aboutsummaryrefslogtreecommitdiff
path: root/train.py
Commit message (Collapse)AuthorAgeFilesLines
* New seed for all the computationsAdeB2015-07-021-0/+2
|
* Step rule & dropout params cleanupAlex Auvolat2015-07-021-1/+6
|
* Unify parameters for joint_simple_tgtcls_111_cswdtx_bigger{,_dropout}Alex Auvolat2015-07-021-1/+4
|
* Load model parameters only when specified in the configAdeB2015-06-241-1/+1
|
* New configs. training step rule out of train.pyAdeB2015-06-241-3/+2
|
* Correct Plot importAdeB2015-06-211-1/+1
|
* Import cPickleAdeB2015-06-211-0/+1
|
* Updates for Blocks compatibilityAlex Auvolat2015-06-211-33/+43
|
* Fix RNN validationÉtienne Simon2015-06-121-3/+9
|
* Add TaxiRemoveTestOnlyClients ; custom dumpmanager enabling multiprocessingAlex Auvolat2015-06-111-4/+29
|
* Fix validation cost computation to not use dropout/noise regularizationAlex Auvolat2015-05-221-4/+2
|
* Add ElementwiseRemoveNotFinite step rule.Alex Auvolat2015-05-221-2/+35
|
* Merge branch 'alex' into newAlex Auvolat2015-05-211-0/+1
|\ | | | | | | | | | | Conflicts: model/joint_simple_mlp_tgtcls.py train.py
| * Model changesAlex Auvolat2015-05-211-3/+3
| |
* | Refactor models, clean the code and separate training from testing.Étienne Simon2015-05-181-117/+57
|/
* Add NaN protectionAlex Auvolat2015-05-131-10/+18
|
* Add model for a network that predicts both time and destination.Alex Auvolat2015-05-081-17/+23
|
* Fusion AddFirstK and AddLastK transformers, and add 'input_time' field.Alex Auvolat2015-05-081-6/+4
|
* Add target class based model for time prediction (seems to work)Alex Auvolat2015-05-071-3/+3
|
* Add Bokeh plottingAlex Auvolat2015-05-061-0/+2
|
* Merge branch 'master' of github.com:adbrebs/taxiAdeB2015-05-051-30/+21
|\
| * Clean data module and generalize use of hdf5.Étienne Simon2015-05-051-30/+21
| |
* | New hyperparameters. Training error is monitored.AdeB2015-05-051-2/+3
|/
* Add models for time predictioAdd models for time predictionAlex Auvolat2015-05-051-11/+16
|
* Add date/time transformer and new model that uses itAlex Auvolat2015-05-051-7/+11
|
* Refactor architecture so that embedding sizes can be easily changed.Alex Auvolat2015-05-051-18/+20
|
* New config (added a hidden layer), small changes to train.pyAlex Auvolat2015-05-051-2/+2
|
* Add model with predefined target classesAlex Auvolat2015-05-041-3/+4
|
* Restructure model & configAlex Auvolat2015-05-041-0/+139