Mercurial > pylearn
comparison algorithms/tests/test_daa.py @ 533:de974b4fc4ea
Bugfix in pylearn.embeddings.length()
author | Joseph Turian <turian@gmail.com> |
---|---|
date | Tue, 18 Nov 2008 03:25:54 -0500 |
parents | 4fb6f7320518 |
children |
comparison
equal
deleted
inserted
replaced
532:34ee3aff3e8f | 533:de974b4fc4ea |
---|---|
26 for l in range(3): | 26 for l in range(3): |
27 for i in range(10): | 27 for i in range(10): |
28 model.local_update[l]([[0, 1, 0, 1]]) | 28 model.local_update[l]([[0, 1, 0, 1]]) |
29 model.local_update[l]([[1, 0, 1, 0]]) | 29 model.local_update[l]([[1, 0, 1, 0]]) |
30 | 30 |
31 for i in range(1): | 31 for i in range(10): |
32 model.update([[0, 1, 0, 1]], [[1]]) | 32 model.update([[0, 1, 0, 1]], [[1]]) |
33 model.update([[1, 0, 1, 0]], [[0]]) | 33 model.update([[1, 0, 1, 0]], [[0]]) |
34 print model.classify([[0, 1, 0, 1]]) | 34 print model.classify([[0, 1, 0, 1]]) |
35 print model.classify([[1, 0, 1, 0]]) | 35 print model.classify([[1, 0, 1, 0]]) |
36 | 36 |
39 | 39 |
40 ndaa = 3 | 40 ndaa = 3 |
41 daa = models.Stacker([(models.SigmoidXEDenoisingAA, 'hidden')] * ndaa + [(pylearn.algorithms.logistic_regression.Module_Nclass, 'pred')], | 41 daa = models.Stacker([(models.SigmoidXEDenoisingAA, 'hidden')] * ndaa + [(pylearn.algorithms.logistic_regression.Module_Nclass, 'pred')], |
42 regularize = False) | 42 regularize = False) |
43 | 43 |
44 model = daa.make([4, 20, 20, 20, 10], | 44 model = daa.make([4] + [20] * ndaa + [10], |
45 lr = 0.01, | 45 lr = 0.01, |
46 mode = mode, | 46 mode = mode, |
47 seed = 10) | 47 seed = 10) |
48 | 48 |
49 model.layers[0].noise_level = 0.3 | 49 for l in range(ndaa): model.layers[l].noise_level = 0.3 |
50 model.layers[1].noise_level = 0.3 | |
51 model.layers[2].noise_level = 0.3 | |
52 | 50 |
53 for l in range(3): | 51 instances = [([[0, 1, 0, 1]], [1]), ([[1, 0, 1, 0]], [0])] |
52 | |
53 for l in range(ndaa): | |
54 for i in range(10): | 54 for i in range(10): |
55 model.local_update[l]([[0, 1, 0, 1]]) | 55 for (input, output) in instances: |
56 model.local_update[l]([[1, 0, 1, 0]]) | 56 model.local_update[l](input) |
57 | 57 |
58 for i in range(1): | 58 for i in range(10): |
59 model.update([[0, 1, 0, 1]], [1]) | 59 for (input, output) in instances: |
60 model.update([[1, 0, 1, 0]], [0]) | 60 # model.update(input, output) |
61 print "OLD:", | |
62 print model.validate(input, output) | |
63 oldloss = model.update(input, output) | |
64 print oldloss | |
65 print "NEW:" | |
66 print model.validate(input, output) | |
67 print | |
68 | |
61 print model.apply([[0, 1, 0, 1]]) | 69 print model.apply([[0, 1, 0, 1]]) |
62 print model.apply([[1, 0, 1, 0]]) | 70 print model.apply([[1, 0, 1, 0]]) |
63 | 71 |
64 | 72 |
65 | 73 |