forked from PaddlePaddle/PaddleRec
-
Notifications
You must be signed in to change notification settings - Fork 0
/
static_model.py
174 lines (149 loc) · 6.71 KB
/
static_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
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import numpy as np
import paddle
from net import ESCMLayer
class StaticModel():
def __init__(self, config):
self.cost = None
self.config = config
self._init_hyper_parameters()
def _init_hyper_parameters(self):
self.max_len = self.config.get("hyper_parameters.max_len", 3)
self.global_w = self.config.get("hyper_parameters.global_w", 0.5)
self.counterfactual_w = self.config.get(
"hyper_parameters.counterfactual_w", 0.5)
self.sparse_feature_number = self.config.get(
"hyper_parameters.sparse_feature_number")
self.sparse_feature_dim = self.config.get(
"hyper_parameters.sparse_feature_dim")
self.num_field = self.config.get("hyper_parameters.num_field")
self.learning_rate = self.config.get(
"hyper_parameters.optimizer.learning_rate")
self.ctr_fc_sizes = self.config.get("hyper_parameters.ctr_fc_sizes")
self.cvr_fc_sizes = self.config.get("hyper_parameters.cvr_fc_sizes")
self.expert_num = self.config.get("hyper_parameters.expert_num")
self.counterfact_mode = self.config.get("runner.counterfact_mode")
self.expert_size = self.config.get("hyper_parameters.expert_size")
self.tower_size = self.config.get("hyper_parameters.tower_size")
self.feature_size = self.config.get("hyper_parameters.feature_size")
def create_feeds(self, is_infer=False):
sparse_input_ids = [
paddle.static.data(
name="field_" + str(i),
shape=[None, self.max_len],
dtype="int64") for i in range(0, 23)
]
label_ctr = paddle.static.data(
name="ctr", shape=[None, 1], dtype="int64")
label_cvr = paddle.static.data(
name="cvr", shape=[None, 1], dtype="int64")
inputs = sparse_input_ids + [label_ctr] + [label_cvr]
if is_infer:
return inputs
else:
return inputs
def counterfact_ipw(self, loss_cvr, ctr_num, O, ctr_out_one):
PS = paddle.multiply(
ctr_out_one, paddle.cast(
ctr_num, dtype="float32"))
min_v = paddle.full_like(PS, 0.000001)
PS = paddle.maximum(PS, min_v)
IPS = paddle.reciprocal(PS)
batch_shape = paddle.full_like(O, 1)
batch_size = paddle.sum(paddle.cast(
batch_shape, dtype="float32"),
axis=0)
#TODO this shoud be a hyparameter
IPS = paddle.clip(IPS, min=-15, max=15) #online trick
IPS = paddle.multiply(IPS, batch_size)
IPS.stop_gradient = True
loss_cvr = paddle.multiply(loss_cvr, IPS)
loss_cvr = paddle.multiply(loss_cvr, O)
return paddle.mean(loss_cvr)
def counterfact_dr(self, loss_cvr, O, ctr_out_one, imp_out):
#dr error part
e = paddle.subtract(loss_cvr, imp_out)
min_v = paddle.full_like(ctr_out_one, 0.000001)
ctr_out_one = paddle.maximum(ctr_out_one, min_v)
IPS = paddle.divide(paddle.cast(O, dtype="float32"), ctr_out_one)
IPS = paddle.clip(IPS, min=-15, max=15) #online trick
IPS.stop_gradient = True
loss_error_second = paddle.multiply(e, IPS)
loss_error = imp_out + loss_error_second
#dr imp part
loss_imp = paddle.square(e)
loss_imp = paddle.multiply(loss_imp, IPS)
loss_dr = loss_error + loss_imp
return paddle.mean(loss_dr)
def net(self, inputs, is_infer=False):
escm_model = ESCMLayer(
self.sparse_feature_number, self.sparse_feature_dim,
self.num_field, self.ctr_fc_sizes, self.cvr_fc_sizes,
self.expert_num, self.expert_size, self.tower_size,
self.counterfact_mode, self.feature_size)
out_list = escm_model.forward(inputs[0:-2])
ctr_out, ctr_out_one, cvr_out, cvr_out_one, ctcvr_prop, ctcvr_prop_one = out_list[
0], out_list[1], out_list[2], out_list[3], out_list[4], out_list[5]
ctr_clk = inputs[-2]
ctcvr_buy = inputs[-1]
ctr_num = paddle.sum(paddle.cast(ctr_clk, dtype="float32"), axis=0)
O = paddle.cast(ctr_clk, 'float32')
auc_ctr, batch_auc_ctr, auc_states_ctr = paddle.static.auc(
input=ctr_out, label=ctr_clk)
auc_ctcvr, batch_auc_ctcvr, auc_states_ctcvr = paddle.static.auc(
input=ctcvr_prop, label=ctcvr_buy)
auc_cvr, batch_auc_cvr, auc_states_cvr = paddle.static.auc(
input=cvr_out, label=ctcvr_buy)
if is_infer:
fetch_dict = {
'auc_ctr': auc_ctr,
'auc_cvr': auc_cvr,
'auc_ctcvr': auc_ctcvr
}
return fetch_dict
loss_ctr = paddle.nn.functional.log_loss(
input=ctr_out_one, label=paddle.cast(
ctr_clk, dtype="float32"))
loss_cvr = paddle.nn.functional.log_loss(
input=cvr_out_one, label=paddle.cast(
ctcvr_buy, dtype="float32"))
if self.counterfact_mode == "DR":
loss_cvr = self.counterfact_dr(loss_cvr, O, ctr_out_one,
out_list[6])
else:
loss_cvr = self.counterfact_ipw(loss_cvr, ctr_num, O, ctr_out_one)
loss_ctcvr = paddle.nn.functional.log_loss(
input=ctcvr_prop_one,
label=paddle.cast(
ctcvr_buy, dtype="float32"))
cost = loss_ctr + loss_cvr * self.counterfactual_w + loss_ctcvr * self.global_w
avg_cost = paddle.mean(x=cost)
self._cost = avg_cost
fetch_dict = {
'cost': avg_cost,
'auc_ctr': auc_ctr,
'auc_cvr': auc_cvr,
'auc_ctcvr': auc_ctcvr
}
return fetch_dict
def create_optimizer(self, strategy=None):
optimizer = paddle.optimizer.Adam(
learning_rate=self.learning_rate, lazy_mode=True)
if strategy != None:
import paddle.distributed.fleet as fleet
optimizer = fleet.distributed_optimizer(optimizer, strategy)
optimizer.minimize(self._cost)
def infer_net(self, input):
return self.net(input, is_infer=True)