Commits

Branch selector

 master

User selector

Xia-Weiwen

Datepicker

All time

Commit History

Commits on Aug 27, 2023

• [Quant][Inductor] add UT of dequant promotion for linear (pytorch#106935)

  Show description for 4b27e46
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  4b27e46

  Copy full SHA for 4b27e46

• [Quant][Inductor] Enable quantization linear pattern fusion inside inductor (pytorch#106934)

  Show description for f3adbab
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  f3adbab

  Copy full SHA for f3adbab

• [Quant][Inductor] Enable qlinear weight prepack inside inductor constant folding (pytorch#106782)

  Show description for 15ceafb
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  15ceafb

  Copy full SHA for 15ceafb

• [Quant][PT2E] Enable linear and linear-unary post-op quant recipe for x86 inductor quantizer (pytorch#106781)

  Show description for e9b0f62
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  e9b0f62

  Copy full SHA for e9b0f62

• [Quant] Add int8 linear op impl for quantization PT2E with Inductor. input is an int8 CPU tensor; weight is an int8 MdkldnnCPU tensor. (pytorch#105818)

  Show description for a6d3da1
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  a6d3da1

  Copy full SHA for a6d3da1

Commits on Aug 25, 2023

• [ONEDNN][BC-breaking] update onednn from v2.7.3 to v3.1.1 (pytorch#97957)

  Show description for 97a291f
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  97a291f

  Copy full SHA for 97a291f

Commits on Jun 1, 2023

• [Quant][PT2E][Inductor] Lower quantized conv to Inductor (pytorch#101164)

  Show description for ce9923a
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  ce9923a

  Copy full SHA for ce9923a

Commits on Mar 31, 2023

• [Quant][FX] Add test case for lowering conv_transpose with kwargs (pytorch#97311)

  Show description for e073979
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  e073979

  Copy full SHA for e073979

• [Quant][FX] lower functional conv_transpose ops (pytorch#97126)

  Show description for e61b842
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  e61b842

  Copy full SHA for e61b842

Commits on Mar 28, 2023

• [Quant][FX] lower ConvTranspose3d (pytorch#97125)

  Show description for 08766b2
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  08766b2

  Copy full SHA for 08766b2

Commits on Mar 21, 2023

• [Quant][FX] Fix issue of lowering weighted functional ops with kwargs (pytorch#95865)

  Show description for e8be6d8
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  e8be6d8

  Copy full SHA for e8be6d8

Commits on Mar 3, 2023

• [Quant][PT2.0] fix issues for rearranging weight observer for decomposed linear (pytorch#94296)

  Show description for f3c25cd
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  f3c25cd

  Copy full SHA for f3c25cd

Commits on Feb 1, 2023

• [Quant] X86 qengine always uses fbgemm kernels on OS other than Linux (pytorch#93218)

  Show description for f77f88f
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  f77f88f

  Copy full SHA for f77f88f

• [Quant][ONEDNN] Fix weight reorder issue for grouped convolution (pytorch#91934)

  Show description for eea752f
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  eea752f

  Copy full SHA for eea752f

Commits on Jan 27, 2023

• [FX][Quant] Enable FX quant for patterns like x.view(x.size(...), ...) (pytorch#90001)

  Show description for 6fa84fd
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  6fa84fd

  Copy full SHA for 6fa84fd

Commits on Jan 26, 2023

• [Quant][Fx] Fix issue: qconfig_mappings of onednn backend are not correctly set for fused modules (pytorch#91297)

  Show description for 1d03a6a
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  1d03a6a

  Copy full SHA for 1d03a6a

Commits on Jan 18, 2023

• [Quant][Eager] Copy MHA's batch_first attribute in prepare() (pytorch#91680)

  Show description for 61a7618
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  61a7618

  Copy full SHA for 61a7618

• [Quant] onednn backend switch to ideep new api without affacting performance (pytorch#91056)

  Show description for 5a2ae88
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  5a2ae88

  Copy full SHA for 5a2ae88

Commits on Jan 12, 2023

• [Quant] make x86 the default quantization backend (qengine) (pytorch#91235)

  Show description for e1bcbbf
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  e1bcbbf

  Copy full SHA for e1bcbbf

• [MKLDNN] Rename pooling_avg to pooling_avg_exclude_padding (pytorch#90247)

  Show description for 3613ff0
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  3613ff0

  Copy full SHA for 3613ff0

Commits on Jan 6, 2023

• [Meta] Register aten.pixel_shuffle.default for meta (pytorch#91605)

  Show description for de9c82f
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  de9c82f

  Copy full SHA for de9c82f

Commits on Dec 20, 2022

• [Quant] lower fused LinearTanh for onednn backend (pytorch#89188)

  Show description for a5eb564
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  a5eb564

  Copy full SHA for a5eb564

Commits on Dec 19, 2022

• [Quant] Add fused LinearTanh module for onednn backend (pytorch#88923)

  Show description for 6686e9b
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  6686e9b

  Copy full SHA for 6686e9b

• [Quant] Add fused linear-tanh op for onednn backend (pytorch#88879)

  Show description for ea49e76
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  ea49e76

  Copy full SHA for ea49e76

• [Quant][FX] Lower QLinearLeakyReLU for onednn backend (pytorch#88668)

  Show description for 9ca41a9
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  9ca41a9

  Copy full SHA for 9ca41a9

Commits on Dec 17, 2022

• [Quant][FX] Add backend config for onednn backend and fuse Linear-LeakyReLU (pytorch#88665)

  Show description for 7b0ec67
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  7b0ec67

  Copy full SHA for 7b0ec67

Commits on Dec 16, 2022

• [Quant] Add fused LinearLeakyReLU module for onednn backend (pytorch#88661)

  Show description for 6ea93b2
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  6ea93b2

  Copy full SHA for 6ea93b2

• [Quant] Add fused LinearLeakyReLU module for onednn backend (pytorch#88661)

  Show description for 353c2e7
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  353c2e7

  Copy full SHA for 353c2e7

Commits on Dec 15, 2022

• [Quant] onednn backend switch to ideep new api without affacting performance (pytorch#90354)

  Show description for 9b89ff0
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  9b89ff0

  Copy full SHA for 9b89ff0

Commits on Dec 6, 2022

• [Quant] Add fused linear-leaky_relu op for onednn backend (pytorch#88478)

  Show description for 97e47a5
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  97e47a5

  Copy full SHA for 97e47a5

Commits on Nov 9, 2022

• [Quant] Support lowering of channel shuffle in FX (pytorch#83731)

  Show description for f6192b7
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  f6192b7

  Copy full SHA for f6192b7

Commits on Oct 28, 2022

• [MKLDNN] Replace pooling algorithm pooling_avg with pooling_avg_exclude_padding for future oneDNN upgrades (pytorch#87851)

  Show description for 07f7c46
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  07f7c46

  Copy full SHA for 07f7c46

Commits on Sep 30, 2022

• [Quant] Make x86 backend default when querying qconfig (pytorch#85461)

  Show description for 4b86a93
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  4b86a93

  Copy full SHA for 4b86a93

Commits on Sep 29, 2022

• [Quant] Add unified x86 quant backend (pytorch#84329)

  Show description for 3a3e200
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  3a3e200

  Copy full SHA for 3a3e200

Commits on Sep 27, 2022

• [UT] Fix random failure of test_qconv_transpose1d by skip using hypothesis (pytorch#85463)

  Show description for f1f6cb0
  

  Xia-Weiwen

  authored and

  pytorchmergebot

  committed
  f1f6cb0

  Copy full SHA for f1f6cb0

Pagination

Previous

Next