def main():
# 加载配置文件
with open('./config.yml') as file_config:
config = yaml.load(file_config)
feature_names = config['model_params']['feature_names']
# 初始化embedding shape, dropouts, 预训练的embedding也在这里初始化)
feature_weight_shape_dict, feature_weight_dropout_dict, \
feature_init_weight_dict = dict(), dict(), dict()
for feature_name in feature_names:
feature_weight_shape_dict[feature_name] = \
config['model_params']['embed_params'][feature_name]['shape']
feature_weight_dropout_dict[feature_name] = \
config['model_params']['embed_params'][feature_name]['dropout_rate']
path_pre_train = config['model_params']['embed_params'][feature_name]['path']
if path_pre_train:
with open(path_pre_train, 'rb') as file_r:
feature_init_weight_dict[feature_name] = pickle.load(file_r)
# 加载数据
# 加载vocs
path_vocs = []
for feature_name in feature_names:
path_vocs.append(config['data_params']['voc_params'][feature_name]['path'])
path_vocs.append(config['data_params']['voc_params']['label']['path'])
vocs = load_vocs(path_vocs)
# 加载训练数据
sep_str = config['data_params']['sep']
assert sep_str in ['table', 'space']
sep = '\t' if sep_str == 'table' else ' '
data_dict = init_data(
path=config['data_params']['path_train'], feature_names=feature_names, sep=sep,
vocs=vocs, max_len=config['model_params']['sequence_length'], model='train')
# 训练模型
model = SequenceLabelingModel(
sequence_length=config['model_params']['sequence_length'],
nb_classes=config['model_params']['nb_classes'],
nb_hidden=config['model_params']['bilstm_params']['num_units'],
feature_weight_shape_dict=feature_weight_shape_dict,
feature_init_weight_dict=feature_init_weight_dict,
feature_weight_dropout_dict=feature_weight_dropout_dict,
dropout_rate=config['model_params']['dropout_rate'],
nb_epoch=config['model_params']['nb_epoch'], feature_names=feature_names,
batch_size=config['model_params']['batch_size'],
train_max_patience=config['model_params']['max_patience'],
use_crf=config['model_params']['use_crf'],
l2_rate=config['model_params']['l2_rate'],
rnn_unit=config['model_params']['rnn_unit'],
learning_rate=config['model_params']['learning_rate'],
clip=config['model_params']['clip'],
path_model=config['model_params']['path_model'])
model.fit(
data_dict=data_dict, dev_size=config['model_params']['dev_size'])
def predict(testlist):
"""Prepare the model input data type
paramters：
testlist: list,[[[u'T'], [u':'],...]
return result_sequences: list,
"""
feature_names, sep, vocs, max_len, use_char_feature, word_len = load_parameters()
data_dict = init_data( feature_names=feature_names,
sep=sep,test_sens=testlist,vocs=vocs, max_len=max_len,
model='test',use_char_feature=use_char_feature,word_len=word_len)
# 生成模型feed data
data_count = data_dict['f1'].shape[0]
nb_test = int(math.ceil(data_count /16.0))
result_sequences = [] # 标记结果
for i in range(nb_test):
feed_dict = dict()
batch_indices = np.arange(i * 16, (i + 1) * 16) \
if (i+1)*16 <= data_count else \
np.arange(i*16, data_count)
batch_data = data_dict['f1'][batch_indices]
item = {'input_x_f1': batch_data}
feed_dict.update(item)
# dropout
item = {'weight_dropout_ph_dict_f1': np.array(0.0, dtype=np.float32)}
feed_dict.update(item)
feed_dict.update({'dropout_rate_ph':np.array(0.,dtype=np.float32), 'rnn_dropout_rate_ph': np.array(0.,dtype=np.float32)})
# viterbi decode procedure
logits, sequence_actual_length, transition_params = load_ner_service(feed_dict)
