Commits

Commit history

1. Commits on Nov 18, 2014

   • Merge https://github.com/torch/cunn into l1-cost

     soumith

     committed
     ea97adbView commit details

     Copy full SHA for ea97adb

     Browse repository at this point

   • Merge pull request #43 from ajtulloch/enforce-device-ownership

     Show description for 51dbc11
     soumith

     committed
     51dbc11View commit details

     Copy full SHA for 51dbc11

     Browse repository at this point

   • Enforce devbice ownership in SpatialConvolution

     ajtulloch

     committed
     98d3239View commit details

     Copy full SHA for 98d3239

     Browse repository at this point

   • L1Cost CUNN implementation

     ajtulloch

     committed
     ba23c90View commit details

     Copy full SHA for ba23c90

     Browse repository at this point
2. Commits on Nov 13, 2014

   • Merge pull request #40 from nicholas-leonard/dropout

     Show description for b07ee65
     soumith

     committed
     b07ee65View commit details

     Copy full SHA for b07ee65

     Browse repository at this point

   • Dropout tests

     nicholas-leonard

     committed
     675ef50View commit details

     Copy full SHA for 675ef50

     Browse repository at this point
3. Commits on Nov 11, 2014

   • fixing global variables in tests

     soumith

     committed
     b269e95View commit details

     Copy full SHA for b269e95

     Browse repository at this point

   • fixing BHWD for new cublasV2 API

     soumith

     committed
     31a705aView commit details

     Copy full SHA for 31a705a

     Browse repository at this point

   • Replaced gemm by gemv for bias gradients.

     Show description for 059d889
     

     Clement Farabet

     committed
     059d889View commit details

     Copy full SHA for 059d889

     Browse repository at this point

   • Cleaned up whitespaces.

     

     Clement Farabet

     committed
     3bc84b2View commit details

     Copy full SHA for 3bc84b2

     Browse repository at this point
4. Commits on Nov 6, 2014

   • fixing temporal convolution for new cadd API (consistent with TH)

     soumith

     committed
     26ded2dView commit details

     Copy full SHA for 26ded2d

     Browse repository at this point

   • Merge pull request #35 from torch/TemporalFix

     Show description for 92f7582
     soumith

     committed
     92f7582View commit details

     Copy full SHA for 92f7582

     Browse repository at this point
5. Commits on Nov 4, 2014

   • fix usage of addmm for new API

     soumith

     committed
     e9730ddView commit details

     Copy full SHA for e9730dd

     Browse repository at this point
6. Commits on Oct 15, 2014

   • moving SpatialConvolutionMM to use THCudaBlas API which fixes some corner cases

     soumith

     committed
     d1cd450View commit details

     Copy full SHA for d1cd450

     Browse repository at this point
7. Commits on Oct 6, 2014

   • Merge pull request #30 from nicholas-leonard/maxpool

     Show description for fe6cdac
     soumith

     committed
     fe6cdacView commit details

     Copy full SHA for fe6cdac

     Browse repository at this point

   • SpatialMaxPooling can handle (dW != kW) or (dH ~= kH)

     nicholas-leonard

     committed
     87d55bfView commit details

     Copy full SHA for 87d55bf

     Browse repository at this point
8. Commits on Sep 22, 2014

   • adding CONTRIBUTING guidelines

     soumith

     committed
     43a9df7View commit details

     Copy full SHA for 43a9df7

     Browse repository at this point
9. Commits on Aug 1, 2014

   • Incorporated rectangular filter support recently added in Caffe repository.

     jzthree

     committed
     1b9f16aView commit details

     Copy full SHA for 1b9f16a

     Browse repository at this point
10. Commits on Jul 22, 2014

    • Deprecated SpatialConvolution.

      clementfarabet

      committed
      2741b7cView commit details

      Copy full SHA for 2741b7c

      Browse repository at this point

    • SpatialConvolutionMM now supports arbitrary batch sizes (and 3-dim

      Show description for 629bd1e
      clementfarabet

      committed
      629bd1eView commit details

      Copy full SHA for 629bd1e

      Browse repository at this point

    • Merge pull request #23 from jonathantompson/kl_divergence

      Show description for b6f892b
      soumith

      committed
      b6f892bView commit details

      Copy full SHA for b6f892b

      Browse repository at this point

    • Merge pull request #24 from jonathantompson/min

      Show description for 1e21b9d
      clementfarabet

      committed
      1e21b9dView commit details

      Copy full SHA for 1e21b9d

      Browse repository at this point

    • added a CUDA version of Min (basically copy-paste from Max).

      jonathantompson

      committed
      1023bdaView commit details

      Copy full SHA for 1023bda

      Browse repository at this point

    • removed redundant kernels from DistKLDivCriterion and modified test.

      jonathantompson

      committed
      927f209View commit details

      Copy full SHA for 927f209

      Browse repository at this point
11. Commits on Jul 21, 2014

    • added DistKLDivCriterion CUDA implementation and a unit test for it. Also added modified the MSE unit test to include sizeAverage tests as well (to get all code paths).

      jonathantompson

      committed
      95f7a57View commit details

      Copy full SHA for 95f7a57

      Browse repository at this point

    • SpatialConvolutionMM supports dim=3 (once gradBias is implementeD).

      clementfarabet

      committed
      2f67641View commit details

      Copy full SHA for 2f67641

      Browse repository at this point

    • SpatialConvolutionMM:updateOutput() supports 3dim

      clementfarabet

      committed
      8a46534View commit details

      Copy full SHA for 8a46534

      Browse repository at this point

    • Removed half-baked (and old) module SpatialConvolutionMap

      clementfarabet

      committed
      b4f9078View commit details

      Copy full SHA for b4f9078

      Browse repository at this point
12. Commits on Jul 20, 2014

    • Merge pull request #20 from nicholas-leonard/mixturetable

      Show description for 5857f7f
      clementfarabet

      committed
      5857f7fView commit details

      Copy full SHA for 5857f7f

      Browse repository at this point

    • Merge branch 'master' of github.com:torch/cunn into mixturetable

      nicholas-leonard

      committed
      8aafbe3View commit details

      Copy full SHA for 8aafbe3

      Browse repository at this point

    • fixed default tensor in test bug

      nicholas-leonard

      committed
      68a8d67View commit details

      Copy full SHA for 68a8d67

      Browse repository at this point
13. Commits on Jul 19, 2014

    • Fixed padding and scale parameter.

      clementfarabet

      committed
      7c0dfbdView commit details

      Copy full SHA for 7c0dfbd

      Browse repository at this point

    • harmonized unit test with torch7/cutorch/nn

      nicholas-leonard

      committed
      305022fView commit details

      Copy full SHA for 305022f

      Browse repository at this point
14. Commits on Jul 18, 2014

    • extended benchmark

      clementfarabet

      committed
      3515a9dView commit details

      Copy full SHA for 3515a9d

      Browse repository at this point

    • Merge branch 'master' of github.com:torch/cunn

      soumith

      committed
      09822f8View commit details

      Copy full SHA for 09822f8

      Browse repository at this point