Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Apr 30, 2016

• Merge pull request karpathy#164 from gdb/master

  Show description for 6f9487a
  karpathy

  committed
  6f9487a

  Copy full SHA for 6f9487a

• Update Readme.md

  gdb

  committed
  672d226

  Copy full SHA for 672d226

Commits on Feb 17, 2016

• update point to torch-rnn

  karpathy

  committed
  f339b7f

  Copy full SHA for f339b7f

Commits on Nov 25, 2015

• Merge pull request karpathy#132 from udibr/master

  Show description for a8c1582
  karpathy

  committed
  a8c1582

  Copy full SHA for a8c1582

Commits on Nov 24, 2015

• fix message

  udibr

  committed
  f7a8e74

  Copy full SHA for f7a8e74

• bug fix

  udibr

  committed
  0d55784

  Copy full SHA for 0d55784

Commits on Oct 26, 2015

• Merge pull request karpathy#125 from skwerlman/master

  Show description for 106a75c
  karpathy

  committed
  106a75c

  Copy full SHA for 106a75c

• Fix unclear errors

  Show description for b3c7e3c
  skwerlman

  committed
  b3c7e3c

  Copy full SHA for b3c7e3c

Commits on Oct 16, 2015

• Readme tweaks and clarifications

  karpathy

  committed
  54b4c66

  Copy full SHA for 54b4c66

Commits on Oct 11, 2015

• speeding up the code by about 15% with faster indexing. I did not imagine this would have such a large effect. Thank you @vseledkin for pointing this out

  karpathy

  committed
  82baeea

  Copy full SHA for 82baeea

Commits on Sep 21, 2015

• adding clarification justifying not normalizing gradient by seqlength

  karpathy

  committed
  4297a9b

  Copy full SHA for 4297a9b

Commits on Sep 20, 2015

• changing default LSTM initialization to use biases of 1.0 for the forget gates, which is a common trick and encourages remembering in the beginning of training. I ran 2 medium-sized experiments and…

  Show description for 0dfeaa4
  karpathy

  committed
  0dfeaa4

  Copy full SHA for 0dfeaa4

Commits on Aug 27, 2015

• Merge pull request karpathy#92 from ajkl/patch-1

  Show description for 065c5ff
  karpathy

  committed
  065c5ff

  Copy full SHA for 065c5ff

Commits on Aug 26, 2015

• adding torch/nn to dependencies

  ajkl

  committed
  82d1eeb

  Copy full SHA for 82d1eeb

Commits on Aug 20, 2015

• Merge pull request karpathy#87 from johnny5550822/master

  Show description for fd1aa47
  karpathy

  committed
  fd1aa47

  Copy full SHA for fd1aa47

• Update Readme.md

  Show description for 69300d5
  johnny5550822

  committed
  69300d5

  Copy full SHA for 69300d5

Commits on Aug 10, 2015

• Merge pull request karpathy#79 from EricZeiberg/patch-1

  Show description for f58a961
  karpathy

  committed
  f58a961

  Copy full SHA for f58a961

• Remove repeated info in Readme

  EricZeiberg

  committed
  5dda019

  Copy full SHA for 5dda019

Commits on Aug 8, 2015

• fixing a bug introduced in previous commit. We have to use doubles not floats when we convert GPU models to CPU, because CPU trained models use doubles. This introduced an error where a CPU checkpo…

  Show description for 0fb9a77
  karpathy

  committed
  0fb9a77

  Copy full SHA for 0fb9a77

Commits on Aug 5, 2015

• quick patch for converting GPU checkpoints to CPU checkpoints

  karpathy

  committed
  86a8edd

  Copy full SHA for 86a8edd

Commits on Aug 1, 2015

• Merge pull request karpathy#74 from ananda621/handle_large_data

  Show description for 25e2a6c
  karpathy

  committed
  25e2a6c

  Copy full SHA for 25e2a6c

• Merge pull request karpathy#66 from donglixp/patch-1

  Show description for c9fbaa2
  karpathy

  committed
  c9fbaa2

  Copy full SHA for c9fbaa2

• Merge pull request karpathy#73 from guillitte/master

  Show description for 5a1793b
  karpathy

  committed
  5a1793b

  Copy full SHA for 5a1793b

• revert to opt.savefile in file name

  guillitte

  committed
  e555fa9

  Copy full SHA for e555fa9

• Load input files chunk by chunk to handle files that are larger than 1GB for which reading whole file will cause luajit to exit with message 'not enough memory'

  

  Ananda

  committed
  43ec480

  Copy full SHA for 43ec480

Commits on Jul 31, 2015

• Added support for GRU and RNN

  guillitte

  committed
  e86e0ad

  Copy full SHA for e86e0ad

• Added support for GRU and RNN

  guillitte

  committed
  ec22113

  Copy full SHA for ec22113

• Update train.lua

  guillitte

  committed
  302a4cc

  Copy full SHA for 302a4cc

• Update train.lua

  Show description for d877830
  guillitte

  committed
  d877830

  Copy full SHA for d877830

• Added oneHot encoding and softmax

  guillitte

  committed
  c54a856

  Copy full SHA for c54a856

• Update GRU.lua

  Show description for 920ef81
  guillitte

  committed
  920ef81

  Copy full SHA for 920ef81

Commits on Jul 27, 2015

• fixing merge conflict

  karpathy

  committed
  ef0373f

  Copy full SHA for ef0373f

• adding gist pointer

  karpathy

  committed
  57c2e2a

  Copy full SHA for 57c2e2a

Commits on Jul 25, 2015

• Merge pull request karpathy#68 from hughperkins/splittable

  Show description for 7b937f3
  karpathy

  committed
  7b937f3

  Copy full SHA for 7b937f3

• migrate from narrow to splittable

  hughperkins

  committed
  08e7606

  Copy full SHA for 08e7606

Pagination

Previous

Next