for logit, seq_len in zip(logits, sequence_actual_length):
logit_actual = logit[:seq_len]
viterbi_sequence, _ = tf.contrib.crf.viterbi_decode(logit_actual, transition_params)
result_sequences.append(viterbi_sequence)
return result_sequences
def pre_feed_data(testlist):
# 加载配置文件
feature_names, sep, vocs, max_len, use_char_feature, word_len = load_parameters(
)
data_dict = init_data(feature_names=feature_names,
sep=sep,
test_sens=testlist,
vocs=vocs,
max_len=max_len,
model='test',
use_char_feature=use_char_feature,
word_len=word_len)
# 生成模型feed data
data_count = data_dict['f1'].shape[0]
nb_test = int(math.ceil(data_count / 16.0))
result_sequences = [] # 标记结果
for i in range(nb_test):
feed_dict = dict()
batch_indices = np.arange(i * 16, (i + 1) * 16) \
if (i+1)*16 <= data_count else \
np.arange(i*16, data_count)
batch_data = data_dict['f1'][batch_indices]
item = {'input_x_f1': batch_data}
feed_dict.update(item)
# dropout
item = {'weight_dropout_ph_dict_f1': np.array(0.0, dtype=np.float32)}
feed_dict.update(item)
feed_dict.update({
'dropout_rate_ph': np.array(0.0, dtype=np.float32),
'rnn_dropout_rate_ph': np.array(0.0, dtype=np.float32)
})
print 'feed_dict_tfserving', feed_dict
yield feed_dict
# print('adj2.size = ', adj2.size())
output = model.forward(input1, input2, adj1, adj2)
_, pre = torch.max(output, dim=1)
tmp = np.zeros((1, 1), dtype=np.int)
tmp[0][0] = label
tmp_label = torch.from_numpy(tmp)
pre = pre.cuda()
tmp_label = tmp_label.cuda()
if pre[0] == tmp_label[0][0]:
acc_num += 1
# print("accuracy : ", acc_num, "of ", test_count)
return acc_num * 1.0 / test_count
if __name__ == '__main__':
my_graphs, max_node_num1, max_node_num2 = init_data(datadir, dataname)
random.shuffle(my_graphs)
print("数据处理完成", time.asctime(time.localtime(time.time())))
model = GraphClassifier(max_node_num1, max_node_num2)
model = model.cuda()
print('model:', model)
crossentropy = torch.nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=learning_rate)
print('开始训练', time.asctime(time.localtime(time.time())))
max_acc = 0
print_loss = 0
for epoch in range(num_epoches):
for i in range(len(my_graphs)):
if i > index:
break
torch.cuda.empty_cache()
def main():
# 加载配置文件
with open('./config.yml') as file_config:
config = yaml.load(file_config)
feature_names = config['model_params']['feature_names']
use_char_feature = config['model_params']['use_char_feature']
# 初始化embedding shape, dropouts, 预训练的embedding也在这里初始化)
feature_weight_shape_dict, feature_weight_dropout_dict, \
feature_init_weight_dict = dict(), dict(), dict()
for feature_name in feature_names:
feature_weight_shape_dict[feature_name] = \
config['model_params']['embed_params'][feature_name]['shape']
feature_weight_dropout_dict[feature_name] = \
config['model_params']['embed_params'][feature_name]['dropout_rate']
path_pre_train = config['model_params']['embed_params'][feature_name]['path']
if path_pre_train:
with open(path_pre_train, 'rb') as file_r:
feature_init_weight_dict[feature_name] = pickle.load(file_r)
# char embedding shape
if use_char_feature:
feature_weight_shape_dict['char'] = \
config['model_params']['embed_params']['char']['shape']
conv_filter_len_list = config['model_params']['conv_filter_len_list']
conv_filter_size_list = config['model_params']['conv_filter_size_list']
else:
conv_filter_len_list = None
