Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Apr 7, 2017

• Add learning rate to adadelta

  

  KeDengMS

  committed
  e7c1609

  Copy full SHA for e7c1609

Commits on Apr 4, 2017

• Fix Adadelta in CPU Sparse

  

  KeDengMS

  committed
  fe9b401

  Copy full SHA for fe9b401

Commits on Apr 3, 2017

• CNTK v2 library: Several sparse data handling fixes

  Show description for fba356f
  amitaga

  committed
  fba356f

  Copy full SHA for fba356f

Commits on Mar 31, 2017

• Make sparse label CE backprop faster

  Show description for 59a2c26
  

  KeDengMS

  committed
  59a2c26

  Copy full SHA for 59a2c26

Commits on Mar 27, 2017

• fix comments and issue

  chentaMS

  committed
  9fe9dfe

  Copy full SHA for 9fe9dfe

• fix comments, add axis paramter, support invalid indices

  chentaMS

  committed
  4f762c3

  Copy full SHA for 4f762c3

• remove hard tab

  chentaMS

  committed
  07bdfe9

  Copy full SHA for 07bdfe9

• update sparse and GPU sparse implementation

  chentaMS

  committed
  fdfbd4c

  Copy full SHA for fdfbd4c

• add sparse support

  chentaMS

  committed
  38e35de

  Copy full SHA for 38e35de

Commits on Mar 21, 2017

• Integrate kedeng/adadelta2 into master

  

  Project Philly

  committed
  469bdda

  Copy full SHA for 469bdda

• Implement AdaDelta learner

  

  KeDengMS

  committed
  b05a6b1

  Copy full SHA for b05a6b1

• optimize memory allocation based on matrix type; remove unnecessary memset, as they have been value initialized.

  zhouwangzw

  committed
  45b349b

  Copy full SHA for 45b349b

Commits on Mar 17, 2017

• CNTK v2 library: Improve NDArrayView creation performance for sparse data

  amitaga

  committed
  bd7df45

  Copy full SHA for bd7df45

Commits on Mar 10, 2017

• CNTK v2 library: Add support for sparse inputs with sample shape > 1D

  amitaga

  committed
  c0e42d8

  Copy full SHA for c0e42d8

Commits on Feb 23, 2017

• Fix a bug that CPUSparseMatrix::DoGatherColumnOf may generate matrix that's not CSC compliant and cause CUDA failure in GEMM

  

  KeDengMS

  committed
  d78332a

  Copy full SHA for d78332a

Commits on Jan 24, 2017

• Speed up TimesNodeBase for sparse by avoid unroll

  Show description for 16daa4a
  

  KeDengMS

  committed
  16daa4a

  Copy full SHA for 16daa4a

Commits on Jan 16, 2017

• Add unit-gain momentum flag

  Show description for 3ab2468
  

  Alexey Reznichenko

  committed
  3ab2468

  Copy full SHA for 3ab2468

Commits on Jan 5, 2017

• Add trace for nodes with abs sum to make compare easier, with trace level 1000

  

  KeDengMS

  committed
  40d0b79

  Copy full SHA for 40d0b79

Commits on Dec 7, 2016

• Fix sparse matrix MultiplyAndAdd bug that old value may not be kept

  

  KeDengMS

  committed
  c12979b

  Copy full SHA for c12979b

Commits on Oct 7, 2016

• fix build errors due to new gcc compiler (5.x from ubuntu 16.4.1)

  fmegen

  committed
  c5f2540

  Copy full SHA for c5f2540

Commits on Sep 19, 2016

• Fix path differing only in case. Fix hard tabs.

  mahilleb-msft

  committed
  74611be

  Copy full SHA for 74611be

Commits on Sep 16, 2016

• Moved the test for emptyness

  thilow

  committed
  d335c24

  Copy full SHA for d335c24

Commits on Sep 15, 2016

• some small fixes

  thilow

  committed
  a13a7e5

  Copy full SHA for a13a7e5

• cr related fixes

  thilow

  committed
  aa553ce

  Copy full SHA for aa553ce

• fixing formating string for unsigned long long

  thilow

  committed
  c2e34f5

  Copy full SHA for c2e34f5

• Fixing simple cr related issues

  thilow

  committed
  6f5fbbe

  Copy full SHA for 6f5fbbe

Commits on Aug 26, 2016

• removing some uneeded (int) casts.

  thilow

  committed
  e7c884a

  Copy full SHA for e7c884a

• Fixing issue that for Sparse * Dense the dense*Dense path was taken afterwards

  thilow

  committed
  3f5f002

  Copy full SHA for 3f5f002

• Fixing some newly introduced bugs

  thilow

  committed
  ad17a78

  Copy full SHA for ad17a78

Commits on Aug 25, 2016

• CPU Sparse*Dense->Dense compiles

  thilow

  committed
  5658594

  Copy full SHA for 5658594

• First implementation of new sparse*dense for CPU

  thilow

  committed
  1810e7c

  Copy full SHA for 1810e7c

Commits on Aug 21, 2016

• CNTK v2 Library: a) Add dynamic axis support b) New primitive functions and some higher level functions and c) Sequence classification test

  amitaga

  committed
  fa4b99d

  Copy full SHA for fa4b99d

Commits on Aug 17, 2016

• more acml removal

  wolfma61

  committed
  79cfcf7

  Copy full SHA for 79cfcf7

Commits on Aug 12, 2016

• Fix of fix

  thilow

  committed
  8d7ed08

  Copy full SHA for 8d7ed08

• Fixing another bug in Times(Dense,Sparse) and restructure code

  thilow

  committed
  6f7505f

  Copy full SHA for 6f7505f

Pagination

Previous

Next