未验证 提交 66c7f380 编写于 作者: W wuzhihua 提交者: GitHub

Merge pull request #114 from overlordmax/pr_6192341

add fibinet
...@@ -59,7 +59,8 @@ ...@@ -59,7 +59,8 @@
| 排序 | [xDeepFM](models/rank/xdeepfm/model.py) | ✓ | x | ✓ | x | [KDD 2018][xDeepFM: Combining Explicit and Implicit Feature Interactions for Recommender Systems](https://dl.acm.org/doi/pdf/10.1145/3219819.3220023) | | 排序 | [xDeepFM](models/rank/xdeepfm/model.py) | ✓ | x | ✓ | x | [KDD 2018][xDeepFM: Combining Explicit and Implicit Feature Interactions for Recommender Systems](https://dl.acm.org/doi/pdf/10.1145/3219819.3220023) |
| 排序 | [DIN](models/rank/din/model.py) | ✓ | x | ✓ | x | [KDD 2018][Deep Interest Network for Click-Through Rate Prediction](https://dl.acm.org/doi/pdf/10.1145/3219819.3219823) | | 排序 | [DIN](models/rank/din/model.py) | ✓ | x | ✓ | x | [KDD 2018][Deep Interest Network for Click-Through Rate Prediction](https://dl.acm.org/doi/pdf/10.1145/3219819.3219823) |
| 排序 | [Wide&Deep](models/rank/wide_deep/model.py) | ✓ | x | ✓ | x | [DLRS 2016][Wide & Deep Learning for Recommender Systems](https://dl.acm.org/doi/pdf/10.1145/2988450.2988454) | | 排序 | [Wide&Deep](models/rank/wide_deep/model.py) | ✓ | x | ✓ | x | [DLRS 2016][Wide & Deep Learning for Recommender Systems](https://dl.acm.org/doi/pdf/10.1145/2988450.2988454) |
| 排序 | [FGCNN](models/rank/fgcnn/model.py) | ✓ | ✓ | ✓ | ✓ | [WWW 2019][Feature Generation by Convolutional Neural Network for Click-Through Rate Prediction](https://arxiv.org/pdf/1904.04447.pdf) | | 排序 | [FGCNN](models/rank/fgcnn/model.py) | ✓ | ✓ | ✓ | ✓ | [WWW 2019][Feature Generation by Convolutional Neural Network for Click-Through Rate Prediction](https://arxiv.org/pdf/1904.04447.pdf)
| 排序 | [Fibinet](models/rank/fibinet/model.py) | ✓ | ✓ | ✓ | ✓ | [RecSys19][FiBiNET: Combining Feature Importance and Bilinear feature Interaction for Click-Through Rate Prediction]( https://arxiv.org/pdf/1905.09433.pdf) |
| 多任务 | [ESMM](models/multitask/esmm/model.py) | ✓ | ✓ | ✓ | ✓ | [SIGIR 2018][Entire Space Multi-Task Model: An Effective Approach for Estimating Post-Click Conversion Rate](https://arxiv.org/abs/1804.07931) | | 多任务 | [ESMM](models/multitask/esmm/model.py) | ✓ | ✓ | ✓ | ✓ | [SIGIR 2018][Entire Space Multi-Task Model: An Effective Approach for Estimating Post-Click Conversion Rate](https://arxiv.org/abs/1804.07931) |
| 多任务 | [MMOE](models/multitask/mmoe/model.py) | ✓ | ✓ | ✓ | ✓ | [KDD 2018][Modeling Task Relationships in Multi-task Learning with Multi-gate Mixture-of-Experts](https://dl.acm.org/doi/abs/10.1145/3219819.3220007) | | 多任务 | [MMOE](models/multitask/mmoe/model.py) | ✓ | ✓ | ✓ | ✓ | [KDD 2018][Modeling Task Relationships in Multi-task Learning with Multi-gate Mixture-of-Experts](https://dl.acm.org/doi/abs/10.1145/3219819.3220007) |
| 多任务 | [ShareBottom](models/multitask/share-bottom/model.py) | ✓ | ✓ | ✓ | ✓ | [1998][Multitask learning](http://reports-archive.adm.cs.cmu.edu/anon/1997/CMU-CS-97-203.pdf) | | 多任务 | [ShareBottom](models/multitask/share-bottom/model.py) | ✓ | ✓ | ✓ | ✓ | [1998][Multitask learning](http://reports-archive.adm.cs.cmu.edu/anon/1997/CMU-CS-97-203.pdf) |
......
doc/imgs/overview.png

217.7 KB | W: | H:

doc/imgs/overview.png

212.4 KB | W: | H:

doc/imgs/overview.png
doc/imgs/overview.png
doc/imgs/overview.png
doc/imgs/overview.png
  • 2-up
  • Swipe
  • Onion skin
# fibinet
以下是本例的简要目录结构及说明:
```
├── data #样例数据
├── sample_data
├── train
├── sample_train.txt
├── download.sh
├── run.sh
├── get_slot_data.py
├── __init__.py
├── README.md # 文档
├── model.py #模型文件
├── config.yaml #配置文件
```
## 简介
[《FiBiNET: Combining Feature Importance and Bilinear feature Interaction for Click-Through Rate Prediction》]( https://arxiv.org/pdf/1905.09433.pdf)是新浪微博机器学习团队发表在RecSys19上的一篇论文,文章指出当前的许多通过特征组合进行CTR预估的工作主要使用特征向量的内积或哈达玛积来计算交叉特征,这种方法忽略了特征本身的重要程度。提出通过使用Squeeze-Excitation network (SENET) 结构动态学习特征的重要性以及使用一个双线性函数来更好的建模交叉特征。
本项目在paddlepaddle上实现FibiNET的网络结构,并在开源数据集Criteo上验证模型效果。
## 数据下载及预处理
数据地址:[Criteo]( https://fleet.bj.bcebos.com/ctr_data.tar.gz)
(1)将原始训练集按9:1划分为训练集和验证集
(2)数值特征(连续特征)进行归一化处理
执行run.sh生成训练集和测试集
```
sh run.sh
```
## 环境
PaddlePaddle 1.7.2
python3.7
PaddleRec
## 单机训练
CPU环境
在config.yaml文件中设置好设备,epochs等。
```
# select runner by name
mode: [single_cpu_train, single_cpu_infer]
