strisunshinewentingwang/pytorch

Pull Requests Commits

Fix binary bitwise converter for when lhs is scalar (#72101)

kflu committed 2 years ago

8f2e7bba

DispatchKeySet perf improvements (#70364)

bdhirsh committed 2 years ago

18cbe80f

free up dispatch key space (in C++) (#69633)

bdhirsh committed 2 years ago

20b8653d

[fx][graph opts] port FoldLayerNormArithmetic from glow to FX (#69715)

alexbeloi committed 2 years ago

1cec7194

Document forward AD interaction with grad mode (#72216)

soulitzer committed 2 years ago

99d490e9

Revert D33992798: Add TORCH_CUDA_CU_API to CUDABlas functions

janeyx99 committed 2 years ago

d832d93e

Add TORCH_CUDA_CU_API to CUDABlas functions (#72305)

malfet committed 2 years ago

07e9f86a

Revert D33994546: [Quant][fx][improvement] Added test for quint4x2 for fx graph mode quantization (reland PR 69846)

malfet committed 2 years ago

cd5ed549

Remove forcing CUDNN_STATIC when CAFFE2_STATIC_LINK_CUDA (#72290)

atalman committed 2 years ago

1e7d20ea

[Quant][fx][improvement] Added test for quint4x2 for fx graph mode quantization (reland PR 69846) (#72278)

dzdang committed 2 years ago

a5dad85c

[Quant][Eager][improvement] Added 4 bit support for eager mode quantization flow (reland PR 69806) (#72277)

dzdang committed 2 years ago

ab1e88e3

[Quant][improvement] Added 4 bit support for embedding quantized module (reland PR 69769) (#72276)

dzdang committed 2 years ago

bfdf45cc

[AutoAccept][Codemod][FBSourceClangFormatLinter] Daily `arc lint --take CLANGFORMAT`

CodemodService FBSourceClangFormatLinterBot committed 2 years ago

e81bfffb

[DataPipe] Extend FileLister to support load multiple directories (#72260)

ejguan committed 2 years ago

6297aa11

Update torch flatbuffer usage to OSS version (#71957)

qihqi committed 2 years ago

03260f85

[FX] Fix default argument handling for Interpreter (#72272)

James Reed committed 2 years ago

3f6643e6

Sparse: Use per-operator headers (#71115)

peterbell10 committed 2 years ago

dbf09bc0

Add support for include-what-you-use (#71114)

peterbell10 committed 2 years ago

e90f5586

Remove CUDA Foreach... files dependency on function operators (#68462)

peterbell10 committed 2 years ago

36385bee

[quant][gpu] Adding quantized conv operator in cudnn (#70622)

jerryzh168 committed 2 years ago

3c266620

Combine install miniconda routes in Windows GHA

janeyx99 committed 2 years ago

aa04c8ed

Make ShardedTensor ctor more inline with torch.Tensor ctor (#72164)

pritamdamania committed 2 years ago

8c505bbc

Add `norm` to LazyTensor codegen (#72119)

ansley committed 2 years ago

Verified bc484561

[NNC] Use index for stride mapping in kernel.cpp (#72266)

Elias Ellison committed 2 years ago

defde3bb

remove alwayslink/link_whole from //c10 library (#70997)

mikey dagitses committed 2 years ago

9d4d782e

Bump torch version to 1.12 (#72221)

atalman committed 2 years ago

0ca0e026

remove //c10:headers dependency from //c10 (#70996)

mikey dagitses committed 2 years ago

e970160c

[DataPipe] Fix OOM when traverse IterDataPipe due to pickling (#72209)

ejguan committed 2 years ago

4fc6ab5e

move //c10/cuda/test to shared build structure (#71429)

mikey dagitses committed 2 years ago

90458004

[PyTorch] MHA: remove TODO about transposing second arg to linear() (#72229)

swolchok committed 2 years ago

2d4513e4

Newer Older