Mercurial > ift6266

annotate datasets/ftfile.py @ 170:89a725d332ae

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
moved deepmlp code into baseline/deep_mlp
author Dumitru Erhan <dumitru.erhan@gmail.com>
date Fri, 26 Feb 2010 14:28:48 -0500
parents 4b28d7382dbf
children 954185d6002a
rev   line source
163
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
1 from pylearn.io.filetensor import _read_header, _prod
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
2 import numpy
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
3 from dataset import DataSet
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
4 from dsetiter import DataIterator
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
5
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
6 class FTFile(object):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
7 def __init__(self, fname):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
8 r"""
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
9 Tests:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
10 >>> f = FTFile('/data/lisa/data/nist/by_class/digits/digits_test_labels.ft')
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
11 """
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
12 self.file = open(fname, 'rb')
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
13 self.magic_t, self.elsize, _, self.dim, _ = _read_header(self.file, False)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
14 self.size = self.dim[0]
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
15
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
16 def skip(self, num):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
17 r"""
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
18 Skips `num` items in the file.
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
19
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
20 Tests:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
21 >>> f = FTFile('/data/lisa/data/nist/by_class/digits/digits_test_labels.ft')
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
22 >>> f.size
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
23 58646
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
24 >>> f.elsize
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
25 4
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
26 >>> f.file.tell()
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
27 20
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
28 >>> f.skip(1000)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
29 >>> f.file.tell()
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
30 4020
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
31 >>> f.size
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
32 57646
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
33 """
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
34 if num >= self.size:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
35 self.size = 0
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
36 else:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
37 self.size -= num
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
38 f_start = self.file.tell()
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
39 self.file.seek(f_start + (self.elsize * _prod(self.dim[1:]) * num))
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
40
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
41 def read(self, num):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
42 r"""
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
43 Reads `num` elements from the file and return the result as a
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
44 numpy matrix. Last read is truncated.
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
45
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
46 Tests:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
47 >>> f = FTFile('/data/lisa/data/nist/by_class/digits/digits_test_labels.ft')
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
48 >>> f.read(1)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
49 array([6], dtype=int32)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
50 >>> f.read(10)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
51 array([7, 4, 7, 5, 6, 4, 8, 0, 9, 6], dtype=int32)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
52 >>> f.skip(58630)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
53 >>> f.read(10)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
54 array([9, 2, 4, 2, 8], dtype=int32)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
55 >>> f.read(10)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
56 array([], dtype=int32)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
57 >>> f = FTFile('/data/lisa/data/nist/by_class/digits/digits_test_data.ft')
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
58 >>> f.read(1)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
59 array([[0, 0, 0, ..., 0, 0, 0]], dtype=uint8)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
60 """
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
61 if num > self.size:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
62 num = self.size
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
63 self.dim[0] = num
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
64 self.size -= num
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
65 return numpy.fromfile(self.file, dtype=self.magic_t, count=_prod(self.dim)).reshape(self.dim)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
66
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
67 class FTSource(object):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
68 def __init__(self, file, skip=0, size=None):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
69 r"""
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
70 Create a data source from a possible subset of a .ft file.
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
71
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
72 Parameters:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
73 `file` (string) -- the filename
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
74 `skip` (int, optional) -- amount of examples to skip from the start of the file
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
75 `size` (int, optional) -- truncates number of examples read (after skipping)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
76
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
77 Tests:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
78 >>> s = FTSource('/data/lisa/data/nist/by_class/digits/digits_test_data.ft')
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
79 >>> s = FTSource('/data/lisa/data/nist/by_class/digits/digits_test_data.ft', size=1000)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
80 >>> s = FTSource('/data/lisa/data/nist/by_class/digits/digits_test_data.ft', skip=10)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
81 >>> s = FTSource('/data/lisa/data/nist/by_class/digits/digits_test_data.ft', skip=100, size=120)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
82 """
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
83 self.file = file
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
84 self.skip = skip
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
85 self.size = size
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
86
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
87 def open(self):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
88 r"""
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
89 Returns an FTFile that corresponds to this dataset.
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
90
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
91 Tests:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
92 >>> s = FTSource('/data/lisa/data/nist/by_class/digits/digits_test_data.ft')
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
93 >>> f = s.open()
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
94 >>> s = FTSource('/data/lisa/data/nist/by_class/digits/digits_test_data.ft', size=1)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
95 >>> len(s.open().read(2))
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
96 1
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
97 >>> s = FTSource('/data/lisa/data/nist/by_class/digits/digits_test_data.ft', skip=57646)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
98 >>> s.open().size
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
99 1000
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
100 >>> s = FTSource('/data/lisa/data/nist/by_class/digits/digits_test_data.ft', skip=57646, size=1)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
101 >>> s.open().size
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
102 1
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
103 """
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
104 f = FTFile(self.file)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
105 if self.skip != 0:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
106 f.skip(self.skip)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
107 if self.size is not None and self.size < f.size:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
108 f.size = self.size
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
109 return f
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
110
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
111 class FTData(object):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
112 r"""
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
113 This is a list of FTSources.
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
114 """
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
115 def __init__(self, datafiles, labelfiles, skip=0, size=None):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
116 self.inputs = [FTSource(f, skip, size) for f in datafiles]
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
117 self.outputs = [FTSource(f, skip, size) for f in labelfiles]
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
118
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
119 def open_inputs(self):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
120 return [f.open() for f in self.inputs]
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
121
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
122 def open_outputs(self):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
123 return [f.open() for f in self.outputs]
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
124
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
125
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
126 class FTDataSet(DataSet):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
127 def __init__(self, train_data, train_lbl, test_data, test_lbl, valid_data=None, valid_lbl=None):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
128 r"""
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
129 Defines a DataSet from a bunch of files.
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
130
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
131 Parameters:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
132 `train_data` -- list of train data files
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
133 `train_label` -- list of train label files (same length as `train_data`)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
134 `test_data`, `test_labels` -- same thing as train, but for
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
135 test. The number of files
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
136 can differ from train.
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
137 `valid_data`, `valid_labels` -- same thing again for validation.
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
138 (optional)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
139
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
140 If `valid_data` and `valid_labels` are not supplied then a sample
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
141 approximately equal in size to the test set is taken from the train
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
142 set.
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
143 """
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
144 if valid_data is None:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
145 total_valid_size = sum(FTFile(td).size for td in test_data)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
146 valid_size = total_valid_size/len(train_data)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
147 self._train = FTData(train_data, train_lbl, skip=valid_size)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
148 self._valid = FTData(train_data, train_lbl, size=valid_size)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
149 else:
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
150 self._train = FTData(train_data, train_lbl)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
151 self._valid = FTData(valid_data, valid_lbl)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
152 self._test = FTData(test_data, test_lbl)
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
153
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
154 def _return_it(self, batchsize, bufsize, ftdata):
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
155 return zip(DataIterator(ftdata.open_inputs(), batchsize, bufsize),
4b28d7382dbf Add inital implementation of datasets.
Arnaud Bergeron <abergeron@gmail.com>
parents:
diff changeset
156 DataIterator(ftdata.open_outputs(), batchsize, bufsize))