conv_filter_size_list = None
# 加载数据
# 加载vocs
path_vocs = []
if use_char_feature:
path_vocs.append(config['data_params']['voc_params']['char']['path'])
for feature_name in feature_names:
path_vocs.append(config['data_params']['voc_params'][feature_name]['path'])
path_vocs.append(config['data_params']['voc_params']['label']['path'])
vocs = load_vocs(path_vocs)
# 加载数据
sep_str = config['data_params']['sep']
assert sep_str in ['table', 'space']
sep = '\t' if sep_str == 'table' else ' '
max_len = config['model_params']['sequence_length']
word_len = config['model_params']['word_length']
data_dict = init_data(
path=config['data_params']['path_test'], feature_names=feature_names, sep=sep,
vocs=vocs, max_len=max_len, model='test', use_char_feature=use_char_feature,
word_len=word_len)
# 加载模型
model = SequenceLabelingModel(
sequence_length=config['model_params']['sequence_length'],
nb_classes=config['model_params']['nb_classes'],
nb_hidden=config['model_params']['bilstm_params']['num_units'],
num_layers=config['model_params']['bilstm_params']['num_layers'],
feature_weight_shape_dict=feature_weight_shape_dict,
feature_init_weight_dict=feature_init_weight_dict,
feature_weight_dropout_dict=feature_weight_dropout_dict,
dropout_rate=config['model_params']['dropout_rate'],
nb_epoch=config['model_params']['nb_epoch'], feature_names=feature_names,
batch_size=config['model_params']['batch_size'],
train_max_patience=config['model_params']['max_patience'],
use_crf=config['model_params']['use_crf'],
l2_rate=config['model_params']['l2_rate'],
rnn_unit=config['model_params']['rnn_unit'],
learning_rate=config['model_params']['learning_rate'],
use_char_feature=use_char_feature,
conv_filter_size_list=conv_filter_size_list,
conv_filter_len_list=conv_filter_len_list,
word_length=word_len,
path_model=config['model_params']['path_model'])
saver = tf.train.Saver()
saver.restore(model.sess, config['model_params']['path_model'])
# 标记
viterbi_sequences = model.predict(data_dict)
# 写入文件
label_voc = dict()
for key in vocs[-1]:
label_voc[vocs[-1][key]] = key
with codecs.open(config['data_params']['path_test'], 'r', encoding='utf-8') as file_r:
sentences = file_r.read().strip().split('\n\n')
file_result = codecs.open(
config['data_params']['path_result'], 'w', encoding='utf-8')
for i, sentence in enumerate(sentences):
for j, item in enumerate(sentence.split('\n')):
if j < len(viterbi_sequences[i]):
file_result.write('%s\t%s\n' % (item, label_voc[viterbi_sequences[i][j]]))
else:
file_result.write('%s\tO\n' % item)
file_result.write('\n')
file_result.close()
def main():
# 加载配置文件
print("config5")
with open('./train_config/config_b2b_tag_5_only_jieba.yml') as file_config:
config = yaml.load(file_config)
feature_names = config['model_params']['feature_names']
use_char_feature = config['model_params']['use_char_feature']
# 初始化embedding shape, dropouts, 预训练的embedding也在这里初始化)
feature_weight_shape_dict, feature_weight_dropout_dict, \
feature_init_weight_dict = dict(), dict(), dict()
for feature_name in feature_names:
feature_weight_shape_dict[feature_name] = \
config['model_params']['embed_params'][feature_name]['shape']
feature_weight_dropout_dict[feature_name] = \
config['model_params']['embed_params'][feature_name]['dropout_rate']
path_pre_train = config['model_params']['embed_params'][feature_name][
'path']
if path_pre_train:
