Commits

Branch selector

 master

User selector

fmassa

Datepicker

All time

Commit History

Commits on Feb 8, 2016

• Fix VolumetricConvolution for C/Cuda interoperability

  fmassa

  committed
  0db32cf

  Copy full SHA for 0db32cf

Commits on Jan 31, 2016

• Unify C/Cuda signatures for SpatialConvolutionMM and Spatial(AdaptiveMax,Max)Pooling

  Show description for dd9fabb
  fmassa

  committed
  dd9fabb

  Copy full SHA for dd9fabb

Commits on Jan 27, 2016

• Add THNN conversion of {Spatial(AdaptiveMax,Average,Max)Pooling}

  fmassa

  committed
  e2ad1f3

  Copy full SHA for e2ad1f3

• Move {Spatial(Adaptive,Average,Max)Pooling.c} to lib/THNN/generic

  fmassa

  committed
  af15ab0

  Copy full SHA for af15ab0

• Add functional convertion of SpatialConvolutionMM

  fmassa

  committed
  2836584

  Copy full SHA for 2836584

• Move SpatialConvolutionMM.c -> lib/THNN/generic

  fmassa

  committed
  fde83d8

  Copy full SHA for fde83d8

Commits on Jan 11, 2016

• Improve error message in SpatialConvolutionMM

  Show description for 3ad9597
  fmassa

  committed
  3ad9597

  Copy full SHA for 3ad9597

Commits on Jan 8, 2016

• Fix batch mode in MarginRankingCriterion

  fmassa

  committed
  605e252

  Copy full SHA for 605e252

Commits on Dec 25, 2015

• Merge pull request #3 from szagoruyko/cross_lrn

  Show description for a4b42a7
  fmassa

  committed
  a4b42a7

  Copy full SHA for a4b42a7

Commits on Dec 14, 2015

• SpatialAveragePooling supports padding, ceil mode and exclude_pad division

  Show description for 738057e
  fmassa

  committed
  738057e

  Copy full SHA for 738057e

Commits on Dec 1, 2015

• Fix type() in Cosine

  fmassa

  committed
  1a8dc44

  Copy full SHA for 1a8dc44

Commits on Nov 16, 2015

• Add tensorCache to Min and Max type method

  fmassa

  committed
  1ab8c88

  Copy full SHA for 1ab8c88

• Normalize supports infinity (maximum) norm

  fmassa

  committed
  84b9248

  Copy full SHA for 84b9248

• Do not change state variables in CosineDistance/CosineEmbeddingCriterion

  fmassa

  committed
  fb2a351

  Copy full SHA for fb2a351

Commits on Oct 24, 2015

• Fix possible division by zero in Normalize

  Show description for 08195f3
  fmassa

  committed
  08195f3

  Copy full SHA for 08195f3

Commits on Oct 20, 2015

• Add note on training shared models

  fmassa

  committed
  d2817dc

  Copy full SHA for d2817dc

Commits on Oct 11, 2015

• Speedup and reduce memory usage in Normalize

  Show description for d09b5a8
  fmassa

  committed
  d09b5a8

  Copy full SHA for d09b5a8

Commits on Oct 6, 2015

• Remove CDiv from doc

  Show description for d060a81
  fmassa

  committed
  d060a81

  Copy full SHA for d060a81

Commits on Oct 5, 2015

• Fix type() in SpatialConvolution

  fmassa

  committed
  b3fc443

  Copy full SHA for b3fc443

• Remove unnecessary mallocs from BCECriterion

  fmassa

  committed
  1d14dcd

  Copy full SHA for 1d14dcd

Commits on Oct 3, 2015

• Adds SpatialCrossLRN

  fmassa

  committed
  b7ce170

  Copy full SHA for b7ce170

Commits on Oct 2, 2015

• Fix error message in SpatialConvolutionMM.c

  fmassa

  committed
  934c3b1

  Copy full SHA for 934c3b1

Commits on Sep 27, 2015

• Adds SpatialDeconvolution

  fmassa

  committed
  9b51443

  Copy full SHA for 9b51443

Commits on Sep 12, 2015

• Adds Normalize

  fmassa

  committed
  b80bda2

  Copy full SHA for b80bda2

Commits on Sep 7, 2015

• Fix non-contiguous test in SpatialConvolutionMM

  fmassa

  committed
  c13a5fb

  Copy full SHA for c13a5fb

Commits on Aug 30, 2015

• CosineEmbeddingCriterion supports batch mode

  fmassa

  committed
  ea97488

  Copy full SHA for ea97488

Commits on Jul 12, 2015

• Optimizations plus bugfix

  fmassa

  committed
  fcb300c

  Copy full SHA for fcb300c

Commits on Jul 8, 2015

• SpatialMaxPooling supports padding and ceil mode

  Show description for 929cfc5
  fmassa

  committed
  929cfc5

  Copy full SHA for 929cfc5

Commits on Jun 21, 2015

• CosineDistance supports batch mode

  Show description for 94f4b62
  fmassa

  committed
  94f4b62

  Copy full SHA for 94f4b62

Commits on Jun 19, 2015

• Assert output size in SpatialConvolutionMM + doc fix

  Show description for d4b9c48
  fmassa

  committed
  d4b9c48

  Copy full SHA for d4b9c48

• Fix print in SpatialConvolution and correct non-contiguous test in SpatialConvolutionMM

  fmassa

  committed
  cc37251

  Copy full SHA for cc37251

Commits on Jun 17, 2015

• SpatialConvolutionMM supports padW/padH != 1

  Show description for 793b6bf
  fmassa

  committed
  793b6bf

  Copy full SHA for 793b6bf

Commits on Apr 22, 2015

• Adding in-place AddConstant and MulConstant

  fmassa

  committed
  f5a9cd5

  Copy full SHA for f5a9cd5

Commits on Apr 21, 2015

• Remove unnecessary buffer in Power

  fmassa

  committed
  81ff14e

  Copy full SHA for 81ff14e

• Adding strides for avoiding copies of non-contiguous tensors

  Show description for 1750204
  fmassa

  committed
  1750204

  Copy full SHA for 1750204

Pagination

Previous

Next