a22a9d72

cuda function for cummax, cummin

anjali411 committed 4 years ago

5624c2dc

cuda code update

anjali411 committed 4 years ago

f69e0dfc

changed the macro to function

anjali411 committed 4 years ago

34102ef5

added TENSOR_DIM_APPLY3 to ATen, working cpu code for cummax, cummin, added dim check to the macro

anjali411 committed 4 years ago

e30b5bbc

removed parallel_for and added helper

anjali411 committed 4 years ago

7250a68a

cuda function -- dispatch

anjali411 committed 4 years ago

89475764

cummax cummin cpu functions

anjali411 committed 4 years ago

d6f987c0

updated cpu fn for cummax and cummin

anjali411 committed 4 years ago

2fc319ea

updated docs and other minor changes

anjali411 committed 4 years ago

86968f20

Updating submodules

svcscm committed 4 years ago

87dc2dbc

[caffe2] make order btw div and mul in adgrad consistent (#32974)

jspark1105 committed 4 years ago

c57f8984

Updating submodules

svcscm committed 4 years ago

d2999737

Revert D19871946: [distributed] pass in timeout to TCP store when initializing

rohan-varma committed 4 years ago

d4e4bedd

[distributed] pass in timeout to TCP store when initializing (#33325)

rohan-varma committed 4 years ago

df47a3ab

Move gating part of SparseFeatureGating to local

huayuli00 committed 4 years ago

c75d06d8

[BC] Temporarily fix the BC check (#33387)

houseroad committed 4 years ago

f6808df7

Fix index truncation in argmin/max for large tensors (#33310)

peterbell10 committed 4 years ago

495bd581

Get rid of some template arguments in GPU loop (#33308)

zasdfgbnm committed 4 years ago

cd038c0a

Use torch.set_default_dtype in test_data_parallel and rename dtype2prec (#32962)

pritamdamania committed 4 years ago

fd684cc3

Revert D19900566: [pytorch][PR] Simplify prim::shape when we have complete tensor types.

houseroad committed 4 years ago

6dd6b0bf

Add support for aten::slice to guard elimination. (#33311)

Owen Anderson committed 4 years ago

d35a4c20

Updating submodules

svcscm committed 4 years ago

c37a9b87

fast setup for output tensor in tensor iterator (#33165)

lixinyu committed 4 years ago

1e76649d

Older