with open(path_pre_train, 'rb') as file_r:
feature_init_weight_dict[feature_name] = pickle.load(file_r)
# char embedding shape
if use_char_feature:
feature_weight_shape_dict['char'] = \
config['model_params']['embed_params']['char']['shape']
conv_filter_len_list = config['model_params']['conv_filter_len_list']
conv_filter_size_list = config['model_params']['conv_filter_size_list']
else:
conv_filter_len_list = None
conv_filter_size_list = None
# 加载数据
# 加载vocs
path_vocs = []
if use_char_feature:
path_vocs.append(config['data_params']['voc_params']['char']['path'])
for feature_name in feature_names:
path_vocs.append(
config['data_params']['voc_params'][feature_name]['path'])
path_vocs.append(config['data_params']['voc_params']['label']['path'])
vocs = load_vocs(path_vocs)
# 加载训练数据
sep_str = config['data_params']['sep']
assert sep_str in ['table', 'space']
sep = '\t' if sep_str == 'table' else ' '
max_len = config['model_params']['sequence_length']
word_len = config['model_params']['word_length']
data_dict = init_data(path=config['data_params']['path_train'],
feature_names=feature_names,
sep=sep,
vocs=vocs,
max_len=max_len,
model='train',
use_char_feature=use_char_feature,
word_len=word_len)
# 训练模型
model = SequenceLabelingModel(
sequence_length=config['model_params']['sequence_length'],
nb_classes=config['model_params']['nb_classes'],
nb_hidden=config['model_params']['bilstm_params']['num_units'],
num_layers=config['model_params']['bilstm_params']['num_layers'],
rnn_dropout=config['model_params']['bilstm_params']['rnn_dropout'],
feature_weight_shape_dict=feature_weight_shape_dict,
feature_init_weight_dict=feature_init_weight_dict,
feature_weight_dropout_dict=feature_weight_dropout_dict,
dropout_rate=config['model_params']['dropout_rate'],
nb_epoch=config['model_params']['nb_epoch'],
feature_names=feature_names,
batch_size=config['model_params']['batch_size'],
train_max_patience=config['model_params']['max_patience'],
use_crf=config['model_params']['use_crf'],
l2_rate=config['model_params']['l2_rate'],
rnn_unit=config['model_params']['rnn_unit'],
learning_rate=config['model_params']['learning_rate'],
clip=config['model_params']['clip'],
use_char_feature=use_char_feature,
conv_filter_size_list=conv_filter_size_list,
conv_filter_len_list=conv_filter_len_list,
cnn_dropout_rate=config['model_params']['conv_dropout'],
word_length=word_len,
path_model=config['model_params']['path_model'])
model.fit(data_dict=data_dict, dev_size=config['model_params']['dev_size'])
def predict(string):
choiceAction = []
choiceTarget = []
choiceData = []
lab = writetxt(string)
# 加载数据
if len(lab[0]) == 0:
return 'ok;None'
sep_str = config['data_params']['sep']
assert sep_str in ['table', 'space']
sep = '\t' if sep_str == 'table' else ' '
data_dict = init_data(path=config['data_params']['path_test'],
feature_names=feature_names,
sep=sep,
vocs=vocs,
max_len=config['model_params']['sequence_length'],
model='test')
saver = tf.train.Saver()
saver.restore(model.sess, config['model_params']['path_model'])
seq = model.predict(data_dict)
print(seq)
for i in range(len(seq)):
if (vocs[-1]['B_ACT'] in seq[i] or vocs[-1]['I_ACT'] in seq[i]
or vocs[-1]['E_ACT'] in seq[i] or vocs[-1]['S_ACT'] in seq[i]):
tem = ""
for j in range(len(seq[i])):
if seq[i][j] == vocs[-1]['B_ACT']:
tem = lab[i][j]
elif seq[i][j] == vocs[-1]['I_ACT']:
