strisunshinewentingwang/pytorch

Pull Requests Commits

[Fakelowp] Open source fake fp16 FC ops (#37923)

Yinghai Lu committed 4 years ago

9143d7fb

Reland [quant][tests] Enable tests to run on all qengine backends (#37943)

supriyar committed 4 years ago

76c964df

[ONNX] Improve error checking for large model export (#37798)

BowenBao committed 4 years ago

122587dc

Report test stats (#37803)

malfet committed 4 years ago

385f7e59

Skip c2_ref_tests on network failures (#37972)

malfet committed 4 years ago

952e0f00

Add option to run python unittests in parallel (#37180)

malfet committed 4 years ago

72e5b7ae

Move complex utilities out of Half.h (#37676)

zasdfgbnm committed 4 years ago

681c6fb6

updated create input and add test methods and added a whitelist for complex (#37835)

anjali411 committed 4 years ago

63428211

Add missing c10::complex::value_type (#37677)

zasdfgbnm committed 4 years ago

14fc83eb

move quantization normalization layers to aten/src/ATen/native/quantized/cpu/ (#37352)

vkuzo committed 4 years ago

09bedec2

add quantized instancenorm operator (#36847)

vkuzo committed 4 years ago

4fa049c5

add quantized groupnorm operator (#36835)

vkuzo committed 4 years ago

b837d5d4

quant: remove hypothesis and int32 from layernorm test (#37947)

vkuzo committed 4 years ago

288dd337

add docker image build ubuntu16.04-cuda9.2-cudnn7-gcc5.4-py3.6 (#37610)

lixinyu committed 4 years ago

675e77e8

Give at::cuda::blas::gemv<at::Half> parity with <float> and <double>. Nature is healing. (#37569)

definitelynotmcarilli committed 4 years ago

35693e9b

[JIT] remove list_with_default op (#37886)

Elias Ellison committed 4 years ago

28ed04c6

Install HugePagesArena to optimize pytorch prediction performance (#37640)

Allan Di Wu committed 4 years ago

f538cd62

Update bazel to 3.1.0 (#37951)

malfet committed 4 years ago

3cc50625

[quant][fix] A typo in quantized::conv2d_relu (#37964)

jerryzh168 committed 4 years ago

56fc347e

Port existing zero_dim_dispatch optimizations from codegen and remove codegen capability. (#37615)

gchanan committed 4 years ago

f29f96d4

[JIT] Peephole optimize list ops (#37612)

Elias Ellison committed 4 years ago

f5b3125a

Migrate some CUDA arithmetic kernels to c10::complex (#37878)

zasdfgbnm committed 4 years ago

bf970bce

[jit][api][refactor] remove redundant deepcopy implementation (#37538)

jerryzh168 committed 4 years ago

4bbf889b

Do not `std::move` returned value (#37891)

malfet committed 4 years ago

cd0724f9

[reland][quant][graphmode] Support a new category of ops in graph mode quantization (#37936)

jerryzh168 committed 4 years ago

72818958

[TensorExpr] Support Bool dtype in Or, Xor, And ops and in TensorExprKernel::bindInput. (#37938)

Mikhail Zolotukhin committed 4 years ago

ec934252

[JIT] Rewrite unaliased if output mutation (#37694)

Elias Ellison committed 4 years ago

a3042ca8

[jit] normalize `getMethod` (#37472)

suo committed 4 years ago

b53e6bfd

fix profiling test (#37961)

Elias Ellison committed 4 years ago

28ac5cdc

Migrate cpu kernel for index and index_put to c10::complex (#37877)

zasdfgbnm committed 4 years ago

6293f1fb

Newer Older