# config of each runner.
# runner is a kind of paddle training class, which wraps the train/infer process.
runner:
- name: single_cpu_train
class: train
# num of epochs
epochs: 4
# device to run training or infer
device: cpu
save_checkpoint_interval: 2 # save model interval of epochs
save_inference_interval: 4 # save inference
save_checkpoint_path: "increment_model" # save checkpoint path
save_inference_path: "inference" # save inference path
save_inference_feed_varnames: [] # feed vars of save inference
save_inference_fetch_varnames: [] # fetch vars of save inference
init_model_path: "" # load model path
print_interval: 10
phases: [phase1]
```
## 单机预测
CPU环境
在config.yaml文件中设置好epochs、device等参数。
```
- name: single_cpu_infer
class: infer
# num of epochs
epochs: 1
# device to run training or infer
device: cpu #选择预测的设备
init_model_path: "increment_dnn" # load model path
phases: [phase2]
```
## 运行
```
python -m paddlerec.run -m paddlerec.models.rank.fibinet
```
## 模型效果
在样例数据上测试模型
训练:
```
Running SingleStartup.
W0623 12:03:35.130075 509 device_context.cc:237] Please NOTE: device: 0, CUDA Capability: 70, Driver API Version: 9.2, Runtime API Version: 9.0
W0623 12:03:35.134771 509 device_context.cc:245] device: 0, cuDNN Version: 7.3.
Running SingleRunner.
batch: 100, AUC: [0.6449976], BATCH_AUC: [0.69029814]
batch: 200, AUC: [0.6769844], BATCH_AUC: [0.70255003]
batch: 300, AUC: [0.67131597], BATCH_AUC: [0.68954499]
batch: 400, AUC: [0.68129822], BATCH_AUC: [0.70892718]
batch: 500, AUC: [0.68242937], BATCH_AUC: [0.69269376]
batch: 600, AUC: [0.68741928], BATCH_AUC: [0.72034578]
...
batch: 1400, AUC: [0.84607023], BATCH_AUC: [0.93358024]
batch: 1500, AUC: [0.84796116], BATCH_AUC: [0.95302841]
batch: 1600, AUC: [0.84949111], BATCH_AUC: [0.92868531]
batch: 1700, AUC: [0.85113661], BATCH_AUC: [0.95452616]
batch: 1800, AUC: [0.85260467], BATCH_AUC: [0.92847032]
epoch 3 done, use time: 1618.1106688976288
```
预测
```
load persistables from increment_model/3
batch: 20, AUC: [0.85304064], BATCH_AUC: [0.94178556]
batch: 40, AUC: [0.85304544], BATCH_AUC: [0.95207907]
batch: 60, AUC: [0.85303907], BATCH_AUC: [0.94782551]
batch: 80, AUC: [0.85298773], BATCH_AUC: [0.93987691]
...
batch: 1780, AUC: [0.866046], BATCH_AUC: [0.96424594]
batch: 1800, AUC: [0.86633785], BATCH_AUC: [0.96900967]
batch: 1820, AUC: [0.86662365], BATCH_AUC: [0.96759972]
```
# 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.
# 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.
# workspace
workspace: "paddlerec.models.rank.fibinet"
# list of dataset
dataset:
- name: dataloader_train # name of dataset to distinguish different datasets
batch_size: 2
type: DataLoader # or QueueDataset
data_path: "{workspace}/data/sample_data/train"
sparse_slots: "click 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"
dense_slots: "dense_var:13"
- name: dataset_infer # name
batch_size: 2
type: DataLoader # or QueueDataset
data_path: "{workspace}/data/sample_data/train"
sparse_slots: "click 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"
dense_slots: "dense_var:13"
# hyper parameters of user-defined network
hyper_parameters:
# optimizer config
optimizer:
class: Adam
learning_rate: 0.001
strategy: async
# user-defined <key, value> pairs
sparse_inputs_slots: 27
sparse_feature_number: 1000001
sparse_feature_dim: 9
dense_input_dim: 13
bilinear_type: 'all'
reduction_ratio: 3
dropout_rate: 0.5
# select runner by name
mode: [single_cpu_train, single_cpu_infer]
# config of each runner.
# runner is a kind of paddle training class, which wraps the train/infer process.
runner:
- name: single_cpu_train
class: train
# num of epochs
epochs: 4
# device to run training or infer
device: cpu
save_checkpoint_interval: 2 # save model interval of epochs
save_inference_interval: 4 # save inference
save_checkpoint_path: "increment_model" # save checkpoint path
save_inference_path: "inference" # save inference path
save_inference_feed_varnames: [] # feed vars of save inference
save_inference_fetch_varnames: [] # fetch vars of save inference
init_model_path: "" # load model path
print_interval: 100
phases: [phase1]
- name: single_gpu_train
class: train
# num of epochs
epochs: 4
# device to run training or infer
device: gpu
save_checkpoint_interval: 1 # save model interval of epochs
save_inference_interval: 4 # save inference
save_checkpoint_path: "increment_model" # save checkpoint path
save_inference_path: "inference" # save inference path
save_inference_feed_varnames: [] # feed vars of save inference
save_inference_fetch_varnames: [] # fetch vars of save inference
init_model_path: "" # load model path
print_interval: 100
phases: [phase1]
- name: single_cpu_infer
class: infer
# device to run training or infer
device: cpu
init_model_path: "increment_model" # load model path
phases: [phase2]
- name: single_gpu_infer
class: infer
# device to run training or infer
device: gpu
init_model_path: "increment_model" # load model path
phases: [phase2]
# runner will run all the phase in each epoch
phase:
- name: phase1
model: "{workspace}/model.py" # user-defined model
dataset_name: dataloader_train # select dataset by name
thread_num: 8
- name: phase2
model: "{workspace}/model.py" # user-defined model
dataset_name: dataset_infer # select dataset by name
thread_num: 8
wget --no-check-certificate https://fleet.bj.bcebos.com/ctr_data.tar.gz
tar -zxvf ctr_data.tar.gz
mv ./raw_data ./train_data_full
mkdir train_data && cd train_data
cp ../train_data_full/part-0 ../train_data_full/part-1 ./ && cd ..