tem += lab[i][j]
elif seq[i][j] == vocs[-1]['E_ACT']:
tem += lab[i][j]
choiceAction.append(tem)
elif seq[i][j] == vocs[-1]['S_ACT']:
choiceAction.append(lab[i][j])
# if len(tem) > 0:
# choiceAction.append(tem)
ch = '***'.join(choiceAction)
finalAction = '' + ch
if finalAction == '':
finalAction = '0'
for i in range(len(seq)):
if (vocs[-1]['B_TAR'] in seq[i] or vocs[-1]['I_TAR'] in seq[i]
or vocs[-1]['E_TAR'] in seq[i] or vocs[-1]['S_TAR'] in seq[i]):
tem = ""
for j in range(len(seq[i])):
if seq[i][j] == vocs[-1]['B_TAR']:
tem = lab[i][j]
elif seq[i][j] == vocs[-1]['I_TAR']:
tem += lab[i][j]
elif seq[i][j] == vocs[-1]['E_TAR']:
tem += lab[i][j]
choiceTarget.append(tem)
elif seq[i][j] == vocs[-1]['S_TAR']:
choiceTarget.append(lab[i][j])
# if seq[i][j] == 11:
# choiceTarget.append(lab[i][j])
# if len(tem) > 0:
# choiceTarget.append(tem)
ch = '***'.join(choiceTarget)
finalTarget = '' + ch
if finalTarget == '':
finalTarget = '0'
for i in range(len(seq)):
if (vocs[-1]['B_DAT'] in seq[i] or vocs[-1]['I_DAT'] in seq[i]
or vocs[-1]['E_DAT'] in seq[i] or vocs[-1]['S_DAT'] in seq[i]):
tem = ""
for j in range(len(seq[i])):
if seq[i][j] == vocs[-1]['B_DAT']:
tem = lab[i][j]
elif seq[i][j] == vocs[-1]['I_DAT']:
tem += lab[i][j]
elif seq[i][j] == vocs[-1]['E_DAT']:
tem += lab[i][j]
choiceData.append(tem)
elif seq[i][j] == vocs[-1]['S_DAT']:
choiceData.append(lab[i][j])
# if len(tem) > 0:
# choiceData.append(tem)
ch = '***'.join(choiceData)
finalData = '' + ch
if finalData == '':
finalData = '0'
return finalAction, finalTarget, finalData
def main():
# 加载配置文件
with open('./config.yml') as file_config:
config = yaml.load(file_config)
feature_names = config['model_params']['feature_names']
logger.info(feature_names)
use_char_feature = config['model_params']['use_char_feature']
logger.info(use_char_feature)
# 初始化embedding shape, dropouts, 预训练的embedding也在这里初始化)
feature_weight_shape_dict = dict()
feature_weight_dropout_dict = dict()
feature_init_weight_dict = dict()
for feature_name in feature_names:
feature_weight_shape_dict[feature_name] = config['model_params'][
'embed_params'][feature_name]['shape']
feature_weight_dropout_dict[feature_name] = config['model_params'][
'embed_params'][feature_name]['dropout_rate']
# embeding mat, 比voc多了两行， 因为voc从2开始编序， 0， 1行用0填充
path_pre_train = config['model_params']['embed_params'][feature_name][
'path'] # 词嵌矩阵位置
# logger.info("%s init mat path: %s" % (feature_name, path_pre_train))
with open(path_pre_train, 'rb') as file_r:
feature_init_weight_dict[feature_name] = pickle.load(file_r)
logger.info(feature_weight_dropout_dict)
logger.info(feature_weight_shape_dict)
logger.info(feature_init_weight_dict)
# char embedding shape
if use_char_feature:
# 暂时不考虑
feature_weight_shape_dict['char'] = config['model_params'][
'embed_params']['char']['shape']
conv_filter_len_list = config['model_params']['conv_filter_len_list']
conv_filter_size_list = config['model_params']['conv_filter_size_list']
else:
# 利用卷集层来提取char的信息
conv_filter_len_list = None
conv_filter_size_list = None
# 加载vocs
path_vocs = []
if use_char_feature:
path_vocs.append(config['data_params']['voc_params']['char']
['path']) # vocs用于将文本数字序列化
for feature_name in feature_names:
path_vocs.append(
config['data_params']['voc_params'][feature_name]['path'])
path_vocs.append(config['data_params']['voc_params']['label']['path'])
