Tue, 08 Mar 2011 12:50:37 -0500 |
Frederic Bastien |
generalized the buildbot script following code review in Theano.
|
Tue, 08 Mar 2011 12:30:00 -0500 |
Frederic Bastien |
fix PYTHONPATH.
|
Mon, 07 Mar 2011 14:53:48 -0500 |
Frederic Bastien |
removed reference to our local filesystem.
|
Mon, 07 Mar 2011 14:35:44 -0500 |
Frederic Bastien |
added buildbot script
|
Sun, 06 Mar 2011 15:14:51 -0500 |
Pascal Lamblin |
Update inline doc.
|
Wed, 02 Mar 2011 13:13:40 -0500 |
James Bergstra |
pca - better comments
|
Wed, 02 Mar 2011 13:05:20 -0500 |
James Bergstra |
pca - added pca_whiten2 with better interface than pca_whiten
|
Fri, 25 Feb 2011 16:38:33 -0500 |
Frederic Bastien |
fixed indentation.
|
Fri, 25 Feb 2011 16:37:48 -0500 |
Frederic Bastien |
put a file to the same license as the rest of the project.
|
Tue, 22 Feb 2011 11:43:39 -0500 |
Razvan Pascanu |
renamed output as key for dictionary ( from field). Now for dictionary we have keys, for tuples/list we have entries and for object fields
|
Tue, 22 Feb 2011 11:40:02 -0500 |
Razvan Pascanu |
added option to order descending /ascending the fields acording to their size
|
Tue, 22 Feb 2011 11:23:32 -0500 |
Razvan Pascanu |
revised version of pkldu that is a bit more structured code wise and outputs in human readable units
|
Tue, 22 Feb 2011 10:04:50 -0500 |
Frederic Bastien |
white space fix.
|
Tue, 22 Feb 2011 10:04:32 -0500 |
Frederic Bastien |
removed import to file that don't exist in this repo.
|
Mon, 14 Feb 2011 19:27:37 -0500 |
Pascal Lamblin |
Add a cast in the chain of transformation of initial data set
|
Tue, 08 Feb 2011 16:19:46 -0500 |
Frederic Bastien |
added comment.
|
Tue, 08 Feb 2011 16:19:18 -0500 |
Frederic Bastien |
'allow to randomize the sparse valid/test utlc dataset at load time'
|
Tue, 08 Feb 2011 16:17:56 -0500 |
Frederic Bastien |
fix loading of sparse utlc dataset when PYLEARN_DATA_ROOT have more then 1 directory.
|
Tue, 08 Feb 2011 12:57:15 -0500 |
Frederic Bastien |
add parameter to randomize the valid and test data.
|
Tue, 08 Feb 2011 12:34:07 -0500 |
Frederic Bastien |
fix whitespace/indentation.
|
Tue, 08 Feb 2011 12:33:33 -0500 |
Frederic Bastien |
in utlc.py, implement a parameter that return a Theano variable that represent the normalized data. Usefull for on the fly normalization.
|
Tue, 08 Feb 2011 11:14:37 -0500 |
Frederic Bastien |
fix whitespace.
|
Tue, 08 Feb 2011 11:14:04 -0500 |
Frederic Bastien |
return the first file in a file that exists in the directory in the PYLEARN_DATA_ROOT.