mv ./test_data ./test_data_full
mkdir test_data && cd test_data
cp ../test_data_full/part-220 ./ && cd ..
echo "Complete data download."
echo "Full Train data stored in ./train_data_full "
echo "Full Test data stored in ./test_data_full "
echo "Rapid Verification train data stored in ./train_data "
echo "Rapid Verification test data stored in ./test_data "
# Copyright (c) 2019 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 paddle.fluid.incubate.data_generator as dg
cont_min_ = [0, -3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
cont_max_ = [20, 600, 100, 50, 64000, 500, 100, 50, 500, 10, 10, 10, 50]
cont_diff_ = [20, 603, 100, 50, 64000, 500, 100, 50, 500, 10, 10, 10, 50]
hash_dim_ = 1000001
continuous_range_ = range(1, 14)
categorical_range_ = range(14, 40)
class CriteoDataset(dg.MultiSlotDataGenerator):
"""
DacDataset: inheritance MultiSlotDataGeneratior, Implement data reading
Help document: http://wiki.baidu.com/pages/viewpage.action?pageId=728820675
"""
def generate_sample(self, line):
"""
Read the data line by line and process it as a dictionary
"""
def reader():
"""
This function needs to be implemented by the user, based on data format
"""
features = line.rstrip('\n').split('\t')
dense_feature = []
sparse_feature = []
for idx in continuous_range_:
if features[idx] == "":
dense_feature.append(0.0)
else:
dense_feature.append(
(float(features[idx]) - cont_min_[idx - 1]) /
cont_diff_[idx - 1])
for idx in categorical_range_:
sparse_feature.append(
[hash(str(idx) + features[idx]) % hash_dim_])
label = [int(features[0])]
process_line = dense_feature, sparse_feature, label
feature_name = ["dense_feature"]
for idx in categorical_range_:
feature_name.append("C" + str(idx - 13))
feature_name.append("label")
s = "click:" + str(label[0])
for i in dense_feature:
s += " dense_feature:" + str(i)
for i in range(1, 1 + len(categorical_range_)):
s += " " + str(i) + ":" + str(sparse_feature[i - 1][0])
print(s.strip())
yield None
return reader
d = CriteoDataset()
d.run_from_stdin()
sh download.sh
mkdir slot_train_data_full
for i in `ls ./train_data_full`
do
cat train_data_full/$i | python get_slot_data.py > slot_train_data_full/$i
done
mkdir slot_test_data_full
for i in `ls ./test_data_full`
do
cat test_data_full/$i | python get_slot_data.py > slot_test_data_full/$i
done
mkdir slot_train_data
for i in `ls ./train_data`
do
cat train_data/$i | python get_slot_data.py > slot_train_data/$i
done
mkdir slot_test_data
for i in `ls ./test_data`
do
cat test_data/$i | python get_slot_data.py > slot_test_data/$i
done
click:0 dense_feature:0.0 dense_feature:0.00497512437811 dense_feature:0.05 dense_feature:0.08 dense_feature:0.207421875 dense_feature:0.028 dense_feature:0.35 dense_feature:0.08 dense_feature:0.082 dense_feature:0.0 dense_feature:0.4 dense_feature:0.0 dense_feature:0.08 1:737395 2:210498 3:903564 4:286224 5:286835 6:906818 7:906116 8:67180 9:27346 10:51086 11:142177 12:95024 13:157883 14:873363 15:600281 16:812592 17:228085 18:35900 19:880474 20:984402 21:100885 22:26235 23:410878 24:798162 25:499868 26:306163
click:1 dense_feature:0.0 dense_feature:0.932006633499 dense_feature:0.02 dense_feature:0.14 dense_feature:0.0395625 dense_feature:0.328 dense_feature:0.98 dense_feature:0.12 dense_feature:1.886 dense_feature:0.0 dense_feature:1.8 dense_feature:0.0 dense_feature:0.14 1:715353 2:761523 3:432904 4:892267 5:515218 6:948614 7:266726 8:67180 9:27346 10:266081 11:286126 12:789480 13:49621 14:255651 15:47663 16:79797 17:342789 18:616331 19:880474 20:984402 21:242209 22:26235 23:669531 24:26284 25:269955 26:187951
click:0 dense_feature:0.0 dense_feature:0.00829187396352 dense_feature:0.08 dense_feature:0.06 dense_feature:0.14125 dense_feature:0.076 dense_feature:0.05 dense_feature:0.22 dense_feature:0.208 dense_feature:0.0 dense_feature:0.2 dense_feature:0.0 dense_feature:0.06 1:737395 2:952384 3:511141 4:271077 5:286835 6:948614 7:903547 8:507110 9:27346 10:56047 11:612953 12:747707 13:977426 14:671506 15:158148 16:833738 17:342789 18:427155 19:880474 20:537425 21:916237 22:26235 23:468277 24:676936 25:751788 26:363967
click:0 dense_feature:0.0 dense_feature:0.124378109453 dense_feature:0.02 dense_feature:0.04 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.08 dense_feature:0.024 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.04 1:210127 2:286436 3:183920 4:507656 5:286835 6:906818 7:199553 8:67180 9:502607 10:708281 11:809876 12:888238 13:375164 14:202774 15:459895 16:475933 17:555571 18:847163 19:26230 20:26229 21:808836 22:191474 23:410878 24:315120 25:26224 26:26223
click:0 dense_feature:0.1 dense_feature:0.0149253731343 dense_feature:0.34 dense_feature:0.32 dense_feature:0.016421875 dense_feature:0.098 dense_feature:0.04 dense_feature:0.96 dense_feature:0.202 dense_feature:0.1 dense_feature:0.2 dense_feature:0.0 dense_feature:0.32 1:230803 2:817085 3:539110 4:388629 5:286835 6:948614 7:586040 8:67180 9:27346 10:271155 11:176640 12:827381 13:36881 14:202774 15:397299 16:411672 17:342789 18:474060 19:880474 20:984402 21:216871 22:26235 23:761351 24:787115 25:884722 26:904135
