Mercurial > ift6266

comparison deep/stacked_dae/v2/config.py.example @ 239:42005ec87747

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Mergé (manuellement) les changements de Sylvain pour utiliser le code de dataset d'Arnaud, à cette différence près que je n'utilse pas les givens. J'ai probablement une approche différente pour limiter la taille du dataset dans mon débuggage, aussi.
author fsavard
date Mon, 15 Mar 2010 18:30:21 -0400
parents
children
comparison
equal deleted inserted replaced
238:9fc641d7adda 239:42005ec87747
1 '''
2 These are parameters used by nist_sda.py. They'll end up as globals in there.
3
4 Rename this file to config.py and configure as needed.
5 DON'T add the renamed file to the repository, as others might use it
6 without realizing it, with dire consequences.
7 '''
8
9 # Set this to True when you want to run cluster tests, ie. you want
10 # to run on the cluster, many jobs, but want to reduce the training
11 # set size and the number of epochs, so you know everything runs
12 # fine on the cluster.
13 # Set this PRIOR to inserting your test jobs in the DB.
14 TEST_CONFIG = False
15
16 NIST_ALL_LOCATION = '/data/lisa/data/nist/by_class/all'
17 NIST_ALL_TRAIN_SIZE = 649081
18 # valid et test =82587 82587
19
20 # change "sandbox" when you're ready
21 JOBDB = 'postgres://ift6266h10@gershwin/ift6266h10_sandbox_db/yourtablenamehere'
22 EXPERIMENT_PATH = "ift6266.deep.stacked_dae.v2.nist_sda.jobman_entrypoint"
23
24 # reduce training set to that many examples
25 REDUCE_TRAIN_TO = None
26 # that's a max, it usually doesn't get to that point
27 MAX_FINETUNING_EPOCHS = 1000
28 # number of minibatches before taking means for valid error etc.
29 REDUCE_EVERY = 100
30
31 if TEST_CONFIG:
32 REDUCE_TRAIN_TO = 1000
33 MAX_FINETUNING_EPOCHS = 2
34 REDUCE_EVERY = 10
35
36
37 # This is to configure insertion of jobs on the cluster.
38 # Possible values the hyperparameters can take. These are then
39 # combined with produit_cartesien_jobs so we get a list of all
40 # possible combinations, each one resulting in a job inserted
41 # in the jobman DB.
42 JOB_VALS = {'pretraining_lr': [0.1, 0.01],#, 0.001],#, 0.0001],
43 'pretraining_epochs_per_layer': [10,20],
44 'hidden_layers_sizes': [300,800],
45 'corruption_levels': [0.1,0.2,0.3],
46 'minibatch_size': [20],
47 'max_finetuning_epochs':[MAX_FINETUNING_EPOCHS],
48 'finetuning_lr':[0.1, 0.01], #0.001 was very bad, so we leave it out
49 'num_hidden_layers':[2,3]}
50
51 # Just useful for tests... minimal number of epochs
52 # (This is used when running a single job, locally, when
53 # calling ./nist_sda.py test_jobman_entrypoint
54 DEFAULT_HP_NIST = DD({'finetuning_lr':0.1,
55 'pretraining_lr':0.1,
56 'pretraining_epochs_per_layer':2,
57 'max_finetuning_epochs':2,
58 'hidden_layers_sizes':800,
59 'corruption_levels':0.2,
60 'minibatch_size':20,
61 'reduce_train_to':10000,
62 'num_hidden_layers':1})
63
64