-
Notifications
You must be signed in to change notification settings - Fork 5
/
gen_model.py
206 lines (180 loc) · 9.88 KB
/
gen_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
from gen_modules import *
import tensorflow as tf
import numpy as np
from tqdm import tqdm
class Gen():
def __init__(self, usernum, itemnum, 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.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)
with tf.variable_scope("SA_gen", reuse=reuse):
# sequence embedding, item embedding table
self.seq, self.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_gen",
with_t=True,
reuse=reuse
)
# Positional Encoding
t, pos_emb_table = 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="dec_pos_gen",
reuse=reuse,
with_t=True
)
self.seq += t
# Dropout
self.seq = tf.layers.dropout(self.seq,
rate=args.gen_dropout_rate,
training=tf.convert_to_tensor(self.is_training))
self.seq *= mask
# Build blocks
for i in range(args.gen_num_blocks):
with tf.variable_scope("num_blocks_gen_%d" % i):
# Feed-forward1
self.seq = feedforward(normalize(self.seq), num_units=[args.hidden_units, args.hidden_units],
dropout_rate=args.gen_dropout_rate, is_training=self.is_training)
# Self-attention
self.seq = multihead_attention(queries=normalize(self.seq),
keys=self.seq,
num_units=args.hidden_units,
num_heads=args.gen_num_heads,
dropout_rate=args.gen_dropout_rate,
is_training=self.is_training,
causality=True,
scope="self_attention_gen")
# Feed forward2
self.seq = feedforward2(normalize(self.seq), num_units=[args.hidden_units, args.hidden_units],
dropout_rate=args.gen_dropout_rate, is_training=self.is_training)
self.seq *= mask
self.seq = normalize(self.seq)
self.rewards = tf.placeholder(tf.float32, shape=(args.gen_batch_size * args.maxlen))
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(self.item_emb_table, pos)
neg_emb = tf.nn.embedding_lookup(self.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(self.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, :]
# self.position = tf.placeholder(tf.int32, shape=None)
# self.seq_emb_i = self.seq[:, self.position, :]
# self.seq_emb_i = tf.reshape(self.seq_emb_i, [tf.shape(self.input_seq)[0], args.hidden_units])
# self.item_logits = tf.matmul(seq_emb, tf.transpose(self.item_emb_table))
# self.item_logits = tf.reshape(self.item_logits, [tf.shape(self.input_seq)[0], args.maxlen, tf.shape(self.item_emb_table)[0]])
# self.item_logits = self.item_logits[:, -1, :]
self.seq_emb_i = self.seq[:, -1, :]
self.last_item_logits = tf.matmul(self.seq_emb_i, tf.transpose(self.item_emb_table))
# 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.pre_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.pre_loss += sum(reg_losses)
self.pre_global_step = tf.Variable(0, name='global_step_gen', trainable=False)
self.pre_optimizer = tf.train.AdamOptimizer(learning_rate=args.gen_lr, beta2=0.98)
self.pre_train_op = self.pre_optimizer.minimize(self.pre_loss, global_step=self.pre_global_step)
self.gen_loss = tf.reduce_sum(
(- tf.log(tf.sigmoid(self.pos_logits) + 1e-24) * istarget * self.rewards -
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.gen_loss += sum(reg_losses)
self.gen_global_step = tf.Variable(0, name='global_step_gen', trainable=False)
self.gen_optimizer = tf.train.AdamOptimizer(learning_rate=args.gen_lr, beta2=0.98)
self.gen_train_op = self.gen_optimizer.minimize(self.gen_loss, global_step=self.gen_global_step)
self.merged = tf.summary.merge_all()
def predict(self, sess, u, seq, item_idx):
return sess.run(self.test_logits,
{self.u: u, self.input_seq: seq, self.test_item: item_idx, self.is_training: False})
def generate_position_k(self, sess, u, seq, k, args, batch=6040):
print("sampling")
sampled_item = np.zeros([len(u), args.maxlen, k], dtype=np.int32)
for i in tqdm(range(batch), total=batch, ncols=70, leave=False, unit='u'):
logit = sess.run(self.item_logits,
{self.u: u, self.input_seq: [seq[i]], self.is_training: False})
logit = -logit
index = logit.argsort()
for position in range(args.maxlen):
if seq[i][args.maxlen - 1 - position] == 0:
break
cnt = 0
for j in range(k):
if index[args.maxlen - 1 - position][cnt] == seq[i][args.maxlen - 1 - position]:
cnt += 1
sampled_item[i][args.maxlen - 1 - position][j] = index[args.maxlen - 1 - position][cnt]
cnt += 1
return sampled_item # user * maxlen * k
def generate_k(self, sess, u, seq, pos, k):
batch = 10
interval = int(len(seq) / batch)
begin = 0
sampled_item = np.zeros([len(u), k], dtype=np.int32)
global_pos = 0
for i in tqdm(range(batch-1), total=batch - 1, ncols=70, leave=False, unit='u'):
logit = sess.run(self.last_item_logits,
{self.u: u, self.input_seq: seq[begin:begin + interval], self.is_training: False})
begin += interval
logit = -logit
index = logit.argsort()
for line in range(len(logit)):
cnt = 0
for rank in range(k):
if index[line][cnt] == pos[global_pos]: cnt += 1
sampled_item[global_pos][rank] = index[line][cnt]
cnt += 1
global_pos += 1
logit = sess.run(self.last_item_logits,
{self.u: u, self.input_seq: seq[begin:len(seq)], self.is_training: False})
for line in range(len(logit)):
cnt = 0
for rank in range(k):
if index[line][cnt] == pos[global_pos]: cnt += 1
sampled_item[global_pos][rank] = index[line][cnt]
cnt += 1
global_pos += 1
return sampled_item
def generate_last_item(self, sess, u, seq):
batch = 1
interval = int(len(seq) / batch)
begin = 0
global_pos = 0
top_item = np.zeros([len(u)])
for i in range(batch-1):
logit = sess.run(self.last_item_logits,
{self.u: u, self.input_seq: seq[begin:begin + interval], self.is_training: False})
begin += interval
logit = -logit
index = logit.argsort()
for line in range(len(logit)):
top_item[global_pos] = index[line][0]
global_pos += 1
logit = sess.run(self.last_item_logits,
{self.u: u, self.input_seq: seq[begin:len(seq)], self.is_training: False})
logit = -logit
index = logit.argsort()
for line in range(len(logit)):
top_item[global_pos] = index[line][0]
global_pos += 1
return top_item