vocs = load_vocs(path_vocs)
# 加载训练数据
sep_str = config['data_params']['sep']
assert sep_str in ['table', 'space'] # 数据的分隔方式
sep = '\t' if sep_str == 'table' else ' '
max_len = config['model_params']['sequence_length']
word_len = config['model_params']['word_length']
# 通过voc 将input f1 和输出 label 数字序列化 得到训练的输入和输出
# data_dict = None
data_dict = init_data(path=config['data_params']['path_train'],
feature_names=feature_names,
sep=sep,
vocs=vocs,
max_len=max_len,
model='train',
use_char_feature=use_char_feature,
word_len=word_len)
logger.info(data_dict) # 每个特征序列化后的数据
# 训练模型
model = SequenceLabelingModel(
sequence_length=config['model_params']['sequence_length'], # 句子被固定长度
nb_classes=config['model_params']['nb_classes'],
nb_hidden=config['model_params']['bilstm_params']['num_units'],
num_layers=config['model_params']['bilstm_params']['num_layers'],
rnn_dropout=config['model_params']['bilstm_params']['rnn_dropout'],
feature_weight_shape_dict=feature_weight_shape_dict,
feature_init_weight_dict=feature_init_weight_dict,
feature_weight_dropout_dict=feature_weight_dropout_dict,
dropout_rate=config['model_params']['dropout_rate'],
nb_epoch=config['model_params']['nb_epoch'],
feature_names=feature_names,
batch_size=config['model_params']['batch_size'],
train_max_patience=config['model_params']['max_patience'],
use_crf=config['model_params']['use_crf'],
l2_rate=config['model_params']['l2_rate'],
rnn_unit=config['model_params']['rnn_unit'],
learning_rate=config['model_params']['learning_rate'],
clip=config['model_params']['clip'],
use_char_feature=use_char_feature,
conv_filter_size_list=conv_filter_size_list,
conv_filter_len_list=conv_filter_len_list,
cnn_dropout_rate=config['model_params']['conv_dropout'],
word_length=word_len,
path_model=config['model_params']['path_model'],
last_train_sess_path=None, # 为了加快训练的速度我们继续载入前面训练的参数
transfer=False) # 是否对前面载入的参数进行迁移学习，True的话就重置LSTM的输出层
model.fit(data_dict=data_dict, dev_size=config['model_params']['dev_size'])
"""
def predict(testlist):
# 加载配置文件
with open('./config.yml') as file_config:
config = yaml.load(file_config)
feature_names = config['model_params']['feature_names']
use_char_feature = config['model_params']['use_char_feature']
# 初始化embedding shape, dropouts, 预训练的embedding也在这里初始化)
feature_weight_shape_dict, feature_weight_dropout_dict, \
feature_init_weight_dict = dict(), dict(), dict()
for feature_name in feature_names:
feature_weight_shape_dict[feature_name] = \
config['model_params']['embed_params'][feature_name]['shape']
feature_weight_dropout_dict[feature_name] = \
config['model_params']['embed_params'][feature_name]['dropout_rate']
path_pre_train = config['model_params']['embed_params'][feature_name][
'path']
if path_pre_train:
with open(path_pre_train, 'rb') as file_r:
feature_init_weight_dict[feature_name] = pickle.load(file_r)
# char embedding shape
if use_char_feature:
feature_weight_shape_dict['char'] = \
config['model_params']['embed_params']['char']['shape']
conv_filter_len_list = config['model_params']['conv_filter_len_list']
conv_filter_size_list = config['model_params']['conv_filter_size_list']
else:
conv_filter_len_list = None
conv_filter_size_list = None
# 加载vocs
path_vocs = []
if use_char_feature:
path_vocs.append(config['data_params']['voc_params']['char']['path'])
for feature_name in feature_names:
path_vocs.append(
