strisunshinewentingwang/pytorch

Pull Requests Commits

Open source fbgemm fp16 pack op (#36791)

Yinghai Lu committed 4 years ago

b245b1d2

Open source fakefp16 BatchMatMul op (#36789)

Yinghai Lu committed 4 years ago

1b1a6a90

Use hypothesis to to control the rand seed (#36717)

yinghai committed 4 years ago

b08494eb

Avoid printing erroneous warning about "MIOpen not found" for ROCm builds (#33837)

jithunnair-amd committed 4 years ago

dc1f9eee

Print GPU info for ROCm test runs (#36827)

jithunnair-amd committed 4 years ago

6963973d

Back out "Vectorize reduction when reducing on fastest striding dimension" (#36854)

ppwwyyxx committed 4 years ago

a64ea8ea

Print all test output while running unit tests in bazel (#36825)

malfet committed 4 years ago

681158e2

[tensorboard] Add strings to image boxes (#30941)

lanpa committed 4 years ago

f767de60

Add build_variable.bzl to CMAKE_RERUN target (#36809)

malfet committed 4 years ago

4668d47d

Python binding api to optimize for mobile model on script module. (#36357)

xcheng16 committed 4 years ago

86f354c5

[pytorch] move prim::TupleIndex from register_prim_ops_fulljit to register_prim_ops (#36808)

kmatzen committed 4 years ago

fac076a8

[RELAND] Port to quantized and other operators to new registration API (#36799)

ezyang committed 4 years ago

0a8a0120

Move DICT ops to lite interpreter (#36816)

linbinyu committed 4 years ago

d7608c7f

[Format] format a few files (#35187)

idning committed 4 years ago

cc5befc4

Reduce binary size of schema inference (#34735)

smessmer committed 4 years ago

54a1e850

[tensorboard] Fix function input parameter for add_hparams (#31301)

lanpa committed 4 years ago

3a400b8d

Vectorize reduction when reducing on fastest striding dimension (#36709)

zasdfgbnm committed 4 years ago

d92005ff

Amp gradient accumulation example (#36601)

definitelynotmcarilli committed 4 years ago

e6bc34f5

Fix hardsigmoid/hardswish for proper device dispatch. (#36704)

Ailing Zhang committed 4 years ago

adca88e8

[ROCm] Use float datatype for RNN test for MIOpen (#36772)

ashish committed 4 years ago

9df9aef9

Handle log_sigmoid(out=) properly. (#36736)

gchanan committed 4 years ago

4c666d42

Print keyword-only arg symbol for function signature suggestions. (#36780)

gchanan committed 4 years ago

46288465

Fix incorrect merge of #34136. (#36760)

gchanan committed 4 years ago

ebdc4f02

Make trivial thread-idx for degenerate statements without thread-idx. (#36480)

zheng-xq committed 4 years ago

32bbf12a

Improve aten::backward handling (#36750)

smessmer committed 4 years ago

31f91d64

Fixes clang format (#36787)

Mike Ruberry committed 4 years ago

b45b9673

Move unboxing for factory ops to after dispatch (#36564)

smessmer committed 4 years ago

a89d1ed5

[pytorch][PR] Re-enable a failing test (#36763)

Krovatkin committed 4 years ago

b5483b82

Revert D21080503: [pytorch][PR] [quant] Add backward compatiblity test

Mike Ruberry committed 4 years ago

f00014b7

Add a test to verify non-contiguous tensors work correctly with RPC. (#36705)

pritamdamania committed 4 years ago

b0227f29

Newer Older