forked from yuanxiaosc/Keras_Attention_Seq2Seq
-
Notifications
You must be signed in to change notification settings - Fork 0
/
run_seq2seq_model.py
209 lines (181 loc) · 10.9 KB
/
run_seq2seq_model.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
203
204
205
206
207
208
209
import keras
import numpy as np
import pickle
import json
import os
import time
from attention_seq2seq_model import train_model_from_scratch,translation_mode,\
train_model_include_Pre_trained_Word_Vector
# 加载预训练好的词向量
def get_word_to_vec_map(glove_file_path=None):
if glove_file_path is None:
return None
with open(glove_file_path, 'r') as f:
words = set()
word_to_vec_map = {}
for line in f:
line = line.strip().split()
curr_word = line[0]
words.add(curr_word)
word_to_vec_map[curr_word] = np.array(line[1:], dtype=np.float64)
return word_to_vec_map
def load_dict():
# 加载字典
with open(os.path.join("preparing_resources", "en_vocab_to_int.pickle"), 'rb') as f:
source_vocab_to_int = pickle.load(f)
with open(os.path.join("preparing_resources", "fa_vocab_to_int.pickle"), 'rb') as f:
target_vocab_to_int = pickle.load(f)
source_vocab_to_word = {idx: word for word, idx in source_vocab_to_int.items()}
target_vocab_to_word = {idx: word for word, idx in target_vocab_to_int.items()}
print("The size of source dict is : {}".format(len(source_vocab_to_int)))
print("The size of target dict is : {}".format(len(target_vocab_to_int)))
return source_vocab_to_int, target_vocab_to_int, source_vocab_to_word, target_vocab_to_word
#加载数据和字典
def load_data_and_dict():
'''
:return:
DATA shape:
source_text_shape: (137861, 20) ndaary
target_text_shape: (137861, 25) ndaary
The size of English Map is : 229 dict
The size of French Map is : 358 dict
'''
# 加载数据
prepared_data = np.load(os.path.join("preparing_resources", "prepared_data.npz"))
source_text = prepared_data['X']
target_text = prepared_data['Y']
print("\nDATA shape:")
print("source_text_shape:\t", source_text.shape)
print("target_text_shape:\t", target_text.shape)
# 加载字典
with open(os.path.join("preparing_resources", "en_vocab_to_int.pickle"), 'rb') as f:
source_vocab_to_int = pickle.load(f)
with open(os.path.join("preparing_resources", "fa_vocab_to_int.pickle"), 'rb') as f:
target_vocab_to_int = pickle.load(f)
print("The size of English Map is : {}".format(len(source_vocab_to_int)))
print("The size of French Map is : {}".format(len(target_vocab_to_int)))
return source_text, target_text, source_vocab_to_int, target_vocab_to_int
def load_data_trained_Word_Vector_and_train_model():
# 加载预训练词向量（非必须）
# Glove官方页面：https://nlp.stanford.edu/projects/glove/
# Glove数据下载地址：https://link.zhihu.com/?target=http%3A//nlp.stanford.edu/data/glove.6B.zip
glove_100d_file_path = "/home/b418/jupyter_workspace/B418_common/袁宵/data/Glove/glove.6B.100d.txt"
word_to_vec_map = get_word_to_vec_map(glove_100d_file_path)
#加载数据和对应的字典
source_text, target_text, source_vocab_to_int, target_vocab_to_int = load_data_and_dict()
source_sequence_lenth = source_text.shape[1] # 20 源序列长度
target_sequence_lenth = target_text.shape[1] # 25 目标序列长度
source_vocab_length = len(source_vocab_to_int)
target_vocab_length = len(target_vocab_to_int) # The size of target vocab is : 358
encoder_Bi_LSTM_units_numbers = 32 # 编码端 Bi-LSTM 的隐藏状态单元个数
decoder_LSTM_units_numbers = 128 # 解码端 LSTM 的隐藏状态单元个数
model, model_name = train_model_include_Pre_trained_Word_Vector(source_vocab_to_int=source_vocab_to_int,
word_to_vec_map=word_to_vec_map,
target_vocab_length=target_vocab_length,
source_sequence_lenth=source_sequence_lenth,
target_sequence_lenth=target_sequence_lenth,
encoder_Bi_LSTM_units_numbers=encoder_Bi_LSTM_units_numbers,
decoder_LSTM_units_numbers=decoder_LSTM_units_numbers)
# 把目标序列转换成onehot形式
target_text_to_onehot = np.array(
list(map(lambda x: keras.utils.to_categorical(x, num_classes=len(target_vocab_to_int)), target_text)))
outputs = list(target_text_to_onehot.swapaxes(0, 1))
# 训练模型
model.fit(source_text,outputs, epochs=2, batch_size=1024)
now_time = time.time()
model_weight_file_name = "WEIGHT_{}.h5".format(model_name)
# 保存模型参数信息
model.save_weights(model_weight_file_name)
model_info = {"model_name":model_name,"train_time":now_time,
"model_weight_file_name":model_weight_file_name,
"trained_Word_Vector":glove_100d_file_path,
"source_sequence_lenth":source_sequence_lenth,"target_sequence_lenth":target_sequence_lenth,
"source_vocab_length":source_vocab_length,"target_vocab_length":target_vocab_length,
"encoder_Bi_LSTM_units_numbers":encoder_Bi_LSTM_units_numbers,
"decoder_LSTM_units_numbers":decoder_LSTM_units_numbers}
with open("model_info.json", "w") as f:
