Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 18, 2014

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

  soumith

  committed
  ea97adb

  Copy full SHA for ea97adb

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

  Show description for 51dbc11
  soumith

  committed
  51dbc11

  Copy full SHA for 51dbc11

• Enforce devbice ownership in SpatialConvolution

  ajtulloch

  committed
  98d3239

  Copy full SHA for 98d3239

• L1Cost CUNN implementation

  ajtulloch

  committed
  ba23c90

  Copy full SHA for ba23c90

Commits on Nov 13, 2014

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

  Show description for b07ee65
  soumith

  committed
  b07ee65

  Copy full SHA for b07ee65

• Dropout tests

  nicholas-leonard

  committed
  675ef50

  Copy full SHA for 675ef50

Commits on Nov 11, 2014

• fixing global variables in tests

  soumith

  committed
  b269e95

  Copy full SHA for b269e95

• fixing BHWD for new cublasV2 API

  soumith

  committed
  31a705a

  Copy full SHA for 31a705a

• Replaced gemm by gemv for bias gradients.

  Show description for 059d889
  

  Clement Farabet

  committed
  059d889

  Copy full SHA for 059d889

• Cleaned up whitespaces.

  

  Clement Farabet

  committed
  3bc84b2

  Copy full SHA for 3bc84b2

Commits on Nov 6, 2014

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

  soumith

  committed
  26ded2d

  Copy full SHA for 26ded2d

• Merge pull request #35 from torch/TemporalFix

  Show description for 92f7582
  soumith

  committed
  92f7582

  Copy full SHA for 92f7582

Commits on Nov 4, 2014

• fix usage of addmm for new API

  soumith

  committed
  e9730dd

  Copy full SHA for e9730dd

Commits on Oct 15, 2014

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

  soumith

  committed
  d1cd450

  Copy full SHA for d1cd450

Commits on Oct 6, 2014

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

  Show description for fe6cdac
  soumith

  committed
  fe6cdac

  Copy full SHA for fe6cdac

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

  nicholas-leonard

  committed
  87d55bf

  Copy full SHA for 87d55bf

Commits on Sep 22, 2014

• adding CONTRIBUTING guidelines

  soumith

  committed
  43a9df7

  Copy full SHA for 43a9df7

Commits on Aug 1, 2014

• Incorporated rectangular filter support recently added in Caffe repository.

  jzthree

  committed
  1b9f16a

  Copy full SHA for 1b9f16a

Commits on Jul 22, 2014

• Deprecated SpatialConvolution.

  clementfarabet

  committed
  2741b7c

  Copy full SHA for 2741b7c

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

  Show description for 629bd1e
  clementfarabet

  committed
  629bd1e

  Copy full SHA for 629bd1e

• Merge pull request #23 from jonathantompson/kl_divergence

  Show description for b6f892b
  soumith

  committed
  b6f892b

  Copy full SHA for b6f892b

• Merge pull request #24 from jonathantompson/min

  Show description for 1e21b9d
  clementfarabet

  committed
  1e21b9d

  Copy full SHA for 1e21b9d

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

  jonathantompson

  committed
  1023bda

  Copy full SHA for 1023bda

• removed redundant kernels from DistKLDivCriterion and modified test.

  jonathantompson

  committed
  927f209

  Copy full SHA for 927f209

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
  95f7a57

  Copy full SHA for 95f7a57

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

  clementfarabet

  committed
  2f67641

  Copy full SHA for 2f67641

• SpatialConvolutionMM:updateOutput() supports 3dim

  clementfarabet

  committed
  8a46534

  Copy full SHA for 8a46534

• Removed half-baked (and old) module SpatialConvolutionMap

  clementfarabet

  committed
  b4f9078

  Copy full SHA for b4f9078

Commits on Jul 20, 2014

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

  Show description for 5857f7f
  clementfarabet

  committed
  5857f7f

  Copy full SHA for 5857f7f

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

  nicholas-leonard

  committed
  8aafbe3

  Copy full SHA for 8aafbe3

• fixed default tensor in test bug

  nicholas-leonard

  committed
  68a8d67

  Copy full SHA for 68a8d67

Commits on Jul 19, 2014

• Fixed padding and scale parameter.

  clementfarabet

  committed
  7c0dfbd

  Copy full SHA for 7c0dfbd

• harmonized unit test with torch7/cutorch/nn

  nicholas-leonard

  committed
  305022f

  Copy full SHA for 305022f

Commits on Jul 18, 2014

• extended benchmark

  clementfarabet

  committed
  3515a9d

  Copy full SHA for 3515a9d

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

  soumith

  committed
  09822f8

  Copy full SHA for 09822f8

Pagination

Previous

Next