|
Mon, 07 Feb 2011 17:04:56 -0500 |
Ian Goodfellow |
added pickle disk usage inspection utility 'pkldu'
|
Fri, 04 Feb 2011 16:07:27 -0500 |
James Bergstra |
adding delta-bar-delta optimization updates to gd module
|
Fri, 04 Feb 2011 16:06:36 -0500 |
James Bergstra |
added some old test_pca file I never committed
|
Fri, 04 Feb 2011 16:06:00 -0500 |
James Bergstra |
dataset_ops/tinyimages - added a tinyimages_op that gives access to the full
|
Fri, 04 Feb 2011 16:05:22 -0500 |
James Bergstra |
TensorFnDataset - added x_ attribute that caches the dataset function return
|
Fri, 04 Feb 2011 16:03:25 -0500 |
James Bergstra |
TensorFnDataset - cleaned up docs a little
|
Fri, 04 Feb 2011 16:02:57 -0500 |
James Bergstra |
TensorDataset - added single_shape and batch_size properties
|
Fri, 04 Feb 2011 16:01:45 -0500 |
James Bergstra |
80char
|
Thu, 03 Feb 2011 18:07:04 -0500 |
James Bergstra |
added datasets.tinyimages.rebuild_numpy_file method to build 5GB memmappable file of all images
|
Thu, 03 Feb 2011 13:50:27 -0500 |
Frederic Bastien |
small fix to new PYLEARN_DATA_ROOT
|
Thu, 03 Feb 2011 13:16:30 -0500 |
Frederic Bastien |
Allow PYLEARN_DATA_ROOT to be a list of directory. created pylearn.datasets.config.get_filepath_in_roots(name) fct to find a file in the list of directory.
|
Wed, 02 Feb 2011 14:33:01 -0500 |
Frederic Bastien |
allow to read gzip filetensor when there is more then 1 tensor in the file.
|
Wed, 02 Feb 2011 12:39:35 -0500 |
Frederic Bastien |
allow to load the ule labels.
|
Wed, 02 Feb 2011 11:20:34 -0500 |
Frederic Bastien |
normalize the utlc ndarray dataset inplace to use less memory.
|
Mon, 31 Jan 2011 12:23:20 -0500 |
Philippe Hamel |
added a pca online estimator
|
Fri, 28 Jan 2011 11:00:11 -0500 |
Frederic Bastien |
allow to load UTLC transfer label data.
|
Fri, 28 Jan 2011 10:58:45 -0500 |
Frederic Bastien |
merge
|
Tue, 25 Jan 2011 04:16:33 -0500 |
Frederic Bastien |
added the normalization of the last UTLC dataset
|
Wed, 26 Jan 2011 16:55:44 -0500 |
Pascal Lamblin |
Add L1 and L2² costs to sigmoidal layer
|
Mon, 24 Jan 2011 13:18:43 -0500 |
Frederic Bastien |
normalize 5 of the UTLC datasets.
|
Fri, 21 Jan 2011 20:40:57 -0500 |
Pascal Lamblin |
int8 should be enough to represent digits from 0 to 9
|
Fri, 21 Jan 2011 17:05:46 -0500 |
Frederic Bastien |
first version of a script to load the utlc datasets.
|
Thu, 20 Jan 2011 11:28:08 -0500 |
Frederic Bastien |
allow pylearn.io.filetensor to accept gzip file handle.
|
Thu, 13 Jan 2011 17:53:38 -0500 |
James Bergstra |
added some comments in dataset_ops/cifar10
|
Thu, 13 Jan 2011 17:52:49 -0500 |
James Bergstra |
adding MultiHingeLoss cost function
|
Wed, 12 Jan 2011 17:03:30 -0500 |
Frederic Bastien |
rst syntax fix and update to what is currently done.
|
Tue, 11 Jan 2011 16:43:10 -0500 |
Frederic Bastien |
small format change.
|
Mon, 10 Jan 2011 14:55:39 -0500 |
Frederic Bastien |
new file about datalearn in pytables.
|
Sat, 08 Jan 2011 17:14:44 -0500 |
James Bergstra |
mcRBM.contrastive_grad accepts optional "consider_constant" arg
|
Fri, 07 Jan 2011 11:42:38 -0500 |
Frederic Bastien |
new version _read_header for filetensor that allow to work gzip file.
|
Thu, 06 Jan 2011 13:33:47 -0500 |
Frederic Bastien |
allow to read header of dataset from a string.
