strisunshinewentingwang/pytorch

Pull Requests Commits

Change to.dtype_layout to c10-full (#41169)

smessmer committed 4 years ago

9daba76b

Reducing size of docker Linux image (#41207)

Steve Sylvain committed 4 years ago

7c143e5d

Improve repr for torch.iinfo & torch.finfo (#40488)

Kiyosora committed 4 years ago

0651887e

Migrate addmm, addbmm and THBlas_gemm to ATen (#40927)

peterbell10 committed 4 years ago

cb6c3526

Self binning histogram (#40875)

Yavuz Yetim committed 4 years ago

16c8146d

[ONNX]Fix export of flatten (#40418)

KsenijaS committed 4 years ago

9b0393fc

add check for duplicated op registration in JIT (#41214)

linbinyu committed 4 years ago

a548c6b1

Wrap Caffe2's SparseLengthsSum into a PyTorch op (#39596)

nqjade committed 4 years ago

75b6dd3d

Small clarification of torch.cuda.amp multi-model example (#41203)

definitelynotmcarilli committed 4 years ago

d927aee3

LogitOp LUT based fake FP16 Op. (#41258)

venkatacrc committed 4 years ago

4a09501f

Modularize parsing NCCL_BLOCKING_WAIT in ProcessGroupNCCL (#41076)

osalpekar committed 4 years ago

33f9fbf8

Autograd Doc for Complex Numbers (#41012)

anjali411 committed 4 years ago

db38487e

test nan and inf in TestTorchMathOps (#41225)

ngimel committed 4 years ago

e568b3fa

[caffe2] Add the dedup implementation of fused RowWiseAdagrad op on GPUs (#40282)

jianyuh committed 4 years ago

62e16934

Benchmark RecordFunction overhead on some models (#40952)

Ilia Cherniavskii committed 4 years ago

08227072

Add add_relu fusion pass to optimize_for_mobile. (#40252)

kimishpatel committed 4 years ago

8a79eec9

Added SiLU activation function (#41034)

heitorschueroff committed 4 years ago

75a4862f

Expose private APIs to enable/disable pickling ScriptModules without RPC (#39631)

mrshenli committed 4 years ago

f6eb92a3

[ROCm] Skip caffe2 unique op test for rocm3.5 (#41219)

rohithkrn committed 4 years ago

df252c05

make Int8 FC bias quantization use round flush to infinity

Hector Yuen committed 4 years ago

a79b4168

Revert D22458928: [pytorch][PR] Use explicit templates in CUDALoops kernels

malfet committed 4 years ago

7c2c752e

JIT pass for add relu fusion. (#39343)

kimishpatel committed 4 years ago

c5dcf056

Add fused add_relu op. (#39342)

kimishpatel committed 4 years ago

82c9f79e

[Vec256][neon] Add neon backend for vec256 (#39341)

kimishpatel committed 4 years ago

d6feb614

Add benchmark for add op. (#40059)

kimishpatel committed 4 years ago

bddba1e3

[RPC docs] Remove mention of TensorPipe's SHM and CMA backends as they're not built (#41200)

lw committed 4 years ago

dde3d5f4

restore old documentation references (#39086)

mattip committed 4 years ago

a88099ba

Preserve CUDA gencode flags (#41173)

malfet committed 4 years ago

b952eaf6

Use explicit templates in CUDALoops kernels (#41059)

malfet committed 4 years ago

e3742807

Revert D21870844: Create lazy_dyndeps to avoid caffe2 import costs.

malfet committed 4 years ago

1f135148

Newer Older