-
Notifications
You must be signed in to change notification settings - Fork 1
/
lstmscript_config.py
184 lines (138 loc) · 4.82 KB
/
lstmscript_config.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
#! /usr/bin/env python
# -*- coding: utf-8 -*-
'''
Author: Xihao Liang
Created: 2016.03.16
'''
import os
import sys
reload(sys)
sys.setdefaultencoding('utf8')
import cPickle
from optparse import OptionParser
import datica
import validatica
from const import N_EMO, DIR_MODEL, DIR_TEST, DIR_DATA, DIR_UNIGRAM
from lstm import LstmClassifier
import wemb_rand
from wordembedder import WordEmbedder
class LstmScriptRand:
def __init__(self):
self.init_default_options()
self.add_extra_options()
def init_default_options(self):
parser = OptionParser()
# necessary
parser.add_option('-p', '--prefix', action='store', type = 'str', dest='prefix')
parser.add_option('-c', '--fname_config', action='store', type = 'str', dest='fname_config')
# optional
parser.add_option('-r', '--resume', action='store_true', dest='resume', default = False)
# debug
parser.add_option('-n', '--n_samples', action='store', dest='n_samples', default = None)
# especially for gpu
parser.add_option('-b', '--batch_size', action='store', type='int', dest='batch_size', default = 16)
self.optparser = parser
def add_extra_options(self):
self.optparser.add_option('-d', '--dim_proj', action='store', type = 'int', dest='dim_proj')
def init_folder(self):
'''
mkdir if the necessary folders do not exist
'''
if not os.path.isdir(DIR_MODEL):
os.mkdir(DIR_MODEL)
if not os.path.isdir(DIR_TEST):
os.mkdir(DIR_TEST)
def init_embedder(self, dataset, fname_embedder):
'''
initialize the embedder by load it from file if available
or build the model by the dataset and save it
'''
if os.path.exists(fname_embedder):
print >> sys.stderr, 'embedding model %s found and loaded'%(fname_embedder)
return WordEmbedder.load(fname_embedder)
else:
def x_iterator(dataset):
for set_x, set_y in dataset:
for x in set_x:
yield x
embedder = WordEmbedder(*wemb_rand.build(x_iterator(dataset), self.opts.dim_proj))
embedder.dump(fname_embedder)
return embedder
def prepare_input(self, dataset, embedder):
'''
turn sequences of string into list of vectors
'''
def index_set(set_x_y):
x, y = set_x_y
new_x = [embedder.index(xi) for xi in x]
print len(new_x)
return (new_x, y)
train, test, valid = dataset
new_dataset = (index_set(train), index_set(test), index_set(valid))
return new_dataset, embedder.get_Wemb()
def run(self):
'''
the function to launch the script
'''
################### Preparation of Variables #######################
print >> sys.stderr, 'lstmscript.run: [info] preparing variables ... ',
opts, args = self.optparser.parse_args() # initialized in init_default_options
self.opts = opts # shared by self. for customized function
datalen = opts.n_samples
dim_proj = opts.dim_proj
prefix = opts.prefix
fname_test = DIR_TEST + '%s_test.pkl'%(prefix)
fname_model = DIR_MODEL + '%s_model.npz'%(prefix)
fname_embedder = DIR_MODEL + '%s_embedder.pkl'%(prefix)
print >> sys.stderr, 'Done'
#################### Preparation of Input ##############
print >> sys.stderr, 'lstmscript.run: [info] loading dataset ... ',
eids_list = datica.load_config(opts.fname_config)
dataset = datica.load_by_config(DIR_UNIGRAM, eids_list, datalen)
n_emo = len(eids_list)
print >> sys.stderr, 'Done'
print >> sys.stderr, 'lstmscript.run: [info] initialization of embedder'
embedder = self.init_embedder(dataset, fname_embedder)
print >> sys.stderr, 'lstmscript.run: [info] preparing input'
dataset, Wemb = self.prepare_input(dataset, embedder)
#################### Preparation for Output ############
self.init_folder()
#################### Training ##########################
print >> sys.stderr, 'lstmscript.run: [info] start training'
classifier = LstmClassifier()
if not opts.resume:
res = classifier.train(
dataset = dataset,
Wemb = Wemb,
ydim = n_emo,
fname_model = fname_model,
batch_size = opts.batch_size,
valid_batch_size = opts.batch_size,
)
elif not os.path.exists(fname_model):
print >> sys.stderr, 'model %s not found'%(fname_model)
return
else:
res = classifier.train(
dataset = dataset,
Wemb = Wemb,
ydim = n_emo,
fname_model = fname_model,
reload_model = True,
batch_size = opts.batch_size,
valid_batch_size = opts.batch_size,
)
#else:
# print >> sys.stderr, 'lstm model %s found and loaded'%(fname_model)
# classifier.load(fname_model)
###################### Test ##############################
test_x, test_y = dataset[2]
preds_prob = classifier.classify(test_x)
cPickle.dump((test_y, preds_prob), open(fname_test, 'w'))
###################### Report ############################
validatica.report(test_y, preds_prob, DIR_TEST + prefix)
def main():
script = LstmScriptRand()
script.run()
if __name__ == '__main__':
main()