/
train_val_test.py
160 lines (132 loc) · 6.65 KB
/
train_val_test.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
# Author: duguiming
# Description: 训练、验证和测试
# Date: 2020-4-8
import os
import sys
import time
import tensorflow as tf
from tensorflow.contrib.crf import viterbi_decode
from data_helper import tag2label, batch_yield, pad_sequences
from eval import conlleval
from utils import get_entity
def feed_data(model, seqs, labels=None, lr=None, dropout=None):
word_ids, seq_len_list = pad_sequences(seqs, pad_mark=0)
feed_dict = {model.word_ids: word_ids,
model.sequence_lengths: seq_len_list}
if labels is not None:
labels_, _ = pad_sequences(labels, pad_mark=0)
feed_dict[model.labels] = labels_
if lr is not None:
feed_dict[model.lr_pl] = lr
if dropout is not None:
feed_dict[model.dropout_pl] = dropout
return feed_dict, seq_len_list
def evaluate_(session, model, val_data, word2id, config):
label_list, seq_len_list = [], []
for seqs, labels in batch_yield(val_data, config.batch_size, word2id, tag2label, shuffle=False):
feed_dict, seq_len_list_ = feed_data(model, seqs, dropout=1.0)
if config.CRF:
logits, transition_params = session.run([model.logits, model.transition_params],
feed_dict=feed_dict)
label_list_ = []
for logit, seq_len in zip(logits, seq_len_list_):
viterbi_seq, _ = viterbi_decode(logit[:seq_len], transition_params)
label_list_.append(viterbi_seq)
else:
label_list_ = session.run(model.labels_softmax_, feed_dict=feed_dict)
label_list.extend(label_list_)
seq_len_list.extend(seq_len_list_)
return label_list, seq_len_list
def evaluate(label_list, seq_len_list, data, config, epoch=None):
label2tag = {}
for tag, label in tag2label.items():
label2tag[label] = tag if label != 0 else label
model_predict = []
for label_, (sent, tag) in zip(label_list, data):
tag_ = [label2tag[label__] for label__ in label_]
sent_res = []
if len(label_) != len(sent):
print(sent)
print(len(label_))
print(tag)
for i in range(len(sent)):
sent_res.append([sent[i], tag[i], tag_[i]])
model_predict.append(sent_res)
epoch_num = str(epoch + 1) if epoch != None else 'test'
label_path = os.path.join(config.result_path, 'label_' + epoch_num)
metric_path = os.path.join(config.result_path, 'result_metric_' + epoch_num)
for _ in conlleval(model_predict, label_path, metric_path):
config.logger.info(_)
def demo_one(session, model, config, word2id, sent):
label_list = []
for seqs, labels in batch_yield(sent, config.batch_size, word2id, tag2label, shuffle=False):
feed_dict, seq_len_list_ = feed_data(model, seqs, dropout=1.0)
if config.CRF:
logits, transition_params = session.run([model.logits, model.transition_params],
feed_dict=feed_dict)
label_list_ = []
for logit, seq_len in zip(logits, seq_len_list_):
viterbi_seq, _ = viterbi_decode(logit[:seq_len], transition_params)
label_list_.append(viterbi_seq)
else:
label_list_ = session.run(model.labels_softmax_, feed_dict=feed_dict)
label_list.extend(label_list_)
label2tag = {}
for tag, label in tag2label.items():
label2tag[label] = tag if label != 0 else label
tag = [label2tag[label] for label in label_list[0]]
return tag
def train(model, config, train_data, val_data, word2id):
session = tf.Session()
session.run(tf.global_variables_initializer())
tf.summary.scalar('loss', model.loss)
merged_summary = tf.summary.merge_all()
writer = tf.summary.FileWriter(config.summary_path)
saver = tf.train.Saver(tf.global_variables())
print("Training and evaling...")
for epoch in range(config.num_epochs):
num_batches = (len(train_data) + config.batch_size - 1) // config.batch_size
start_time = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())
batches = batch_yield(train_data, config.batch_size, word2id, tag2label, shuffle=config.shuffle)
for step, (seqs, labels) in enumerate(batches):
sys.stdout.write(' processing: {} batch / {} batches.'.format(step + 1, num_batches) + '\r')
step_num = epoch * num_batches + step + 1
feed_dict, _ = feed_data(model, seqs, labels, config.lr, config.dropout_keep_prob)
_, loss_train, summary, step_num_ = session.run([model.train_op, model.loss, merged_summary, model.global_step],
feed_dict=feed_dict)
if step + 1 == 1 or (step + 1) % 300 == 0 or step + 1 == num_batches:
config.logger.info('{} epoch {}, step {}, loss: {:.4}, global_step: {}'.format(start_time, epoch + 1, step + 1,
loss_train, step_num))
writer.add_summary(summary, step_num_)
if step + 1 == num_batches:
saver.save(session, config.model_path, global_step=step_num)
config.logger.info('=========== validation ===========')
label_list_dev, seq_len_list_dev = evaluate_(session, model, val_data, word2id, config)
evaluate(label_list_dev, seq_len_list_dev, val_data, config, epoch)
def test(model, config, test_data, word2id):
ckpt_file = tf.train.latest_checkpoint(config.model_dir)
session = tf.Session()
session.run(tf.global_variables_initializer())
saver = tf.train.Saver()
saver.restore(sess=session, save_path=ckpt_file) # 读取保存的模型
config.logger.info('=========== test ===========')
label_list_dev, seq_len_list_dev = evaluate_(session, model, test_data, word2id, config)
evaluate(label_list_dev, seq_len_list_dev, test_data, config)
def demo(model, config, word2id):
ckpt_file = tf.train.latest_checkpoint(config.model_dir)
session = tf.Session()
session.run(tf.global_variables_initializer())
saver = tf.train.Saver()
saver.restore(sess=session, save_path=ckpt_file) # 读取保存的模型
while True:
print('Please input your sentence:')
demo_sent = input()
if demo_sent == '' or demo_sent.isspace():
print('See you next time!')
break
else:
demo_sent = list(demo_sent.strip())
demo_data = [(demo_sent, ['O'] * len(demo_sent))]
tag = demo_one(session, model, config, word2id, demo_data)
PER, LOC, ORG = get_entity(tag, demo_sent)
print('PER: {}\nLOC: {}\nORG: {}'.format(PER, LOC, ORG))