Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on May 3, 2017

• Fix memory leak introduced by 72e8190 (pytorch#1464)

  

  apaszke

  authored and

  soumith

  committed
  e9d648c

  Copy full SHA for e9d648c

• Use at most one shared_ptr block at a time to manage THPFunctions (pytorch#1454)

  Show description for 72e8190
  

  apaszke

  authored and

  soumith

  committed
  72e8190

  Copy full SHA for 72e8190

Commits on May 1, 2017

• Fix bugs, rename differentiate to grad, make it more flexible

  

  apaszke

  authored and

  soumith

  committed
  5c74534

  Copy full SHA for 5c74534

• Bug fixes

  

  apaszke

  authored and

  soumith

  committed
  87164f5

  Copy full SHA for 87164f5

• Fix memory issues with Conv and BatchNorm

  

  apaszke

  authored and

  soumith

  committed
  267e7c0

  Copy full SHA for 267e7c0

• Add torch.autograd.differentiate

  

  apaszke

  authored and

  soumith

  committed
  e5db8f9

  Copy full SHA for e5db8f9

• Convert some of the functions to new format

  Show description for 20aa5b0
  

  apaszke

  authored and

  soumith

  committed
  20aa5b0

  Copy full SHA for 20aa5b0

• Add support for the new Function format

  

  apaszke

  authored and

  soumith

  committed
  de9998e

  Copy full SHA for de9998e

• Make Variables not subclass Function anymore

  Show description for 702a2e3
  

  apaszke

  authored and

  soumith

  committed
  702a2e3

  Copy full SHA for 702a2e3

• Refactor attribute names in autograd

  

  apaszke

  authored and

  soumith

  committed
  2ca787f

  Copy full SHA for 2ca787f

Commits on Apr 28, 2017

• Refactor Python string utility function

  

  colesbury

  authored and

  apaszke

  committed
  4c1cdb6

  Copy full SHA for 4c1cdb6

• re-enable dilated convolutions on Kepler (pytorch#1394)

  

  ngimel

  authored and

  soumith

  committed
  775481e

  Copy full SHA for 775481e

• temp fix for transposed dilated convolution (pytorch#1388)

  

  ngimel

  authored and

  soumith

  committed
  8db2cf6

  Copy full SHA for 8db2cf6

Commits on Apr 23, 2017

• Fix errors when no CUDA devices are available (pytorch#1334)

  Show description for aab30d4
  

  colesbury

  authored and

  soumith

  committed
  aab30d4

  Copy full SHA for aab30d4

Commits on Apr 20, 2017

• Disable cudnn dilated convolutions for kepler. (pytorch#1308)

  

  csarofeen

  authored and

  soumith

  committed
  d2472d1

  Copy full SHA for d2472d1

Commits on Apr 17, 2017

• Fix group-convolution w/o biases on CPU. (pytorch#1273)

  Show description for a220f2c
  

  lucasb-eyer

  authored and

  soumith

  committed
  a220f2c

  Copy full SHA for a220f2c

Commits on Apr 12, 2017

• Autograd deadlock for recent glibc fix (pytorch#1243)

  

  albanD

  authored and

  apaszke

  committed
  71303b8

  Copy full SHA for 71303b8

Commits on Apr 7, 2017

• dont use cudnn batchnorm for cudnn < 5.1.10

  soumith

  committed
  7f2553b

  Copy full SHA for 7f2553b

Commits on Apr 6, 2017

• fix precedence problem when building with debug python (pytorch#1201)

  

  albanD

  authored and

  soumith

  committed
  a7ae04a

  Copy full SHA for a7ae04a

Commits on Mar 31, 2017

• check for nvidia driver's sufficiency before checking for number of CUDA devices (pytorch#1156)

  soumith

  authored
  b13b701

  Copy full SHA for b13b701

Commits on Mar 29, 2017

• Fix deadlock in autograd (pytorch#1140)

  

  apaszke

  authored and

  soumith

  committed
  8aa1cef

  Copy full SHA for 8aa1cef

Commits on Mar 22, 2017

• Check torch.backends.cudnn.enabled, padding, and output_padding (pytorch#996)

  Show description for b3ab4b1
  

  ngimel

  authored and

  colesbury

  committed
  b3ab4b1

  Copy full SHA for b3ab4b1

Commits on Mar 14, 2017

• Fix conv1d backward segfault (pytorch#999)

  

  ngimel

  authored and

  colesbury

  committed
  e86db38

  Copy full SHA for e86db38

Commits on Mar 7, 2017

• Fix some simple build error on MacOS (pytorch#949)

  Show description for bd0e9a7
  

  Teaonly

  authored and

  soumith

  committed
  bd0e9a7

  Copy full SHA for bd0e9a7

Commits on Mar 6, 2017

• Implement 'pre' and 'post' hooks at the C++ autograd level

  colesbury

  committed
  5073132

  Copy full SHA for 5073132

Commits on Mar 3, 2017

• Parallelize backwards

  colesbury

  committed
  34ce58c

  Copy full SHA for 34ce58c

• sparse tensor operations (pytorch#735)

  

  martinraison

  authored and

  apaszke

  committed
  f17cfe4

  Copy full SHA for f17cfe4

Commits on Mar 1, 2017

• Handle duplicate backward roots in autograd

  apaszke

  committed
  977630b

  Copy full SHA for 977630b

• Add support for variable length sequences in RNNs (pytorch#873)

  apaszke

  authored
  da72583

  Copy full SHA for da72583

Commits on Feb 27, 2017

• Fix one more reference cycle and ensure correct flag propagation (pytorch#868)

  

  apaszke

  authored and

  soumith

  committed
  502ebed

  Copy full SHA for 502ebed

• Improve autograd memory usage (pytorch#859)

  

  apaszke

  authored and

  soumith

  committed
  2b23712

  Copy full SHA for 2b23712

Commits on Feb 26, 2017

• Prevent creation of reference cycles with leaf Variables that don't require grad

  Show description for 3194191
  apaszke

  committed
  3194191

  Copy full SHA for 3194191

Commits on Feb 17, 2017

• Fix Engine::compute_dependencies

  

  apaszke

  authored and

  soumith

  committed
  aaf41c6

  Copy full SHA for aaf41c6

• Fix previous_functions when it contains Variables

  

  colesbury

  authored and

  soumith

  committed
  dd844f7

  Copy full SHA for dd844f7

• Fix a bug in Engine::compute_dependencies

  

  apaszke

  authored and

  soumith

  committed
  5e150ca

  Copy full SHA for 5e150ca

Pagination

Previous

Next