click:0 dense_feature:0.0 dense_feature:0.00829187396352 dense_feature:0.13 dense_feature:0.04 dense_feature:0.246203125 dense_feature:0.108 dense_feature:0.05 dense_feature:0.04 dense_feature:0.03 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.04 1:737395 2:64837 3:259267 4:336976 5:515218 6:154084 7:847938 8:67180 9:27346 10:708281 11:776766 12:964800 13:324323 14:873363 15:212708 16:637238 17:681378 18:895034 19:673458 20:984402 21:18600 22:26235 23:410878 24:787115 25:884722 26:355412
click:0 dense_feature:0.0 dense_feature:0.028192371476 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0245625 dense_feature:0.016 dense_feature:0.04 dense_feature:0.12 dense_feature:0.016 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.0 1:737395 2:554760 3:661483 4:263696 5:938478 6:906818 7:786926 8:67180 9:27346 10:245862 11:668197 12:745676 13:432600 14:413795 15:751427 16:272410 17:342789 18:422136 19:26230 20:26229 21:452501 22:26235 23:51381 24:776636 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00497512437811 dense_feature:1.95 dense_feature:0.28 dense_feature:0.092828125 dense_feature:0.57 dense_feature:0.06 dense_feature:0.4 dense_feature:0.4 dense_feature:0.0 dense_feature:0.2 dense_feature:0.0 dense_feature:0.4 1:371155 2:817085 3:773609 4:555449 5:938478 6:906818 7:166117 8:507110 9:27346 10:545822 11:316654 12:172765 13:989600 14:255651 15:792372 16:606361 17:342789 18:566554 19:880474 20:984402 21:235256 22:191474 23:700326 24:787115 25:884722 26:569095
click:0 dense_feature:0.0 dense_feature:0.0912106135987 dense_feature:0.01 dense_feature:0.02 dense_feature:0.06625 dense_feature:0.018 dense_feature:0.05 dense_feature:0.06 dense_feature:0.098 dense_feature:0.0 dense_feature:0.4 dense_feature:0.0 dense_feature:0.04 1:230803 2:531472 3:284417 4:661677 5:938478 6:553107 7:21150 8:49466 9:27346 10:526914 11:164508 12:631773 13:882348 14:873363 15:523948 16:687081 17:342789 18:271301 19:26230 20:26229 21:647160 22:26235 23:410878 24:231695 25:26224 26:26223
click:1 dense_feature:0.0 dense_feature:0.00663349917081 dense_feature:0.01 dense_feature:0.02 dense_feature:0.02153125 dense_feature:0.092 dense_feature:0.05 dense_feature:0.68 dense_feature:0.472 dense_feature:0.0 dense_feature:0.3 dense_feature:0.0 dense_feature:0.02 1:737395 2:532829 3:320762 4:887282 5:286835 6:25207 7:640357 8:67180 9:27346 10:695831 11:739268 12:835325 13:402539 14:873363 15:125813 16:168896 17:342789 18:374414 19:26230 20:26229 21:850229 22:26235 23:410878 24:480027 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00497512437811 dense_feature:0.05 dense_feature:0.04 dense_feature:0.086125 dense_feature:0.098 dense_feature:0.15 dense_feature:0.06 dense_feature:0.228 dense_feature:0.0 dense_feature:0.2 dense_feature:0.0 dense_feature:0.04 1:210127 2:999497 3:646348 4:520638 5:938478 6:906818 7:438398 8:67180 9:27346 10:975902 11:532544 12:708828 13:815045 14:255651 15:896230 16:663630 17:342789 18:820094 19:687226 20:537425 21:481536 22:26235 23:761351 24:888170 25:250729 26:381125
click:1 dense_feature:0.1 dense_feature:0.00331674958541 dense_feature:0.02 dense_feature:0.02 dense_feature:0.00078125 dense_feature:0.002 dense_feature:0.73 dense_feature:0.08 dense_feature:0.254 dense_feature:0.1 dense_feature:1.4 dense_feature:0.0 dense_feature:0.02 1:715353 2:342833 3:551901 4:73418 5:286835 6:446063 7:219517 8:67180 9:27346 10:668726 11:40711 12:921745 13:361076 14:15048 15:214564 16:400893 17:228085 18:393370 19:26230 20:26229 21:383046 22:26235 23:700326 24:369764 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.142620232172 dense_feature:0.04 dense_feature:0.1 dense_feature:0.08853125 dense_feature:0.028 dense_feature:0.01 dense_feature:0.1 dense_feature:0.028 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.1 1:737395 2:583707 3:519411 4:19103 5:286835 6:906818 7:801403 8:67180 9:27346 10:35743 11:626052 12:142351 13:988058 14:873363 15:617333 16:850339 17:276641 18:696084 19:26230 20:26229 21:121620 22:191474 23:468277 24:18340 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00995024875622 dense_feature:0.0 dense_feature:0.22 dense_feature:0.0251875 dense_feature:0.0 dense_feature:0.0 dense_feature:0.8 dense_feature:0.182 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.84 1:737395 2:19359 3:166075 4:381832 5:286835 6:446063 7:816009 8:67180 9:27346 10:708281 11:619790 12:524128 13:826787 14:202774 15:371495 16:392894 17:644532 18:271180 19:26230 20:26229 21:349978 22:26235 23:761351 24:517170 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.0149253731343 dense_feature:0.52 dense_feature:0.1 dense_feature:6.25153125 dense_feature:0.0 dense_feature:0.0 dense_feature:0.3 dense_feature:0.03 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.1 1:230803 2:24784 3:519411 4:19103 5:843054 6:948614 7:529143 8:67180 9:502607 10:708281 11:430027 12:142351 13:529101 14:202774 15:618316 16:850339 17:644532 18:95370 19:880474 20:31181 21:121620 22:26235 23:744389 24:18340 25:269955 26:683431
