-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathscore.py
107 lines (77 loc) · 4.06 KB
/
score.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
# -*- coding: utf-8 -*-
"""
Created on Wed Jan 3 22:38:45 2018
@author: tianle
"""
import pickle
import time
import numpy as np
import tensorflow as tf
# import tensorflow.contrib.layers.fully_connected as linear
class score:
def __init__(self,sequence_length,batch_size,vocab_size,embedding_size,hidden_size,l2_reg_lambda=0.0,paras=None,learning_rate=1e-2,embeddings=None,loss_type="pair",trainable=True):
self.learning_rate=learning_rate
self.paras=paras
self.l2_reg_lambda=l2_reg_lambda
self.embeddings=embeddings
self.sequence_length=sequence_length
self.embedding_size=embedding_size
self.batch_size=batch_size
self.model_type="base"
self.hidden_size=hidden_size
self.is_train=trainable
# self.lstm_keep_prob=lstm_keep_prob
self.input_x_1 = tf.placeholder(tf.int32, [None, sequence_length], name="input_x_1")
self.input_x_2 = tf.placeholder(tf.int32, [None,sequence_length], name="input_x_2")
self.label=tf.placeholder(tf.float32, [None], name="label")
self.lengths=tf.placeholder(tf.int32,[None],name='lengths')
self.updated_paras=[]
# Embedding layer
with tf.name_scope("embedding"):
if self.paras==None:
if self.embeddings ==None:
print ("random embedding")
#w是一个全部词行，和嵌入行的矩阵,
self.Embedding_W = tf.Variable(
tf.random_uniform([vocab_size, embedding_size], -1.0, 1.0),
name="random_W")
else:
self.Embedding_W = tf.Variable(np.array(self.embeddings),name="embedding_W" ,dtype="float32",trainable=False)
with tf.name_scope("prepare"):
q =tf.nn.embedding_lookup(self.Embedding_W, self.input_x_1)
pos=tf.nn.embedding_lookup(self.Embedding_W, self.input_x_2)
self.x12=self.data_concat(q,pos)
fw_cell = tf.nn.rnn_cell.LSTMCell(hidden_size,forget_bias=1.0)
bw_cell = tf.nn.rnn_cell.LSTMCell(hidden_size,forget_bias=1.0)
with tf.variable_scope("dis"):
outputs, states=tf.nn.bidirectional_dynamic_rnn(fw_cell,bw_cell,self.x12,sequence_length=self.lengths,dtype=tf.float32)
states_fw, states_bw = states
output_fw, output_bw = outputs
self.lstm_output = tf.reduce_max(tf.concat([output_fw,output_bw],2),1)
c_f, h_f = states_fw
c_b, h_b= states_bw
#可能还需要改动
self.lstm_out=tf.concat([h_f,h_b],1)
with tf.variable_scope('MLP'):
self.W1 = tf.get_variable('w1',[2*hidden_size,1], initializer=tf.truncated_normal_initializer(mean=0.0, stddev=0.1))
self.b1 = tf.get_variable('b1',[1],initializer=tf.constant_initializer(0.0))
#self.W2 = tf.get_variable('w2',[self.hidden_size,1],initializer=tf.truncated_normal_initializer(mean=0.0, stddev=0.1))
#self.score=tf.matmul(tf.nn.tanh(tf.nn.xw_plus_b(outputs12, self.W1, self.b1)), self.W2)
self.score=tf.nn.xw_plus_b(self.lstm_output, self.W1, self.b1)
self.score=tf.cast(tf.squeeze(self.score),tf.float32)
self.updated_paras.append(self.W1)
# self.l2_loss = tf.constant(0.0)
# #给矩阵算一个平均值
# for para in self.updated_paras:
# self.l2_loss+= tf.nn.l2_loss(para)
def data_concat(self,q,a):
mulq=tf.multiply(q,a)
subq=tf.abs(tf.subtract(q,a))
x=tf.concat([q,mulq,subq],2)
return x
# def attention(self,x):
# z = tf.reduce_sum(tf.contrib.layers.fully_connected(inputs=x,num_outputs=1,activation_fn=None),2)
# mulq=tf.multiply(q,a)
# subq=tf.abs(tf.subtract(q,a))
# attention = tf.nn.softmax(x)
# return attention