CMeKG_tools/train_ner.py at main · king-yyf/CMeKG_tools · GitHub

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
# -*- coding: utf-8 -*-

from tqdm import tqdm
import torch
import torch.nn as nn
from torch.autograd import Variable
import torch.optim as optim
from torch.utils.data import TensorDataset
from torch.utils.data import DataLoader

from utils import load_vocab, load_data, recover_label, get_ner_fmeasure, save_model, load_model
from ner_constant import *
from model_ner import BERT_LSTM_CRF


print('device',device)
# if torch.cuda.is_available():
#     device = torch.device("cuda", 2)
#     print('device',device)
#     use_cuda = True
# else:
#     device = torch.device("cpu")
#     use_cuda = False
vocab = load_vocab(vocab_file)
vocab_reverse = {v:k for k, v in vocab.items()}

print('max_length',max_length)


train_data = load_data(train_file, max_length=max_length, label_dic=l2i_dic, vocab=vocab)
train_ids = torch.LongTensor([temp.input_id for temp in train_data[1500:]])
train_masks = torch.LongTensor([temp.input_mask for temp in train_data[1500:]])
train_tags = torch.LongTensor([temp.label_id for temp in train_data[1500:]])
train_lenghts = torch.LongTensor([temp.lenght for temp in train_data[1500:]])
train_dataset = TensorDataset(train_ids, train_masks, train_tags,train_lenghts)
train_loader = DataLoader(train_dataset, shuffle=True, batch_size=batch_size)

dev_data = load_data(dev_file, max_length=max_length, label_dic=l2i_dic, vocab=vocab)
dev_ids = torch.LongTensor([temp.input_id for temp in dev_data[:1500]])
dev_masks = torch.LongTensor([temp.input_mask for temp in dev_data[:1500]])
dev_tags = torch.LongTensor([temp.label_id for temp in dev_data[:1500]])
dev_lenghts = torch.LongTensor([temp.lenght for temp in dev_data[:1500]])
dev_dataset = TensorDataset(dev_ids, dev_masks, dev_tags,dev_lenghts)
dev_loader = DataLoader(dev_dataset, shuffle=True, batch_size=batch_size)


test_data = load_data(test_file, max_length=max_length, label_dic=l2i_dic, vocab=vocab)
test_ids = torch.LongTensor([temp.input_id for temp in test_data])
test_masks = torch.LongTensor([temp.input_mask for temp in test_data])
test_tags = torch.LongTensor([temp.label_id for temp in test_data])
test_lenghts = torch.LongTensor([temp.lenght for temp in test_data])


test_dataset = TensorDataset(test_ids, test_masks, test_tags,test_lenghts)
test_loader = DataLoader(test_dataset, shuffle=False, batch_size=batch_size)


######测试函数
def evaluate(medel, dev_loader):
    medel.eval()
    pred = []
    gold = []
    print('evaluate')
    with torch.no_grad():
        for i, dev_batch in enumerate(dev_loader):
            sentence, masks, tags , lengths = dev_batch
            sentence, masks, tags, lengths = Variable(sentence), Variable(masks), Variable(tags), Variable(lengths)
            if use_cuda:
                sentence = sentence.to(device)
                masks = masks.to(device)
                tags = tags.to(device)

            predict_tags = medel(sentence, masks)
            loss = model.neg_log_likelihood_loss(sentence, masks, tags)

            pred.extend([t for t in predict_tags.tolist()])
            gold.extend([t for t in tags.tolist()])

        pred_label,gold_label = recover_label(pred, gold, l2i_dic,i2l_dic)
        print('dev loss {}'.format(loss.item()))
        pred_label_1 = [t[1:] for t in pred_label]
        gold_label_1 = [t[1:] for t in gold_label]
        acc,p, r, f = get_ner_fmeasure(gold_label_1,pred_label_1)
        print('p: {}，r: {}, f: {}'.format(p, r, f))
        return p, r, f

# test 函数
def evaluate_test(medel,test_loader,dev_f):
    medel.eval()
    pred = []
    gold = []
    print('test')
    with torch.no_grad():
        for i, dev_batch in enumerate(test_loader):
            sentence, masks, tags, lengths = dev_batch
            sentence, masks, tags , lengths = Variable(sentence), Variable(masks), Variable(tags),Variable(lengths)
            if use_cuda:
                sentence = sentence.to(device)
                masks = masks.to(device)
                tags = tags.to(device)
            predict_tags = medel(sentence, masks)

            pred.extend([t for t in predict_tags.tolist()])
            gold.extend([t for t in tags.tolist()])

        pred_label, gold_label = recover_label(pred, gold, l2i_dic,i2l_dic)
        pred_label_2 = [t[1:] for t in pred_label]
        gold_label_2 = [t[1:] for t in gold_label]
        fw = open('data/predict_result'+str(float('%.3f'%dev_f))+'bert.txt','w')
        for i in pred_label_2:
            for j in range(len(i)-1):
                fw.write(i[j])
                fw.write(' ')
            fw.write(i[len(i)-1])
            fw.write('\n')
        acc,p, r, f = get_ner_fmeasure(gold_label_2,pred_label_2)
        print('p: {}，r: {}, f: {}'.format(p, r, f))
    return p, r, f


model = BERT_LSTM_CRF('./data/my_bert', tagset_size, 768, 200, 2,
                      dropout_ratio=0.5, dropout1=0.5, use_cuda = use_cuda)

if use_cuda:
    model.to(device)

optimizer = getattr(optim, 'Adam')
optimizer = optimizer(model.parameters(), lr=0.000005, weight_decay=0.00005)

best_f = -100
model_name = save_model_dir + '0518' + str(float('%.3f' % best_f)) + ".pkl"
print(model_name)

for epoch in range(epochs):
    print('epoch: {}，train'.format(epoch))
    for i, train_batch in enumerate(tqdm(train_loader)):
        sentence, masks, tags , lengths= train_batch

        sentence, masks, tags , lengths = Variable(sentence), Variable(masks), Variable(tags), Variable(lengths)

        if use_cuda:
            sentence = sentence.to(device)
            masks = masks.to(device)
            tags = tags.to(device)
        model.train()
        optimizer.zero_grad()
        loss = model.neg_log_likelihood_loss(sentence, masks, tags)
        loss.backward()
        optimizer.step()

    print('epoch: {}，train loss: {}'.format(epoch, loss.item()))
    p, r, f = evaluate(model,dev_loader)


    if f > best_f:
        best_f = f
        _, _, _ = evaluate_test(model, test_loader, loss.item())
        model_name = save_model_dir + 'new' + str(float('%.3f' % best_f)) + ".pkl"
        torch.save(model.state_dict(), model_name)