click:0 dense_feature:0.0 dense_feature:0.0480928689884 dense_feature:0.12 dense_feature:0.22 dense_feature:0.541703125 dense_feature:1.062 dense_feature:0.01 dense_feature:0.24 dense_feature:0.054 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.22 1:737395 2:378661 3:21539 4:552097 5:286835 6:553107 7:512138 8:67180 9:27346 10:708281 11:91094 12:516991 13:150114 14:873363 15:450569 16:353024 17:228085 18:539379 19:26230 20:26229 21:410733 22:26235 23:700326 24:272703 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.016583747927 dense_feature:0.06 dense_feature:0.0 dense_feature:0.209625 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.09 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 1:737395 2:750131 3:807749 4:905739 5:286835 6:906818 7:11935 8:67180 9:27346 10:708281 11:505199 12:285350 13:724106 14:255651 15:625913 16:511836 17:644532 18:102288 19:26230 20:26229 21:726818 22:179327 23:744389 24:176417 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00663349917081 dense_feature:0.05 dense_feature:0.14 dense_feature:0.226703125 dense_feature:0.12 dense_feature:0.05 dense_feature:0.14 dense_feature:0.112 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.14 1:736218 2:690313 3:757279 4:763330 5:286835 6:553107 7:89560 8:642551 9:27346 10:128328 11:281593 12:246510 13:200341 14:255651 15:899145 16:807138 17:342789 18:659853 19:26230 20:26229 21:399608 22:26235 23:669531 24:787115 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00829187396352 dense_feature:0.3 dense_feature:0.2 dense_feature:0.021296875 dense_feature:0.83 dense_feature:0.2 dense_feature:0.56 dense_feature:1.122 dense_feature:0.0 dense_feature:0.5 dense_feature:0.0 dense_feature:0.2 1:715353 2:283434 3:523722 4:590869 5:286835 6:948614 7:25472 8:67180 9:27346 10:340404 11:811342 12:679454 13:897590 14:813514 15:578769 16:962576 17:342789 18:267210 19:310188 20:537425 21:746185 22:179327 23:761351 24:416923 25:253255 26:249672
click:1 dense_feature:0.05 dense_feature:0.0149253731343 dense_feature:0.03 dense_feature:0.24 dense_feature:0.0 dense_feature:0.008 dense_feature:0.4 dense_feature:0.62 dense_feature:0.82 dense_feature:0.1 dense_feature:1.4 dense_feature:0.0 dense_feature:0.08 1:715353 2:532829 3:716475 4:940968 5:286835 6:948614 7:38171 8:67180 9:27346 10:619455 11:515541 12:779426 13:711791 14:255651 15:881750 16:408550 17:342789 18:612540 19:26230 20:26229 21:23444 22:26235 23:410878 24:88425 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00497512437811 dense_feature:0.11 dense_feature:0.08 dense_feature:0.135265625 dense_feature:0.426 dense_feature:0.06 dense_feature:0.06 dense_feature:0.42 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.08 1:737395 2:817085 3:506158 4:48876 5:286835 6:948614 7:95506 8:67180 9:27346 10:75825 11:220591 12:613471 13:159874 14:255651 15:121379 16:889290 17:681378 18:532453 19:880474 20:537425 21:717912 22:26235 23:270873 24:450199 25:884722 26:382723
click:0 dense_feature:0.0 dense_feature:0.0829187396352 dense_feature:0.0 dense_feature:0.0 dense_feature:0.555859375 dense_feature:0.318 dense_feature:0.03 dense_feature:0.0 dense_feature:0.02 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.0 1:715353 2:465222 3:974451 4:892661 5:938478 6:948614 7:651987 8:67180 9:27346 10:708281 11:229311 12:545057 13:875629 14:149134 15:393524 16:213237 17:681378 18:540092 19:26230 20:26229 21:483290 22:26235 23:700326 24:946673 25:26224 26:26223
click:1 dense_feature:0.05 dense_feature:0.854063018242 dense_feature:0.01 dense_feature:0.04 dense_feature:0.000171875 dense_feature:0.004 dense_feature:0.01 dense_feature:0.04 dense_feature:0.004 dense_feature:0.1 dense_feature:0.1 dense_feature:0.0 dense_feature:0.04 1:737395 2:99294 3:681584 4:398205 5:914075 6:906818 7:620358 8:67180 9:27346 10:147441 11:364583 12:535262 13:516341 14:813514 15:281303 16:714384 17:276641 18:443922 19:26230 20:26229 21:948746 22:26235 23:700326 24:928903 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00331674958541 dense_feature:0.0 dense_feature:0.0 dense_feature:0.45190625 dense_feature:0.048 dense_feature:0.01 dense_feature:0.16 dense_feature:0.044 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.0 1:737395 2:792512 3:676584 4:995262 5:938478 6:906818 7:888723 8:67180 9:27346 10:708281 11:310529 12:951172 13:885793 14:873363 15:62698 16:672021 17:276641 18:11502 19:880474 20:984402 21:501083 22:191474 23:744389 24:398029 25:218743 26:991064
click:0 dense_feature:0.0 dense_feature:0.00663349917081 dense_feature:0.51 dense_feature:0.0 dense_feature:0.2689375 dense_feature:0.0 dense_feature:0.0 dense_feature:0.02 dense_feature:0.006 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 1:230803 2:239052 3:323170 4:474182 5:140103 6:553107 7:757837 8:524745 9:27346 10:743444 11:883533 12:123023 13:621127 14:255651 15:570872 16:883618 17:924903 18:984920 19:964183 20:984402 21:260134 22:179327 23:410878 24:787860 25:269955 26:949924