|
Mon, 20 Dec 2010 18:09:11 -0500 |
gdesjardins |
merge
|
Mon, 20 Dec 2010 18:08:48 -0500 |
gdesjardins |
* removed temporary caltech_silhouette2 dataset
|
Mon, 20 Dec 2010 18:08:04 -0500 |
gdesjardins |
added comments to hmc code (old outstanding changes)
|
Mon, 20 Dec 2010 17:59:49 -0500 |
gdesjardins |
pylearn dataset wrapper for entire NIST dataset (digits + lower + upper)
|
Tue, 14 Dec 2010 14:53:48 -0500 |
Olivier Delalleau |
Many small fixes in code review proposal
|
Tue, 14 Dec 2010 14:22:16 -0500 |
Olivier Delalleau |
Merged
|
Tue, 14 Dec 2010 14:18:41 -0500 |
Olivier Delalleau |
Added another open question in datalearn
|
Tue, 14 Dec 2010 12:53:58 -0500 |
Frederic Bastien |
update the proposed politics.
|
Tue, 14 Dec 2010 12:32:42 -0500 |
Frederic Bastien |
small old modif not commtited.
|
Wed, 08 Dec 2010 14:35:53 -0500 |
Frederic Bastien |
small rst fix.
|
Wed, 08 Dec 2010 14:30:13 -0500 |
Frederic Bastien |
First try to use numpy documentation syntax.
|
Mon, 06 Dec 2010 13:33:07 -0500 |
boulanni |
Added regularization terms
|
Fri, 03 Dec 2010 09:09:00 -0500 |
James Bergstra |
added a fn to datasets/tiny_images to output a mosaic of images from the dataset
|
Thu, 02 Dec 2010 12:46:50 -0500 |
Frederic Bastien |
remove reference to fringant2 that don't exist anymore.
|
Fri, 19 Nov 2010 13:24:16 -0500 |
Frederic Bastien |
make failing test less verbose.
|
Thu, 18 Nov 2010 14:14:48 -0500 |
Olivier Delalleau |
Minor fixes for better Sphinx doc output
|
Thu, 18 Nov 2010 14:00:49 -0500 |
Olivier Delalleau |
Rewrote the Theano-Like Data Flow section in datalearn.txt
|
Thu, 18 Nov 2010 11:45:10 -0500 |
Olivier Delalleau |
Commit guidelines have been added in API_coding_style
|
Thu, 18 Nov 2010 09:53:38 -0500 |
Frederic Bastien |
small bug fix in test discovered by making Variable class raise an error about logical comparaison then checking if it is zero or not.
|
Wed, 17 Nov 2010 13:21:21 -0500 |
Frederic Bastien |
fix and URL that was moved.
|
Mon, 15 Nov 2010 16:49:12 -0500 |
Olivier Delalleau |
Replied to RP
|
Mon, 15 Nov 2010 16:28:02 -0500 |
Razvan Pascanu |
comment on OD idea
|
Mon, 15 Nov 2010 16:17:59 -0500 |
Olivier Delalleau |
Added a suggestion to solve the problem of Fixed vs. Varying parameters
|
Mon, 15 Nov 2010 15:20:49 -0500 |
Olivier Delalleau |
datalearn: Added another comment on James' numeric iterator function
|
Mon, 15 Nov 2010 13:49:25 -0500 |
Razvan Pascanu |
Answered Yoshua's question
|
Mon, 15 Nov 2010 11:51:33 -0500 |
Olivier Delalleau |
Refactored datalearn committee file to be easier to read
|
Fri, 12 Nov 2010 13:49:13 -0500 |
Yoshua Bengio |
comment by YB
|
Fri, 12 Nov 2010 11:49:00 -0500 |
Razvan Pascanu |
reply to OD
|
Fri, 12 Nov 2010 11:36:30 -0500 |
Olivier Delalleau |
Reply to Razvan
|
Fri, 12 Nov 2010 11:11:49 -0500 |
Razvan Pascanu |
Reply to a comment of OD
|
Fri, 12 Nov 2010 10:39:19 -0500 |
Olivier Delalleau |
Datalearn replies / comments
|
Thu, 11 Nov 2010 22:40:01 -0500 |
Razvan Pascanu |
Some question/suggestions to datalearn
|
Thu, 11 Nov 2010 18:08:05 -0500 |
James Bergstra |
corrected use of epsilons in pca
