Commits

Branch selector

 v0.1.9

User selector

All users

Datepicker

All time

Commit History

Commits on Feb 17, 2017

• Fix previous_functions when it contains Variables

  

  colesbury

  authored and

  soumith

  committed
  dd844f7

  Copy full SHA for dd844f7

• Fix flaky non-contig test

  

  apaszke

  authored and

  soumith

  committed
  7117a90

  Copy full SHA for 7117a90

• Add torch.__version__

  

  apaszke

  authored and

  soumith

  committed
  1bdc281

  Copy full SHA for 1bdc281

• Fix a bug in Engine::compute_dependencies

  

  apaszke

  authored and

  soumith

  committed
  5e150ca

  Copy full SHA for 5e150ca

• Make detach() actually remove the creator

  

  apaszke

  authored and

  soumith

  committed
  c0c62d0

  Copy full SHA for c0c62d0

• Make torch.Size methods return torch.Size, not tuple

  

  apaszke

  authored and

  soumith

  committed
  b9ece39

  Copy full SHA for b9ece39

Commits on Feb 16, 2017

• allow DataParallel to have tuple inputs on a single GPU

  

  Natalia Gimelshein

  authored and

  apaszke

  committed
  7c44506

  Copy full SHA for 7c44506

• Improve nn.legacy compatibility with Torch7 (#738)

  

  mvitez

  authored and

  soumith

  committed
  937ba58

  Copy full SHA for 937ba58

• setup.cfg -> tox.ini (#761)

  soumith

  authored
  2ae54f1

  Copy full SHA for 2ae54f1

Commits on Feb 15, 2017

• Update rnn.py

  Show description for a217fef
  

  bdfhjk

  authored and

  apaszke

  committed
  a217fef

  Copy full SHA for a217fef

• add test for bias=False for 3d convolution

  

  Eli Stevens

  authored and

  soumith

  committed
  5221745

  Copy full SHA for 5221745

• Merge commit '797544c47a4e9bdff02137a127f883a6df9b3dfe'

  soumith

  committed
  000ca44

  Copy full SHA for 000ca44

• Merge commit '0426f2f3ec2b932cb83d64101081244c2a1451b1'

  soumith

  committed
  8f3d440

  Copy full SHA for 8f3d440

• Merge commit '07f5b21ef1bd29d1451c616062dcbfc3f8fd7c6a'

  soumith

  committed
  7cc14c5

  Copy full SHA for 7cc14c5

• implementation of bias=False for VolConv.cu

  

  Eli Stevens

  authored and

  soumith

  committed
  797544c

  Copy full SHA for 797544c

• implementation of bias=False for VolConv.c

  Show description for 0426f2f
  

  Eli Stevens

  authored and

  soumith

  committed
  0426f2f

  Copy full SHA for 0426f2f

• kernel_size as the default stride for avg_pool1d (#744)

  Show description for 336eeee
  

  supakjk

  authored and

  soumith

  committed
  336eeee

  Copy full SHA for 336eeee

• Fixed a simple compiling erroin mac OS #745. (#746)

  Show description for 593f867
  

  Teaonly

  authored and

  soumith

  committed
  593f867

  Copy full SHA for 593f867

• Fix class torch.nn.ConvTransposeNd documentation (#739)

  Show description for 385913b
  

  Atcold

  authored and

  soumith

  committed
  385913b

  Copy full SHA for 385913b

• Fix LSTMCell Doc Typo (#743)

  

  dexter1691

  authored and

  soumith

  committed
  6aaa14f

  Copy full SHA for 6aaa14f

• Merge pull request #702 from gchanan/conservativeAllocator

  Show description for 07f5b21
  soumith

  authored
  07f5b21

  Copy full SHA for 07f5b21

Commits on Feb 14, 2017

• Free set of stored streams and handle NULL streams.

  gchanan

  committed
  e454870

  Copy full SHA for e454870

• Fix flaky tests

  apaszke

  committed
  2822013

  Copy full SHA for 2822013

• Add some more asserts to cuDNN RNN

  apaszke

  committed
  72c1982

  Copy full SHA for 72c1982

• Support retain_variables in cuDNN RNN

  apaszke

  committed
  0de2ea3

  Copy full SHA for 0de2ea3

• Raise error when too small input is given to conv

  apaszke

  committed
  d899385

  Copy full SHA for d899385

• Check that all tensors are on the same GPU in cuDNN bindings

  apaszke

  committed
  c6d6cbe

  Copy full SHA for c6d6cbe

• Fix bug in zero_grad, when some parameters didn't require grad

  apaszke

  committed
  85e82e8

  Copy full SHA for 85e82e8

• Fix auto-gpu in cat

  apaszke

  committed
  a1534cc

  Copy full SHA for a1534cc

• Load half and double THCUNN backends

  apaszke

  committed
  8c8dc79

  Copy full SHA for 8c8dc79

• Add tests for non-contiguous inputs and gradients

  apaszke

  committed
  63edca4

  Copy full SHA for 63edca4

• compile with cudart (#737)

  

  kashif

  authored and

  soumith

  committed
  8d90ab2

  Copy full SHA for 8d90ab2

• Refactor autograd package to separate Python dependencies. (#662)

  Show description for bd53030
  colesbury

  authored
  bd53030

  Copy full SHA for bd53030

Commits on Feb 13, 2017

• make networks converted with loadcaffe loadable

  

  szagoruyko

  authored and

  apaszke

  committed
  16d2c3d

  Copy full SHA for 16d2c3d

• std::min() requires same type (#732)

  Show description for 407a92d
  

  kashif

  authored and

  apaszke

  committed
  407a92d

  Copy full SHA for 407a92d

Pagination

Previous

Next