click:0 dense_feature:0.0 dense_feature:0.273631840796 dense_feature:0.0 dense_feature:0.0 dense_feature:0.066453125 dense_feature:0.052 dense_feature:0.04 dense_feature:0.06 dense_feature:0.01 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.0 1:737395 2:531472 3:747313 4:362684 5:843054 6:553107 7:863980 8:718499 9:27346 10:881217 11:371751 12:168971 13:290788 14:202774 15:316669 16:269663 17:342789 18:136775 19:26230 20:26229 21:76865 22:26235 23:761351 24:441421 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.116086235489 dense_feature:0.43 dense_feature:0.36 dense_feature:0.000953125 dense_feature:0.0 dense_feature:0.0 dense_feature:0.36 dense_feature:0.036 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.36 1:737395 2:24784 3:677469 4:820784 5:286835 6:553107 7:715520 8:718499 9:27346 10:708281 11:670424 12:122926 13:724619 14:873363 15:845517 16:488791 17:644532 18:183573 19:880474 20:31181 21:46761 22:26235 23:700326 24:629361 25:269955 26:862373
click:0 dense_feature:2.55 dense_feature:0.0348258706468 dense_feature:0.01 dense_feature:0.38 dense_feature:0.001453125 dense_feature:0.046 dense_feature:1.11 dense_feature:0.44 dense_feature:2.312 dense_feature:0.2 dense_feature:1.1 dense_feature:0.0 dense_feature:0.46 1:594517 2:194636 3:496284 4:323209 5:286835 6:553107 7:259696 8:760861 9:27346 10:698046 11:478868 12:576074 13:635369 14:201966 15:926692 16:972906 17:342789 18:409802 19:26230 20:26229 21:395694 22:26235 23:410878 24:844671 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.144278606965 dense_feature:0.43 dense_feature:0.22 dense_feature:0.00309375 dense_feature:0.15 dense_feature:0.14 dense_feature:0.54 dense_feature:0.152 dense_feature:0.0 dense_feature:0.2 dense_feature:0.1 dense_feature:0.22 1:737395 2:239052 3:456744 4:736474 5:286835 6:948614 7:13277 8:67180 9:27346 10:958384 11:778183 12:497627 13:136915 14:201966 15:757961 16:747483 17:228085 18:984920 19:905920 20:537425 21:472149 22:179327 23:410878 24:709155 25:269955 26:618673
click:0 dense_feature:0.0 dense_feature:0.0132669983416 dense_feature:0.4 dense_feature:0.3 dense_feature:0.36440625 dense_feature:1.492 dense_feature:0.07 dense_feature:0.3 dense_feature:1.048 dense_feature:0.0 dense_feature:0.3 dense_feature:0.0 dense_feature:0.3 1:737395 2:19959 3:661391 4:748753 5:286835 6:948614 7:848540 8:67180 9:27346 10:708281 11:703964 12:72024 13:336272 14:255651 15:835686 16:703858 17:342789 18:274368 19:26230 20:26229 21:765452 22:26235 23:700326 24:815200 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.0116086235489 dense_feature:0.01 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 1:210127 2:662691 3:334228 4:857003 5:286835 6:25207 7:280499 8:67180 9:502607 10:708281 11:195094 12:870026 13:783566 14:873363 15:139595 16:214259 17:555571 18:208248 19:880474 20:984402 21:471770 22:26235 23:744389 24:507551 25:383787 26:797121
click:1 dense_feature:0.0 dense_feature:0.0348258706468 dense_feature:0.03 dense_feature:0.02 dense_feature:0.066140625 dense_feature:0.006 dense_feature:0.17 dense_feature:0.02 dense_feature:0.236 dense_feature:0.0 dense_feature:0.5 dense_feature:0.0 dense_feature:0.02 1:230803 2:999497 3:25361 4:892267 5:286835 6:906818 7:356528 8:67180 9:27346 10:5856 11:157692 12:554754 13:442501 14:255651 15:896230 16:248781 17:342789 18:820094 19:905920 20:984402 21:916436 22:26235 23:669531 24:26284 25:884722 26:187951
click:0 dense_feature:0.0 dense_feature:4.62852404643 dense_feature:0.07 dense_feature:0.0 dense_feature:0.022671875 dense_feature:0.0 dense_feature:0.01 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.0 1:624252 2:344887 3:238747 4:308366 5:286835 6:553107 7:69291 8:67180 9:27346 10:781054 11:258240 12:546906 13:772337 14:873363 15:807640 16:525695 17:276641 18:613203 19:438655 20:984402 21:415123 22:191474 23:700326 24:729290 25:218743 26:953507
click:0 dense_feature:0.0 dense_feature:0.00663349917081 dense_feature:0.06 dense_feature:0.02 dense_feature:0.06878125 dense_feature:0.044 dense_feature:0.01 dense_feature:0.22 dense_feature:0.044 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.02 1:737395 2:7753 3:871178 4:183530 5:286835 6:906818 7:273988 8:507110 9:27346 10:708281 11:942072 12:775997 13:612590 14:873363 15:669921 16:639940 17:681378 18:421122 19:880474 20:984402 21:410471 22:26235 23:410878 24:228420 25:269955 26:616000
click:0 dense_feature:0.0 dense_feature:0.212271973466 dense_feature:0.02 dense_feature:0.28 dense_feature:0.113421875 dense_feature:0.06 dense_feature:0.02 dense_feature:0.28 dense_feature:0.194 dense_feature:0.0 dense_feature:0.2 dense_feature:0.0 dense_feature:0.28 1:210127 2:228963 3:692240 4:389834 5:938478 6:948614 7:125690 8:507110 9:27346 10:708281 11:549232 12:308284 13:262461 14:255651 15:629185 16:280660 17:276641 18:886164 19:26230 20:26229 21:367919 22:191474 23:700326 24:520083 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:1.01658374793 dense_feature:0.01 dense_feature:0.02 dense_feature:0.11759375 dense_feature:0.08 dense_feature:0.02 dense_feature:0.02 dense_feature:0.024 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.02 1:230803 2:7753 3:194720 4:831884 5:286835 6:553107 7:620358 8:67180 9:27346 10:843010 11:424144 12:615986 13:516341 14:813514 15:782575 16:775856 17:342789 18:421122 19:880474 20:984402 21:110090 22:191474 23:700326 24:784174 25:269955 26:101161
