-
Notifications
You must be signed in to change notification settings - Fork 54
/
train.py
92 lines (72 loc) · 3.94 KB
/
train.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
import tensorflow as tf
from data_util import getBatches,get_data,get_sentence_int_to_vocab,get_posint
from transformer_CRF import LSTM_CRFModel
from tqdm import tqdm
import math
import os
import random
tf.app.flags.DEFINE_integer("rnn_size",100,"Number of hidden units in each layer")
tf.app.flags.DEFINE_integer("num_blocks",1,"Number of transformer block")
tf.app.flags.DEFINE_integer("num_heads",8,"Number of attention head")
tf.app.flags.DEFINE_integer("batch_size",30,"Batch Size")
tf.app.flags.DEFINE_integer("embedding_size",100,"Embedding dimensions of encoder and decoder inputs")
tf.app.flags.DEFINE_float("learning_rate",0.01,"Learning rate")
tf.app.flags.DEFINE_float("keep_prob",0.5,"keep_prob")
tf.app.flags.DEFINE_integer("numEpochs",30,"Maximum # of training epochs")
tf.app.flags.DEFINE_string("model_dir","saves/","Path to save model checkpoints")
tf.app.flags.DEFINE_string("model_name","ner.ckpt","File name used for model checkpoints")
FLAGS = tf.app.flags.FLAGS
train_datapath = "dataset/train.txt"
test_datapath = "dataset/test.txt"
sentence_int_to_vocab, sentence_vocab_to_int, tags_vocab_to_int, tags_int_to_vocab = get_sentence_int_to_vocab(train_datapath,test_datapath)
pos_vocab_to_int,pos_int_to_vocab = get_posint()
data = get_data(train_datapath)
train_data = data
vali_data = get_data(test_datapath)
with tf.Session() as sess:
model = LSTM_CRFModel(FLAGS.num_heads,FLAGS.num_blocks,FLAGS.rnn_size,FLAGS.embedding_size,FLAGS.learning_rate,
sentence_vocab_to_int,tags_vocab_to_int,tags_int_to_vocab,pos_vocab_to_int,FLAGS.keep_prob)
ckpt = tf.train.get_checkpoint_state(FLAGS.model_dir)
if ckpt and tf.train.checkpoint_exists(ckpt.model_checkpoint_path):
print("reloading model parameters....")
model.saver.restore(sess, ckpt.model_checkpoint_path)
else:
print("create new model parameters...")
sess.run(tf.global_variables_initializer())
sess.run(tf.local_variables_initializer())
current_step = 0
avef1list = []
micf1list = []
train_summary_writer = tf.summary.FileWriter("saves/train", graph=sess.graph)
for e in range(FLAGS.numEpochs):
print("Epoch{}/{}-------------".format(e+1,FLAGS.numEpochs))
train_batches = getBatches(train_data,FLAGS.batch_size)
vali_batches = getBatches(vali_data,FLAGS.batch_size)
for train_Batch in tqdm(train_batches,desc = "Training"):
trainloss,acc,trainsummary = model.train(sess,train_Batch)
current_step += 1
tqdm.write("----Step %d -- trainloss %.2f --acc %.2f " %(current_step,trainloss,acc))
train_summary_writer.add_summary(trainsummary, current_step)
if(current_step % 100 == 0):
ave_f1 = 0
mic_f1 = 0
vali_current_step = 0
checkpoint_path = os.path.join(FLAGS.model_dir, FLAGS.model_name)
model.saver.save(sess, checkpoint_path, global_step=current_step)
for vali_Batch in tqdm(vali_batches, desc="Vali"):
vali_current_step+=1
vali_acc,valiloss,valisummary,vali_f1,mirco_f1 = model.vali(sess, vali_Batch)
tqdm.write("----Step %d -- valiloss %.2f -- valiacc %.3f --valif1 %.3f --micf1 %.3f " %(vali_current_step,valiloss,vali_acc*100,vali_f1*100,mirco_f1*100))
ave_f1+=vali_f1
mic_f1+=mirco_f1
ave_f1 = ave_f1/len(vali_batches)
mic_f1 = mic_f1/len(vali_batches)
# if avef1list != [] and ave_f1 * 100 >= max(avef1list):
print("ave_f1",ave_f1*100)
print("mic_f1",mic_f1*100)
avef1list.append(ave_f1*100)
micf1list.append(mic_f1*100)
print("最大avef1为:",max(avef1list))
print("最大avef1索引为:",avef1list.index(max(avef1list)))
print("最大micf1为:",max(micf1list))
print("最大micf1索引为:",micf1list.index(max(micf1list)))