-
Notifications
You must be signed in to change notification settings - Fork 26
/
Copy pathmodel.py
162 lines (145 loc) · 7.97 KB
/
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
from modules import *
class Model():
def __init__(self, usernum, itemnum, timenum, args, reuse=None):
self.is_training = tf.placeholder(tf.bool, shape=())
self.u = tf.placeholder(tf.int32, shape=(None))
self.input_seq = tf.placeholder(tf.int32, shape=(None, args.maxlen))
self.time_matrix = tf.placeholder(tf.int32, shape=(None, args.maxlen, args.maxlen))
self.pos = tf.placeholder(tf.int32, shape=(None, args.maxlen))
self.neg = tf.placeholder(tf.int32, shape=(None, args.maxlen))
pos = self.pos
neg = self.neg
mask = tf.expand_dims(tf.to_float(tf.not_equal(self.input_seq, 0)), -1)
self.time_matrix = tf.reshape(self.time_matrix, [tf.shape(self.input_seq)[0], args.maxlen, args.maxlen])
with tf.variable_scope("SASRec", reuse=reuse):
# sequence embedding, item embedding table
self.seq, item_emb_table = embedding(self.input_seq,
vocab_size=itemnum + 1,
num_units=args.hidden_units,
zero_pad=True,
scale=True,
l2_reg=args.l2_emb,
scope="input_embeddings",
with_t=True,
reuse=reuse
)
absolute_pos_K = embedding(
tf.tile(tf.expand_dims(tf.range(tf.shape(self.input_seq)[1]), 0), [tf.shape(self.input_seq)[0], 1]),
vocab_size=args.maxlen,
num_units=args.hidden_units,
zero_pad=False,
scale=False,
l2_reg=args.l2_emb,
scope="abs_pos_K",
reuse=reuse,
with_t=False
)
absolute_pos_V = embedding(
tf.tile(tf.expand_dims(tf.range(tf.shape(self.input_seq)[1]), 0), [tf.shape(self.input_seq)[0], 1]),
vocab_size=args.maxlen,
num_units=args.hidden_units,
zero_pad=False,
scale=False,
l2_reg=args.l2_emb,
scope="abs_pos_V",
reuse=reuse,
with_t=False
)
# Time Encoding
time_matrix_emb_K = embedding(
self.time_matrix,
vocab_size=args.time_span+1,
num_units=args.hidden_units,
zero_pad=False,
scale=False,
l2_reg=args.l2_emb,
scope="dec_time_K",
reuse=reuse,
with_t=False
)
time_matrix_emb_V = embedding(
self.time_matrix,
vocab_size=args.time_span+1,
num_units=args.hidden_units,
zero_pad=False,
scale=False,
l2_reg=args.l2_emb,
scope="dec_time_V",
reuse=reuse,
with_t=False
)
# Dropout
self.seq = tf.layers.dropout(self.seq,
rate=args.dropout_rate,
training=tf.convert_to_tensor(self.is_training))
self.seq *= mask
time_matrix_emb_K = tf.layers.dropout(time_matrix_emb_K,
rate=args.dropout_rate,
training=tf.convert_to_tensor(self.is_training))
time_matrix_emb_V = tf.layers.dropout(time_matrix_emb_V,
rate=args.dropout_rate,
training=tf.convert_to_tensor(self.is_training))
absolute_pos_K = tf.layers.dropout(absolute_pos_K,
rate=args.dropout_rate,
training=tf.convert_to_tensor(self.is_training))
absolute_pos_V = tf.layers.dropout(absolute_pos_V,
rate=args.dropout_rate,
training=tf.convert_to_tensor(self.is_training))
# Build blocks
for i in range(args.num_blocks):
with tf.variable_scope("num_blocks_%d" % i):
# Self-attention
self.seq = multihead_attention(queries=normalize(self.seq),
keys=self.seq,
time_matrix_K=time_matrix_emb_K,
time_matrix_V=time_matrix_emb_V,
absolute_pos_K=absolute_pos_K,
absolute_pos_V=absolute_pos_V,
num_units=args.hidden_units,
num_heads=args.num_heads,
dropout_rate=args.dropout_rate,
is_training=self.is_training,
causality=True,
scope="self_attention",
)
# Feed forward
self.seq = feedforward(normalize(self.seq), num_units=[args.hidden_units, args.hidden_units],
dropout_rate=args.dropout_rate, is_training=self.is_training)
self.seq *= mask
self.seq = normalize(self.seq)
pos = tf.reshape(pos, [tf.shape(self.input_seq)[0] * args.maxlen])
neg = tf.reshape(neg, [tf.shape(self.input_seq)[0] * args.maxlen])
pos_emb = tf.nn.embedding_lookup(item_emb_table, pos)
neg_emb = tf.nn.embedding_lookup(item_emb_table, neg)
seq_emb = tf.reshape(self.seq, [tf.shape(self.input_seq)[0] * args.maxlen, args.hidden_units])
self.test_item = tf.placeholder(tf.int32, shape=(101))
test_item_emb = tf.nn.embedding_lookup(item_emb_table, self.test_item)
self.test_logits = tf.matmul(seq_emb, tf.transpose(test_item_emb))
self.test_logits = tf.reshape(self.test_logits, [tf.shape(self.input_seq)[0], args.maxlen, 101])
self.test_logits = self.test_logits[:, -1, :]
# prediction layer
self.pos_logits = tf.reduce_sum(pos_emb * seq_emb, -1)
self.neg_logits = tf.reduce_sum(neg_emb * seq_emb, -1)
# ignore padding items (0)
istarget = tf.reshape(tf.to_float(tf.not_equal(pos, 0)), [tf.shape(self.input_seq)[0] * args.maxlen])
self.loss = tf.reduce_sum(
- tf.log(tf.sigmoid(self.pos_logits) + 1e-24) * istarget -
tf.log(1 - tf.sigmoid(self.neg_logits) + 1e-24) * istarget
) / tf.reduce_sum(istarget)
reg_losses = tf.get_collection(tf.GraphKeys.REGULARIZATION_LOSSES)
self.loss += sum(reg_losses)
tf.summary.scalar('loss', self.loss)
self.auc = tf.reduce_sum(
((tf.sign(self.pos_logits - self.neg_logits) + 1) / 2) * istarget
) / tf.reduce_sum(istarget)
if reuse is None:
tf.summary.scalar('auc', self.auc)
self.global_step = tf.Variable(0, name='global_step', trainable=False)
self.optimizer = tf.train.AdamOptimizer(learning_rate=args.lr, beta2=0.98)
self.train_op = self.optimizer.minimize(self.loss, global_step=self.global_step)
else:
tf.summary.scalar('test_auc', self.auc)
self.merged = tf.summary.merge_all()
def predict(self, sess, u, seq, time_matrix, item_idx):
return sess.run(self.test_logits,
{self.u: u, self.input_seq: seq, self.time_matrix: time_matrix, self.test_item: item_idx, self.is_training: False})