strisunshinewentingwang/pytorch

Pull Requests Commits

Enabling concat fast path for channels last inputs (#39448)

kimishpatel committed 4 years ago

6a421d50

fixes to layernorm emulation (#40422)

Hector Yuen committed 4 years ago

27982d57

Increase shapes column length (#40440)

Ilia Cherniavskii committed 4 years ago

b82bd654

Destroy CUDA events after profiling (#39962)

Ilia Cherniavskii committed 4 years ago

d8c38454

Fix demangle 't' issue in profiler (#40416)

Ilia Cherniavskii committed 4 years ago

a54bb4e9

Make custom_fwd a no-op when not executed under autocast (#36171)

definitelynotmcarilli committed 4 years ago

3b040c47

[jit] Enable `copy.deepcopy` and `copy.copy` for RecursiveScriptModule (#32685)

jerryzh168 committed 4 years ago

f652abc1

quant docs: add and clean up ELU (#40377)

vkuzo committed 4 years ago

9bf25557

quant docs: add and clean up BatchNorm{n}d (#40346)

vkuzo committed 4 years ago

d71ec51c

quant docs: add and clean up InstanceNorm{n}d (#40345)

vkuzo committed 4 years ago

5e683517

quant docs: add and clean up GroupNorm (#40343)

vkuzo committed 4 years ago

6e3fdd77

quant docs: add and clean up LayerNorm (#40342)

vkuzo committed 4 years ago

d15fcc7e

quant docs: add and clean up hardtanh (#40341)

vkuzo committed 4 years ago

d27f8eaf

quant docs: add and clean up hardsigmoid (#40340)

vkuzo committed 4 years ago

8e74fb6a

quant docs: clean up hardswish (#40323)

vkuzo committed 4 years ago

c4594a97

Simplify complex case for `div_cpu` (#39996)

ShawnZhong committed 4 years ago

79736ff9

Testcppextensionjit rebuild once (#40169)

nolifenameless committed 4 years ago

3e6fa778

Add basic GPU support to distributed autograd. (#40312)

pritamdamania committed 4 years ago

54c05fa3

Set C++14 compatibility flag in torch_compile_options (#40399)

malfet committed 4 years ago

e509c58a

Revert D22124313: Use Int8QuantParamsBlob to pass the scale and zeropoint params

lw committed 4 years ago

2acee6dc

[Codemod][FBSourceGoogleJavaFormatLinter] Daily `arc lint --take GOOGLEJAVAFORMAT`

generatedunixname89002005287564 committed 4 years ago

08ae7d3a

Use Int8QuantParamsBlob to pass the scale and zeropoint params (#40390)

csummersea committed 4 years ago

1ec4337b

[TensorPipe] Register multiplexing channel over UV (#40389)

lw committed 4 years ago

78b3d5f8

[ROCM][CI] Skip fp16 bench and 2-GPU runs (#40243)

xw285cornell committed 4 years ago

e9efad68

[quant][graphmode][refactor] InsertQuantDeQuantHelper (#40384)

jerryzh168 committed 4 years ago

ba89a893

Revert D22165477: [pytorch][PR] [JIT] Fork/Join inline docs

zhaojuanmao committed 4 years ago

6c40ec55

Fix Cuda IPC deadlock (#40347)

VitalyFedyunin committed 4 years ago

7bf1dd58

[quant][graphmode] Add warning for debug option for add_scalar/mul_scalar (#40383)

jerryzh168 committed 4 years ago

18122fac

Device name should be a string, not bytes (#40322)

malfet committed 4 years ago

5766da50

Add note to torch.save (#40394)

James Reed committed 4 years ago

0d24ed0c

Older