click:0 dense_feature:0.0 dense_feature:0.00663349917081 dense_feature:0.59 dense_feature:0.06 dense_feature:0.04321875 dense_feature:0.192 dense_feature:0.02 dense_feature:0.08 dense_feature:0.014 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.06 1:230803 2:532829 3:26258 4:853241 5:938478 6:948614 7:877607 8:67180 9:27346 10:613723 11:246387 12:538673 13:377975 14:873363 15:659013 16:601478 17:681378 18:199271 19:26230 20:26229 21:300137 22:26235 23:410878 24:372458 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.06135986733 dense_feature:0.0 dense_feature:0.0 dense_feature:0.294671875 dense_feature:0.212 dense_feature:0.26 dense_feature:0.0 dense_feature:0.034 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.0 1:737395 2:154478 3:982044 4:501457 5:819883 6:906818 7:445051 8:67180 9:27346 10:976970 11:783630 12:609883 13:358461 14:15048 15:409791 16:756307 17:342789 18:480228 19:26230 20:26229 21:845147 22:26235 23:669531 24:124290 25:26224 26:26223
click:1 dense_feature:0.05 dense_feature:0.537313432836 dense_feature:0.0 dense_feature:0.02 dense_feature:0.018578125 dense_feature:0.016 dense_feature:0.16 dense_feature:0.22 dense_feature:0.192 dense_feature:0.1 dense_feature:0.3 dense_feature:0.0 dense_feature:0.02 1:737395 2:194636 3:274597 4:418981 5:286835 6:553107 7:553528 8:67180 9:27346 10:901359 11:110700 12:108037 13:915461 14:255651 15:951604 16:421384 17:342789 18:728110 19:26230 20:26229 21:772733 22:191474 23:761351 24:844671 25:26224 26:26223
click:0 dense_feature:0.1 dense_feature:0.00663349917081 dense_feature:0.16 dense_feature:0.26 dense_feature:0.00509375 dense_feature:0.122 dense_feature:0.03 dense_feature:0.94 dense_feature:0.526 dense_feature:0.1 dense_feature:0.1 dense_feature:0.0 dense_feature:1.1 1:210127 2:344887 3:343793 4:917598 5:286835 6:948614 7:220413 8:67180 9:27346 10:912799 11:370606 12:722621 13:569604 14:255651 15:499545 16:159495 17:342789 18:613203 19:305384 20:984402 21:844602 22:26235 23:410878 24:695516 25:218743 26:729263
click:0 dense_feature:0.0 dense_feature:0.00497512437811 dense_feature:0.09 dense_feature:0.16 dense_feature:0.11221875 dense_feature:0.51 dense_feature:0.09 dense_feature:0.48 dense_feature:0.088 dense_feature:0.0 dense_feature:0.4 dense_feature:0.0 dense_feature:0.16 1:737395 2:532829 3:579624 4:980109 5:286835 6:948614 7:927736 8:67180 9:27346 10:970644 11:931289 12:377125 13:539272 14:873363 15:555779 16:405069 17:342789 18:701770 19:26230 20:26229 21:201088 22:26235 23:410878 24:113994 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.182421227197 dense_feature:0.01 dense_feature:0.02 dense_feature:0.000109375 dense_feature:0.978 dense_feature:0.01 dense_feature:0.02 dense_feature:0.062 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.02 1:478318 2:158107 3:508317 4:452336 5:286835 6:948614 7:620358 8:67180 9:27346 10:147441 11:364583 12:34025 13:516341 14:873363 15:502825 16:683439 17:681378 18:889198 19:26230 20:26229 21:234451 22:26235 23:700326 24:256238 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.469320066335 dense_feature:0.2 dense_feature:0.2 dense_feature:0.0705 dense_feature:0.102 dense_feature:0.05 dense_feature:0.22 dense_feature:0.194 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.2 1:715353 2:846239 3:573061 4:508181 5:286835 6:553107 7:892443 8:718499 9:27346 10:639370 11:866496 12:791636 13:895012 14:873363 15:362079 16:16082 17:228085 18:994402 19:880474 20:984402 21:35513 22:26235 23:669531 24:520197 25:934391 26:625657
click:0 dense_feature:0.0 dense_feature:0.0729684908789 dense_feature:0.06 dense_feature:0.04 dense_feature:5.620296875 dense_feature:0.0 dense_feature:0.0 dense_feature:0.06 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.04 1:399845 2:239052 3:334610 4:593315 5:286835 6:948614 7:751495 8:67180 9:502607 10:111048 11:244081 12:115252 13:915518 14:873363 15:817451 16:296052 17:276641 18:984920 19:774721 20:984402 21:930636 22:26235 23:700326 24:975048 25:269955 26:266439
click:1 dense_feature:0.05 dense_feature:0.0265339966833 dense_feature:0.07 dense_feature:0.22 dense_feature:1.5625e-05 dense_feature:0.008 dense_feature:0.04 dense_feature:0.36 dense_feature:0.088 dense_feature:0.1 dense_feature:0.3 dense_feature:0.0 dense_feature:0.08 1:737395 2:64837 3:534435 4:555449 5:286835 6:25207 7:661236 8:67180 9:27346 10:708281 11:785752 12:47348 13:524553 14:117289 15:776971 16:293528 17:681378 18:102169 19:758208 20:31181 21:27506 22:26235 23:410878 24:787115 25:884722 26:605635
click:1 dense_feature:0.1 dense_feature:0.0464344941957 dense_feature:0.0 dense_feature:0.04 dense_feature:0.00059375 dense_feature:0.004 dense_feature:0.02 dense_feature:0.04 dense_feature:0.004 dense_feature:0.1 dense_feature:0.1 dense_feature:0.0 dense_feature:0.04 1:230803 2:7753 3:529866 4:437169 5:938478 6:948614 7:17274 8:67180 9:27346 10:461781 11:452641 12:302471 13:49621 14:873363 15:543432 16:858509 17:681378 18:402164 19:880474 20:984402 21:650184 22:191474 23:410878 24:492581 25:269955 26:217228
