Commits

Branch selector

 master

User selector

XiaobingSuper

Datepicker

All time

Commit History

Commits on Dec 2, 2022

• TorchDynamo: don't compute index for max_pooling when return_index is false (pytorch#89838)

  Show description for 42f27c3
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  42f27c3

  Copy full SHA for 42f27c3

• TorchDynamo: enable convolution bn folding for functional bn (pytorch#89746)

  Show description for b058a02
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  b058a02

  Copy full SHA for b058a02

• update quantization doc: add x86 backend as default backend of server inference (pytorch#86794)

  Show description for 8b2f988
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  8b2f988

  Copy full SHA for 8b2f988

Commits on Dec 1, 2022

• quantization: make x86 as default backend (pytorch#88799)

  Show description for 4bae860
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  4bae860

  Copy full SHA for 4bae860

• fix mkldnn quantization issue for weight reorder error (pytorch#86876)

  Show description for 0e7918b
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  0e7918b

  Copy full SHA for 0e7918b

Commits on Nov 25, 2022

• TorchDynamo: weight prepack for single conv (pytorch#89209)

  Show description for 3b3ebcd
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  3b3ebcd

  Copy full SHA for 3b3ebcd

• TorchDynamo: weight prepack for mkl linear (pytorch#89109)

  Show description for 0c4f3db
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  0c4f3db

  Copy full SHA for 0c4f3db

• TorchDynamo: weight prepack for onednn convolution external call (pytorch#88988)

  Show description for 07151a6
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  07151a6

  Copy full SHA for 07151a6

Commits on Nov 21, 2022

• TorchDynamo: enable conv+silu fusion (pytorch#89278)

  Show description for 31708a7
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  31708a7

  Copy full SHA for 31708a7

• TorchDynamo: enable conv+relu6 fusion (pytorch#89265)

  Show description for 79770d3
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  79770d3

  Copy full SHA for 79770d3

Commits on Nov 15, 2022

• Take input striding for conv fusion op based on eager output (pytorch#88864)

  Show description for 7a37bbe
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  7a37bbe

  Copy full SHA for 7a37bbe

Commits on Nov 14, 2022

• TorchDynamo: skip convolution fusion when convolution's padding is string (pytorch#88794)

  Show description for 9943d46
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  9943d46

  Copy full SHA for 9943d46

• Fake Tensor For (ConvFusion) Propagation (pytorch#88414)

  Show description for 15ef066
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  15ef066

  Copy full SHA for 15ef066

• TorchDynamo: Add convolution binary+unary fusion for cpu in inference mode (pytorch#88412)

  Show description for 072920c
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  072920c

  Copy full SHA for 072920c

• TorchDynamo: Add convolution binary(inplace) fusion for cpu in inference mode (pytorch#88403)

  Show description for 4ad7b17
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  4ad7b17

  Copy full SHA for 4ad7b17

Commits on Nov 10, 2022

• torchdynamo: add convolution add(relu) inplace fusion kernel (pytorch#88048)

  Show description for 3e43ff2
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  3e43ff2

  Copy full SHA for 3e43ff2

Commits on Nov 4, 2022

• TorchDynamo: enable convolution and batchnorm folding for inference path (pytorch#87435)

  Show description for b320626
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  b320626

  Copy full SHA for b320626

• TorchDynamo: Add linear binary fusion for cpu in BF16 inference mode (pytorch#87066)

  Show description for 71f793d
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  71f793d

  Copy full SHA for 71f793d

• TorchDynamo: Add linear unary fusion for cpu in BF16 inference mode (pytorch#87065)

  Show description for e4efea4
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  e4efea4

  Copy full SHA for e4efea4

• TorchDynamo: Add convolution binary fusion for cpu in inference mode (pytorch#87064)

  Show description for 5217318
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  5217318

  Copy full SHA for 5217318

Commits on Nov 3, 2022

• add parameters check for mkldnn_transpose (pytorch#85318)

  Show description for d70bc22
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  d70bc22

  Copy full SHA for d70bc22

Commits on Oct 27, 2022

• TorchDynamo: Add convolution unary fusion for cpu in inference mode (pytorch#87063)

  Show description for c36db82
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  c36db82

  Copy full SHA for c36db82

Commits on Oct 19, 2022

• [TorchDynamo]: fused bias for cpu convolution path (pytorch#87050)

  Show description for 232fbd9
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  232fbd9

  Copy full SHA for 232fbd9

Commits on Oct 15, 2022

• torchdynamo: add linear pointwise(binary) fusion kernel (pytorch#86583)

  Show description for 46aaae9
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  46aaae9

  Copy full SHA for 46aaae9

• torchdynamo: add convolution pointwise(binary) fusion kernel (pytorch#86582)

  Show description for 5210fab
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  5210fab

  Copy full SHA for 5210fab

• torchdynamo: add convolution pointwise(unary) fusion kernel (pytorch#86581)

  Show description for 9a7a49b
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  9a7a49b

  Copy full SHA for 9a7a49b

Commits on Sep 13, 2022

• disable onednn gelu for empty input (pytorch#84926)

  Show description for c61e895
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  c61e895

  Copy full SHA for c61e895

Commits on Aug 25, 2022

• add qscheme check for quantization observer (pytorch#80126)

  Show description for 31f1517
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  31f1517

  Copy full SHA for 31f1517

• fix oneDNN channels_last path issue (pytorch#83653)

  Show description for a013597
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  a013597

  Copy full SHA for a013597

Commits on Aug 23, 2022

• fix upsample bf16 issue for channels last path by using high pricsion to compute index (pytorch#83847)

  Show description for 658f958
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  658f958

  Copy full SHA for 658f958

Commits on Aug 1, 2022

• directly init a zero immediate buffer to reduce overhead for batch_norm cpu path (pytorch#82558)

  Show description for d08157d
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  d08157d

  Copy full SHA for d08157d

Commits on Jul 17, 2022

• fix torch.config can't respect USE_MKLDNN flag issue (pytorch#75001)

  Show description for 86b8620
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  86b8620

  Copy full SHA for 86b8620

Commits on Apr 20, 2022

• clarify hardtanh's definition

  Show description for 27014a8
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  27014a8

  Copy full SHA for 27014a8

Commits on Mar 17, 2022

• report an error if num_channels is not divisible by num_groups for nn.GroupNorm

  Show description for 4a0f6e6
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  4a0f6e6

  Copy full SHA for 4a0f6e6

Commits on Mar 14, 2022

• remove redundant index check for index_select_out_cpu_dim1_ (pytorch#74093)

  Show description for f810d96
  

  XiaobingSuper

  authored and

  pytorchmergebot

  committed
  f810d96

  Copy full SHA for f810d96

Pagination

Previous

Next