json.dump(model_info, f)
def load_data_and_translate_from_scrach():
#加载数据和对应的字典
source_text, target_text, source_vocab_to_int, target_vocab_to_int = load_data_and_dict()
source_sequence_lenth = source_text.shape[1] # 20 源序列长度
target_sequence_lenth = target_text.shape[1] # 25 目标序列长度
source_vocab_length = len(source_vocab_to_int) # The size of source vocab is : 229
target_vocab_length = len(target_vocab_to_int) # The size of target vocab is : 358
encoder_Bi_LSTM_units_numbers = 32 # 编码端 Bi-LSTM 的隐藏状态单元个数
decoder_LSTM_units_numbers = 128 # 解码端 LSTM 的隐藏状态单元个数
# 把目标序列转换成onehot形式
target_text_to_onehot = np.array(
list(map(lambda x: keras.utils.to_categorical(x, num_classes=len(target_vocab_to_int)), target_text)))
outputs = list(target_text_to_onehot.swapaxes(0, 1))
# 获取模型结构和模型名称
model, model_name = train_model_from_scratch(source_vocab_length=source_vocab_length,
target_vocab_length=target_vocab_length,
source_sequence_lenth=source_sequence_lenth,
target_sequence_lenth=target_sequence_lenth,
encoder_Bi_LSTM_units_numbers=encoder_Bi_LSTM_units_numbers,
decoder_LSTM_units_numbers=decoder_LSTM_units_numbers)
# 训练模型
model.fit(source_text,outputs, epochs=2, batch_size=1024)
now_time = time.time()
model_weight_file_name = "WEIGHT_{}.h5".format(model_name)
# 保存模型参数信息
model.save_weights(model_weight_file_name)
model_info = {"model_name":model_name,"train_time":now_time,
"model_weight_file_name":model_weight_file_name,
"source_sequence_lenth":source_sequence_lenth,"target_sequence_lenth":target_sequence_lenth,
"source_vocab_length":source_vocab_length,"target_vocab_length":target_vocab_length,
"encoder_Bi_LSTM_units_numbers":encoder_Bi_LSTM_units_numbers,
"decoder_LSTM_units_numbers":decoder_LSTM_units_numbers}
with open("model_info.json", "w") as f:
json.dump(model_info, f)
def load_dict_and_translate():
with open("model_info.json") as f:
train_info = json.load(f)
print("正在加载训练模型参数，模型参数如下：")
print(train_info)
encoder_Bi_LSTM_units_numbers = train_info['encoder_Bi_LSTM_units_numbers']
decoder_LSTM_units_numbers = train_info['decoder_LSTM_units_numbers']
model_weights_file = train_info['model_weight_file_name']
# 加载字典
source_vocab_to_int, _, _, target_vocab_to_word = load_dict()
translate_tool = translation_mode(model_weights_file=model_weights_file,
source_vocab_to_int=source_vocab_to_int,
target_vocab_to_word=target_vocab_to_word,
source_sequence_lenth=20,
target_sequence_lenth=25,
source_vocab_length=len(source_vocab_to_int),
target_vocab_length=len(target_vocab_to_word),
encoder_Bi_LSTM_units_numbers=encoder_Bi_LSTM_units_numbers,
decoder_LSTM_units_numbers=decoder_LSTM_units_numbers)
while True:
source_sentence = input("Please input your sentences: ")
if "exit()" in source_sentence:
break
target_sentence = translate_tool.traslate_a_sentence(source_sentence, delet_PAD=False)
print(target_sentence)
if __name__=='__main__':
is_translation_mode = True
if is_translation_mode:
load_dict_and_translate()
is_load_data_and_translate_from_scrach = False
if is_load_data_and_translate_from_scrach:
load_data_and_translate_from_scrach()
is_load_data_trained_Word_Vector_and_train_model = False
if is_load_data_trained_Word_Vector_and_train_model:
load_data_trained_Word_Vector_and_train_model()
'''
Using TensorFlow backend.
正在加载训练模型参数，模型参数如下：
{'model_name': 'attention_seq2seq_model_include_Pre_trained_Word_Vector', 'train_time': 1544089042.470125, 'model_weight_file_name': 'WEIGHT_attention_seq2seq_model_include_Pre_trained_Word_Vector.h5', 'trained_Word_Vector': '/home/b418/jupyter_workspace/B418_common/袁宵/data/Glove/glove.6B.100d.txt', 'source_sequence_lenth': 20, 'target_sequence_lenth': 25, 'source_vocab_length': 229, 'target_vocab_length': 358, 'encoder_Bi_LSTM_units_numbers': 32, 'decoder_LSTM_units_numbers': 128}
The size of source dict is : 229
The size of target dict is : 358
2018-12-06 19:53:06.278300: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
Please input your sentences: california is usually quiet during march , and it is usually hot in june .
california is usually quiet during march , and it is usually hot in june .
chine est généralement agréable en mois , et il est généralement en en . . <EOS> <PAD> <PAD> <PAD> <PAD> <PAD> <PAD> <PAD> <PAD> <PAD>
Please input your sentences: china is usually dry during march , but it is nice in november .
china is usually dry during march , but it is nice in november .
chine est parfois agréable en mois , et il est généralement en en . . <EOS> <PAD> <PAD> <PAD> <PAD> <PAD> <PAD> <PAD> <PAD> <PAD>
Please input your sentences: exit()
'''