config['data_params']['voc_params'][feature_name]['path'])
path_vocs.append(config['data_params']['voc_params']['label']['path'])
vocs = load_vocs(path_vocs)
# 加载数据
sep_str = config['data_params']['sep']
assert sep_str in ['table', 'space']
sep = '\t' if sep_str == 'table' else ' '
max_len = config['model_params']['sequence_length']
word_len = config['model_params']['word_length']
data_dict = init_data(path=config['data_params']['path_test'],
feature_names=feature_names,
sep=sep,
test_sens=testlist,
vocs=vocs,
max_len=max_len,
model='test',
use_char_feature=use_char_feature,
word_len=word_len)
# 加载模型
model = SequenceLabelingModel(
sequence_length=config['model_params']['sequence_length'],
nb_classes=config['model_params']['nb_classes'],
nb_hidden=config['model_params']['bilstm_params']['num_units'],
num_layers=config['model_params']['bilstm_params']['num_layers'],
feature_weight_shape_dict=feature_weight_shape_dict,
feature_init_weight_dict=feature_init_weight_dict,
feature_weight_dropout_dict=feature_weight_dropout_dict,
dropout_rate=config['model_params']['dropout_rate'],
nb_epoch=config['model_params']['nb_epoch'],
feature_names=feature_names,
batch_size=config['model_params']['batch_size'],
train_max_patience=config['model_params']['max_patience'],
use_crf=config['model_params']['use_crf'],
l2_rate=config['model_params']['l2_rate'],
rnn_unit=config['model_params']['rnn_unit'],
learning_rate=config['model_params']['learning_rate'],
use_char_feature=use_char_feature,
conv_filter_size_list=conv_filter_size_list,
conv_filter_len_list=conv_filter_len_list,
word_length=word_len,
path_model=config['model_params']['path_model'])
saver = tf.train.Saver()
saver.restore(model.sess, config['model_params']['path_model'])
# print('data_dict', data_dict)
# 标记
result_sequences = model.predict(data_dict)
#print('result_sequences', result_sequences)
# 输出结果
label_voc = dict()
for key in vocs[-1]:
label_voc[vocs[-1][key]] = key
outlist = []
for i, sentence in enumerate(testlist):
templist = []
for j, item in enumerate(sentence):
#char = recheck_char(item[0])
char = item[0]
if j < len(result_sequences[i]):
out = [char, label_voc[result_sequences[i][j]]]
else:
out = [char, 'O']
templist.append(out)
outlist.append(templist)
return outlist
# 加载数据
# 加载 vocs
path_vocs = []
for feature_name in feature_names:
path_vocs.append(config['data_params']['voc_params'][feature_name]['path'])
path_vocs.append(config['data_params']['voc_params']['label']['path'])
vocs = load_vocs(path_vocs)
# 加载训练数据
sep_str = config['data_params']['sep']
assert sep_str in ['table', 'space']
sep = '\t' if sep_str == 'table' else ' '
data_dict = init_data(path=config['data_params']['path_train'],
feature_names=feature_names,
sep=sep,
vocs=vocs,
max_len=config['model_params']['sequence_length'],
model='train')
# 训练模型
model = SequenceLabelingModel(
sequence_length=config['model_params']['sequence_length'],
nb_classes=config['model_params']['nb_classes'],
nb_hidden=config['model_params']['bilstm_params']['num_units'],
feature_weight_shape_dict=feature_weight_shape_dict,
feature_init_weight_dict=feature_init_weight_dict,
feature_weight_dropout_dict=feature_weight_dropout_dict,
dropout_rate=config['model_params']['dropout_rate'],
nb_epoch=config['model_params']['nb_epoch'],
feature_names=feature_names,
batch_size=config['model_params']['batch_size'],