|
Thu, 11 Nov 2010 17:53:13 -0500 |
James Bergstra |
comments on datalearn
|
Thu, 11 Nov 2010 17:36:39 -0500 |
James Bergstra |
merge
|
Thu, 11 Nov 2010 16:34:38 -0500 |
Olivier Delalleau |
Added datalearn committee discussion file
|
Fri, 05 Nov 2010 23:06:27 -0400 |
James Bergstra |
pca - added some assertions
|
Fri, 05 Nov 2010 13:29:07 -0400 |
James Bergstra |
icml07 - added comment
|
Wed, 03 Nov 2010 12:58:04 -0400 |
James Bergstra |
icml07 loaders works
|
Wed, 03 Nov 2010 12:49:24 -0400 |
James Bergstra |
adding icml07 dataset
|
Wed, 03 Nov 2010 12:49:12 -0400 |
James Bergstra |
dataset_ops/tinyimages - added an img_shape optional flag to save_filters fns
|
Thu, 28 Oct 2010 16:15:47 -0400 |
James Bergstra |
mcRBM - added to docstring
|
Thu, 21 Oct 2010 23:12:38 -0400 |
James Bergstra |
mcRBM - added norm_doctoring parameter
|
Thu, 21 Oct 2010 14:47:49 -0400 |
James Bergstra |
mcRBM - changes to creation of topo_P matrix
|
Thu, 21 Oct 2010 11:43:36 -0400 |
James Bergstra |
mcRBM - fixed error in normalization of quadratic term
|
Thu, 21 Oct 2010 11:42:36 -0400 |
James Bergstra |
removed checkpoint draft
|
Wed, 27 Oct 2010 10:37:18 -0400 |
Frederic Bastien |
don't consider __version__ when it is none.
|
Wed, 27 Oct 2010 10:36:55 -0400 |
Frederic Bastien |
make version detection work around a strange behavior of imp.find_module
|
Wed, 27 Oct 2010 10:33:16 -0400 |
Frederic Bastien |
give the NotImplementedError a string message
|
Tue, 26 Oct 2010 16:34:20 -0400 |
Frederic Bastien |
record_version work with module that are not checkout and have __version__ defined(as numpy).
|
Fri, 22 Oct 2010 11:11:51 -0400 |
gdesjardins |
Fixed comments in leapfrog method
|
Fri, 22 Oct 2010 10:58:57 -0400 |
gdesjardins |
HMC: perform half-step on velocity first (instead of position).
|
Thu, 21 Oct 2010 16:18:52 -0400 |
Olivier Delalleau |
Reply to Razvan
|
Thu, 21 Oct 2010 14:36:36 -0400 |
Razvan Pascanu |
comment on dataset proposal by Olivier
|
Thu, 21 Oct 2010 14:03:12 -0400 |
Olivier Delalleau |
Added a comment on the dataset vs. task issue
|
Thu, 21 Oct 2010 13:00:49 -0400 |
Olivier Delalleau |
Added proposal for dataset API as discussed on pylearn-dev
|
Mon, 18 Oct 2010 19:31:17 -0400 |
James Bergstra |
adding new idea to arch_src
|
Mon, 18 Oct 2010 15:01:00 -0400 |
James Bergstra |
removed sandbox code from test_mcRBM
|
Mon, 18 Oct 2010 14:58:52 -0400 |
James Bergstra |
merge
|
Mon, 18 Oct 2010 14:58:39 -0400 |
James Bergstra |
modifs to test_mcRBM to reproduce mcRBM classif results
|
Mon, 18 Oct 2010 08:53:08 -0400 |
James Bergstra |
added test idea to test_mcRBM
|
Thu, 14 Oct 2010 23:55:55 -0400 |
gdesjardins |
merge
|
Thu, 14 Oct 2010 23:55:41 -0400 |
gdesjardins |
small changes to mode dataset (used for tempering work)
|
Thu, 14 Oct 2010 23:54:28 -0400 |
gdesjardins |
forgot to commit... 2 versions: one for /data/lisa6 the other for
|
Thu, 14 Oct 2010 13:33:33 -0400 |
Frederic Bastien |
added function doc and made the interface more constant between fct.
|
Wed, 13 Oct 2010 15:05:16 -0400 |
Frederic Bastien |
make function documentation more clear.
|