Commits

Branch selector

 intel_scnn

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 16, 2018

• adding explicit type cast parameters to valid make_pair(...)

  

  wenyuan95

  committed
  4566adf

  Copy full SHA for 4566adf

Commits on Jan 8, 2018

• Update README.md

  jspark1105

  authored
  49ac6f6

  Copy full SHA for 49ac6f6

Commits on Oct 24, 2017

• Update README.md

  jspark1105

  authored
  c71a7ca

  Copy full SHA for c71a7ca

Commits on Jul 13, 2017

• add description on pruning

  

  Jongsoo Park

  committed
  3cf6efa

  Copy full SHA for 3cf6efa

Commits on Jun 30, 2017

• add neccesary files to reproduce resnet pruning results

  

  Jongsoo Park

  committed
  b6f4576

  Copy full SHA for b6f4576

Commits on Jun 23, 2017

• add Python scripts for accuracy-sparsity plots

  

  Jongsoo Park

  committed
  e72ebe9

  Copy full SHA for e72ebe9

Commits on Jun 14, 2017

• don't use Fontran when compiling libxsmm to remove dependency to Fortran compilers

  

  Jongsoo Park

  committed
  c34a194

  Copy full SHA for c34a194

Commits on Jun 12, 2017

• remove unused functions

  

  Jongsoo Park

  committed
  142d7c4

  Copy full SHA for 142d7c4

Commits on Jun 8, 2017

• add solver files used for googlenet pruning

  

  Jongsoo Park

  committed
  66f9ed1

  Copy full SHA for 66f9ed1

Commits on Jun 2, 2017

• make libxsmm easier

  

  Jongsoo Park

  committed
  44d9157

  Copy full SHA for 44d9157

• make libxsmm easier

  

  Jongsoo Park

  committed
  a43cb62

  Copy full SHA for a43cb62

Commits on May 31, 2017

• remove deprecated winograd_sparsity_factor option

  

  Jongsoo Park

  committed
  b8d1e84

  Copy full SHA for b8d1e84

Commits on May 24, 2017

• resolve undefined symbol error when compiling in ubuntu 16.04

  

  Jongsoo Park

  committed
  aa555b3

  Copy full SHA for aa555b3

• Merge branch 'intel_sparse_winograd' into intel_scnn

  Show description for cf658d5
  

  Jongsoo Park

  committed
  cf658d5

  Copy full SHA for cf658d5

• include omp.h

  

  Jongsoo Park

  committed
  462720a

  Copy full SHA for 462720a

• no longer support old cuda version

  

  Jongsoo Park

  committed
  c615f23

  Copy full SHA for c615f23

• support cudnnv6

  

  Jongsoo Park

  committed
  366b981

  Copy full SHA for 366b981

Commits on May 20, 2017

• add resnet pruning results

  

  Jongsoo Park

  committed
  bde8f8f

  Copy full SHA for bde8f8f

Commits on May 11, 2017

• add compile option for SKX; use new sconv function in sconv_test

  

  Jongsoo Park

  committed
  089608b

  Copy full SHA for 089608b

Commits on Apr 19, 2017

• support cudnnv6

  

  Jongsoo Park

  committed
  331c2a5

  Copy full SHA for 331c2a5

Commits on Mar 29, 2017

• Merge branch 'intel_scnn' of github.com:IntelLabs/SkimCaffe into intel_scnn

  

  Jongsoo Park

  committed
  75b60a8

  Copy full SHA for 75b60a8

• bug fix in ConvolutionReLU layer when DIRECT_SCONV is not used

  

  Jongsoo Park

  committed
  c72d898

  Copy full SHA for c72d898

• avoid nested parallelism in mkl

  

  Jongsoo Park

  committed
  4fd876e

  Copy full SHA for 4fd876e

• add sparsity information of pruned models of AlexNet and GoogLeNet

  

  Jongsoo Park

  committed
  3df975a

  Copy full SHA for 3df975a

• add sparsity information of pruned models of AlexNet and GoogLeNet

  

  Jongsoo Park

  committed
  56482bb

  Copy full SHA for 56482bb

• more description on architecture dependent options

  

  Jongsoo Park

  committed
  6f11bf7

  Copy full SHA for 6f11bf7

• add SKL option

  

  Jongsoo Park

  committed
  63a3219

  Copy full SHA for 63a3219

• remove fft layers for now to reduce dependency to MKL

  

  Jongsoo Park

  committed
  20b0dce

  Copy full SHA for 20b0dce

• add paper references

  

  Jongsoo Park

  committed
  b72048f

  Copy full SHA for b72048f

• add prototxt for testing with direct sparse convolution and MKL with batch size 1

  

  Jongsoo Park

  committed
  3725aff

  Copy full SHA for 3725aff

• remove fft layers for now to reduce dependency to MKL

  

  Jongsoo Park

  committed
  fb8114a

  Copy full SHA for fb8114a

• patch spmdm in libxsmm for less memory usage

  

  Jongsoo Park

  committed
  cc0faf7

  Copy full SHA for cc0faf7

• enable pruning in Winograd domain from scratch

  

  Jongsoo Park

  committed
  7f41033

  Copy full SHA for 7f41033

Commits on Feb 17, 2017

• option to dump sparsity histogram

  

  Jongsoo Park

  committed
  8124e72

  Copy full SHA for 8124e72

Commits on Feb 16, 2017

• get rid of warning

  

  Jongsoo Park

  committed
  4898f6f

  Copy full SHA for 4898f6f

Pagination

Previous

Next