-
Notifications
You must be signed in to change notification settings - Fork 3
/
train_network.py
202 lines (155 loc) · 7.58 KB
/
train_network.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
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
'''
Training interface for Neural network model to detect and classify TLINKS between entities.
'''
import sys
import os
from code.config import env_paths
if env_paths()["PY4J_DIR_PATH"] is None:
sys.exit("PY4J_DIR_PATH environment variable not specified")
import argparse
import glob
import cPickle
import json
from code.learning.network import Network
from code.notes.TimeNote import TimeNote
from code.learning.word2vec import load_word2vec_binary
from keras.models import model_from_json
from keras.models import load_model
from keras.callbacks import ModelCheckpoint, EarlyStopping
from keras.optimizers import Adam, SGD
N_CLASSES = 13
def main():
'''
Process command line arguments and then generate trained models (One for detection of links, one for classification)
'''
parser = argparse.ArgumentParser()
parser.add_argument("train_dir",
help="Directory containing training annotations")
parser.add_argument("model_destination",
help="Where to store the trained model")
parser.add_argument("newsreader_annotations",
help="Where newsreader pipeline parsed file objects go")
parser.add_argument("--val_dir",
default=None,
help="Directory containing validation annotations")
parser.add_argument("--load_model",
action='store_true',
default=False,
help="Load saved model and resume training from there")
parser.add_argument("--no_val",
action='store_true',
default=False,
help="No validation. Use all training data to train.")
parser.add_argument("--pair_type",
default='both',
help="specify the entity type to train: intra, cross or both")
parser.add_argument("--pair_ordered",
action='store_true',
default=False,
help="Only consider pairs in their narrative order (order in text)")
parser.add_argument("--nolink_ratio",
default=1.0,
type=float,
help="no link downsampling ratio. e.g. 0.5 means # of nolinks are 50% of # positive tlinks")
args = parser.parse_args()
assert args.pair_type in ('intra', 'cross', 'both', 'dct')
# validate file paths
if os.path.isdir(args.newsreader_annotations) is False:
sys.exit("invalid path for time note dir")
if os.path.isdir(args.train_dir) is False:
sys.exit("invalid path to directory containing training data")
if os.path.isdir(os.path.dirname(args.model_destination)) is False:
sys.exit("directory for model destination does not exist")
# get files in directory
files = glob.glob(os.path.join(args.train_dir, '*'))
gold_files = []
tml_files = []
for f in files:
if "E3input" in f:
tml_files.append(f)
elif f.endswith('.tml'):
gold_files.append(f)
gold_files.sort()
tml_files.sort()
if args.val_dir is None:
val_files = None
else:
val_files = glob.glob(os.path.join(args.val_dir, '*'))
val_files.sort()
model_destination = os.path.join(args.model_destination, args.pair_type) + '/'
if not os.path.exists(model_destination):
os.makedirs(model_destination)
if args.no_val:
earlystopping = EarlyStopping(monitor='loss', patience=30, verbose=0, mode='auto')
checkpoint = ModelCheckpoint(model_destination + 'model.h5', monitor='loss', save_best_only=True)
else:
earlystopping = EarlyStopping(monitor='loss', patience=30, verbose=0, mode='auto')
checkpoint = ModelCheckpoint(model_destination + 'model.h5', monitor='loss', save_best_only=True)
if args.load_model:
try:
NNet = load_model(model_destination + 'model.h5')
except:
NNet = model_from_json(open(model_destination + '.arch.json').read())
opt = SGD(lr=0.003, momentum=0.9, decay=0.0, nesterov=False)
NNet.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])
NNet.load_weights(model_destination + '.weights.h5')
else:
NNet = None
NN, history = trainNetwork(gold_files, val_files, args.newsreader_annotations, args.pair_type, ordered=args.pair_ordered,
no_val=args.no_val, nolink_ratio=args.nolink_ratio, callbacks=[checkpoint, earlystopping], train_dir=args.train_dir)
architecture = NN.to_json()
open(model_destination + '.arch.json', "wb").write(architecture)
NN.save_weights(model_destination + '.weights.h5')
json.dump(history, open(model_destination + 'training_history.json', 'w'))
def basename(name):
name = os.path.basename(name)
name = name.replace('.TE3input', '')
name = name.replace('.tml', '')
return name
def get_notes(files, newsreader_dir):
if not files:
return None
notes = []
for i, tml in enumerate(files):
if i % 10 == 0:
print 'processing file {}/{} {}'.format(i + 1, len(files), tml)
if os.path.isfile(os.path.join(newsreader_dir, basename(tml) + ".parsed.pickle")):
tmp_note = cPickle.load(open(os.path.join(newsreader_dir, basename(tml) + ".parsed.pickle"), "rb"))
else:
tmp_note = TimeNote(tml, tml)
cPickle.dump(tmp_note, open(newsreader_dir + "/" + basename(tml) + ".parsed.pickle", "wb"))
notes.append(tmp_note)
return notes
def trainNetwork(gold_files, val_files, newsreader_dir, pair_type, ordered=False, no_val=False, nolink_ratio=1.0, callbacks=[], train_dir='./'):
'''
Train a neural network for classification of temporal realtions.
'''
print "Called trainNetwork"
global N_CLASSES
if not os.path.isfile(train_dir+'training_data.pkl'):
notes = get_notes(gold_files, newsreader_dir)
if not no_val:
val_notes = get_notes(val_files, newsreader_dir)
network = Network()
print "loading word vectors..."
network.word_vectors = load_word2vec_binary(os.environ["TEA_PATH"] + '/GoogleNews-vectors-negative300.bin', verbose=0)
if os.path.isfile(train_dir+'training_data.pkl'):
print "loading pkl file... this may take over 10 minutes"
training_data = cPickle.load(open(train_dir+'training_data.pkl'))
print "training data size:", training_data[0].shape, training_data[1].shape, len(training_data[2])
else:
# nolink_ration = # no tlink cases / # tlink cases
training_data = network._get_training_input(notes, pair_type=pair_type, nolink_ratio=nolink_ratio, shuffle=True, ordered=ordered)
print "training data size:", training_data[0].shape, training_data[1].shape, len(training_data[2])
if not no_val and val_notes is not None:
val_data = network._get_test_input(val_notes, pair_type=pair_type, ordered=ordered)
print "validation data size:", val_data[0].shape, val_data[1].shape, len(val_data[2])
else:
val_data = None
del network.word_vectors
NNet, history = network.train_model(None, epochs=200, training_input=training_data, val_input=val_data, no_val=no_val, weight_classes=False, batch_size=100,
encoder_dropout=0, decoder_dropout=0.5, input_dropout=0.6, reg_W=0, reg_B=0, reg_act=0, LSTM_size=256,
dense_size=100, maxpooling=True, data_dim=300, max_len='auto', nb_classes=N_CLASSES, callbacks=callbacks, ordered=ordered)
return NNet, history
if __name__ == "__main__":
main()