Commits

Breadcrumbs

History for

onmaster

User selector

All users

All time

Commit History

Commits on Nov 28, 2013

• Code simplication following code review.

  nouiz

  committed
  6069c87

  Copy full SHA for 6069c87

Commits on Nov 25, 2013

• Fix the dataset loading when in the code directory.

  nouiz

  committed
  8a9a83d

  Copy full SHA for 8a9a83d

• Always load the data from the data folder.

  Show description for c475870
  nouiz

  committed
  c475870

  Copy full SHA for c475870

Commits on Mar 27, 2013

• Fix off-by-one error in numbering of epochs in most tutorials

  Show description for 2dd943a
  lamblin

  committed
  2dd943a

  Copy full SHA for 2dd943a

Commits on Jan 31, 2013

• Use list of pairs for updates, not dict, to remove warning

  lamblin

  committed
  0440226

  Copy full SHA for 0440226

Commits on Sep 6, 2012

• use borrow=True when making shared variable to make the example work with 32 bits python.

  nouiz

  committed
  8596e69

  Copy full SHA for 8596e69

Commits on Jun 19, 2012

• Add doc about the format of the datasets.

  nouiz

  committed
  ded1ad6

  Copy full SHA for ded1ad6

Commits on Mar 14, 2012

• pep8

  nouiz

  committed
  19625bc

  Copy full SHA for 19625bc

Commits on Jul 17, 2011

• Use numpy.inf instead of float('inf').

  Show description for 230e17f
  lamblin

  committed
  230e17f

  Copy full SHA for 230e17f

Commits on Apr 1, 2011

• Automatically download the MNIST dataset.

  lamblin

  committed
  c4ca8f1

  Copy full SHA for c4ca8f1

Commits on Mar 15, 2011

• Remove use of .value in logistic regression

  lamblin

  committed
  72b5062

  Copy full SHA for 72b5062

Commits on Apr 1, 2010

• removed the expected time from the tutorial code. Now into test.py:speed

  nouiz

  committed
  a8ee2e6

  Copy full SHA for a8ee2e6

Commits on Mar 28, 2010

• updated logreg.py

  

  Razvan Pascanu

  committed
  7d8be54

  Copy full SHA for 7d8be54

• updated logistic sgd outputs

  pascanur

  committed
  87f7831

  Copy full SHA for 87f7831

Commits on Mar 26, 2010

• merging

  

  Razvan Pascanu

  committed
  a59a2bf

  Copy full SHA for a59a2bf

Commits on Mar 19, 2010

• moved the timing into the code to time only the exec time.

  nouiz

  committed
  223e870

  Copy full SHA for 223e870

Commits on Feb 19, 2010

• I merged my version 2 into the official deep learning tutorials

  

  Razvan Pascanu

  committed
  d71dce9

  Copy full SHA for d71dce9

Commits on Feb 18, 2010

• trying to make variable names and comments easier for the reader

  

  Yoshua Bengio

  committed
  1e9844e

  Copy full SHA for 1e9844e

Commits on Feb 16, 2010

• added comment to help parse [T.arange(y.shape[0]),y]

  

  Yoshua Bengio

  committed
  08476ed

  Copy full SHA for 08476ed

Commits on Feb 12, 2010

• a few typos + a few updates to rbm

  

  Razvan Pascanu

  committed
  cffce58

  Copy full SHA for cffce58

Commits on Feb 8, 2010

• working version of SdA.py

  

  Razvan Pascanu

  committed
  c8fc56d

  Copy full SHA for c8fc56d

Commits on Feb 5, 2010

• Using index instead of offset ; some bug fixes for the SdA.py; partially finished SdA tutorial ( I got stucked in trying to fix the code and didn't manage to finish everything )

  

  Razvan Pascanu

  committed
  9d2f5d3

  Copy full SHA for 9d2f5d3

Commits on Feb 3, 2010

• I have made all tutorials (code and documentation) to use shared variable to store dataset as James suggested. I also changed the validation_frequency to Yoshua's suggestion. One should still need…

  Show description for a074b32
  pascanur

  committed
  a074b32

  Copy full SHA for a074b32

Commits on Feb 2, 2010

• Changing to the new way of storing the dataset (using shared variables)

  pascanur

  committed
  ffa5bf9

  Copy full SHA for ffa5bf9

Commits on Jan 21, 2010

• SdA.py implements a stacked de-noising autoencoder in the tutorial style. I think DBN/SdA are a bit different from the rest, so if anybody has the time to read through it and give some feedback I w…

  Show description for 76f20c3
  pascanur

  committed
  76f20c3

  Copy full SHA for 76f20c3

Commits on Jan 17, 2010

• Change 'T.sum' to 'T.mean' in computations of NLL, change the documentation accordingly

  

  Pascal Lamblin

  committed
  3ed2a7f

  Copy full SHA for 3ed2a7f

Commits on Jan 15, 2010

• switching to T.sum when computing negative log likelihood (though T.mean should also be supported) for mlp and logistic_sgd. Second bug fixed was when initializing the weights, we need 6./(n_in+n_o…

  Show description for ff4bd33
  pascanur

  committed
  ff4bd33

  Copy full SHA for ff4bd33

Commits on Jan 14, 2010

• Removed casting of learning rate to numpy array

  pascanur

  committed
  5c2d4f3

  Copy full SHA for 5c2d4f3

• moved patience improvement 'if-block' inside validation loss improvement 'if-block'

  

  Yoshua Bengio

  committed
  d712ce8

  Copy full SHA for d712ce8

• I changed the pickled file such that it is not divided in minibatches before hand. Note that the code now generates some warning .. this is because of an issue with advanced indexing that Pascal w…

  Show description for 36c0080
  pascanur

  committed
  36c0080

  Copy full SHA for 36c0080

• Several TODO solved

  pascanur

  committed
  2d10ca2

  Copy full SHA for 2d10ca2

Commits on Jan 13, 2010

• merge

  

  James Bergstra

  committed
  882766f

  Copy full SHA for 882766f

• removed categorical_softmax and inserted advanced indexing

  

  James Bergstra

  committed
  092c587

  Copy full SHA for 092c587

Commits on Jan 12, 2010

• I changed the code to do a validation test every epoch. While this is not the best option, it was straight forward to implement, and from the discussion in the emails I got the feeling there is no…

  Show description for 4e572e8
  pascanur

  committed
  4e572e8

  Copy full SHA for 4e572e8

Commits on Jan 11, 2010

• Minibatches of size 20

  pascanur

  committed
  2aceb63

  Copy full SHA for 2aceb63