click:0 dense_feature:0.55 dense_feature:0.00829187396352 dense_feature:0.03 dense_feature:0.0 dense_feature:0.0014375 dense_feature:0.004 dense_feature:0.36 dense_feature:0.0 dense_feature:0.042 dense_feature:0.1 dense_feature:0.4 dense_feature:0.0 dense_feature:0.0 1:26973 2:817085 3:961160 4:355882 5:843054 6:906818 7:417593 8:67180 9:27346 10:708281 11:402889 12:899379 13:552051 14:202774 15:532679 16:545549 17:342789 18:562805 19:880474 20:31181 21:355920 22:26235 23:700326 24:787115 25:884722 26:115004
click:1 dense_feature:0.0 dense_feature:0.00663349917081 dense_feature:0.01 dense_feature:0.02 dense_feature:0.089296875 dense_feature:0.362 dense_feature:0.23 dense_feature:0.04 dense_feature:0.338 dense_feature:0.0 dense_feature:0.4 dense_feature:0.0 dense_feature:0.02 1:230803 2:977337 3:853759 4:880273 5:515218 6:25207 7:414263 8:437731 9:27346 10:205124 11:108170 12:676869 13:388798 14:255651 15:247232 16:172895 17:228085 18:543219 19:26230 20:26229 21:860937 22:179327 23:669531 24:959959 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.0945273631841 dense_feature:0.62 dense_feature:0.24 dense_feature:0.11840625 dense_feature:0.368 dense_feature:0.07 dense_feature:0.24 dense_feature:0.144 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.48 1:737395 2:532829 3:805087 4:186661 5:286835 6:154084 7:468059 8:718499 9:27346 10:708281 11:968875 12:8177 13:47822 14:255651 15:979316 16:956543 17:342789 18:541633 19:26230 20:26229 21:646669 22:26235 23:410878 24:184909 25:26224 26:26223
click:0 dense_feature:0.3 dense_feature:0.00497512437811 dense_feature:0.12 dense_feature:0.12 dense_feature:0.002890625 dense_feature:0.074 dense_feature:0.06 dense_feature:0.14 dense_feature:0.074 dense_feature:0.1 dense_feature:0.1 dense_feature:0.0 dense_feature:0.74 1:737395 2:64837 3:967865 4:249418 5:938478 6:948614 7:228716 8:67180 9:27346 10:627362 11:722606 12:193782 13:348283 14:255651 15:928582 16:221557 17:342789 18:895034 19:384556 20:984402 21:475712 22:26235 23:410878 24:492875 25:884722 26:468964
click:0 dense_feature:0.0 dense_feature:0.177446102819 dense_feature:0.01 dense_feature:0.02 dense_feature:0.041859375 dense_feature:0.0 dense_feature:0.0 dense_feature:0.16 dense_feature:0.036 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.02 1:154064 2:834620 3:25206 4:25205 5:938478 6:948614 7:134101 8:92608 9:27346 10:708281 11:505199 12:25711 13:724106 14:671506 15:42927 16:25723 17:644532 18:1957 19:26230 20:26229 21:26236 22:26235 23:744389 24:26233 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:5.61691542289 dense_feature:0.0 dense_feature:0.1 dense_feature:0.043796875 dense_feature:0.302 dense_feature:0.13 dense_feature:0.22 dense_feature:0.3 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.22 1:154184 2:19359 3:166075 4:381832 5:286835 6:906818 7:348227 8:49466 9:27346 10:645596 11:951584 12:524128 13:277250 14:255651 15:853732 16:392894 17:342789 18:619939 19:26230 20:26229 21:349978 22:26235 23:700326 24:517170 25:26224 26:26223
click:1 dense_feature:0.0 dense_feature:0.00331674958541 dense_feature:0.0 dense_feature:0.0 dense_feature:0.093234375 dense_feature:0.022 dense_feature:0.04 dense_feature:0.02 dense_feature:0.02 dense_feature:0.0 dense_feature:0.2 dense_feature:0.0 dense_feature:0.0 1:715353 2:485136 3:386313 4:208181 5:286835 6:25207 7:227715 8:49466 9:27346 10:437476 11:733250 12:721260 13:389832 14:255651 15:47178 16:761962 17:342789 18:813169 19:26230 20:26229 21:464938 22:26235 23:410878 24:833196 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.134328358209 dense_feature:0.0 dense_feature:0.14 dense_feature:0.00015625 dense_feature:0.0 dense_feature:0.0 dense_feature:0.14 dense_feature:0.014 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.14 1:737395 2:488655 3:221719 4:442408 5:286835 6:25207 7:898902 8:718499 9:27346 10:457066 11:290973 12:533168 13:949027 14:873363 15:270294 16:934635 17:924903 18:763017 19:880474 20:31181 21:517486 22:26235 23:410878 24:588215 25:499868 26:980179
click:1 dense_feature:0.0 dense_feature:0.00331674958541 dense_feature:0.0 dense_feature:0.0 dense_feature:0.023578125 dense_feature:0.0 dense_feature:0.04 dense_feature:0.0 dense_feature:0.046 dense_feature:0.0 dense_feature:0.3 dense_feature:0.0 dense_feature:0.0 1:737395 2:729012 3:691820 4:351286 5:938478 6:553107 7:21150 8:67180 9:27346 10:947459 11:164508 12:205079 13:882348 14:255651 15:178324 16:282716 17:342789 18:193902 19:880474 20:31181 21:604480 22:191474 23:669531 24:727223 25:499868 26:236426
click:1 dense_feature:0.1 dense_feature:0.00331674958541 dense_feature:0.0 dense_feature:0.0 dense_feature:0.00859375 dense_feature:0.006 dense_feature:1.55 dense_feature:0.16 dense_feature:0.06 dense_feature:0.2 dense_feature:1.6 dense_feature:0.0 dense_feature:0.0 1:712372 2:235347 3:483718 4:382039 5:914075 6:906818 7:727609 8:154004 9:27346 10:116648 11:40711 12:658199 13:361076 14:15048 15:15058 16:644988 17:342789 18:544170 19:26230 20:26229 21:251535 22:26235 23:700326 24:114111 25:26224 26:26223
click:1 dense_feature:0.25 dense_feature:0.192371475954 dense_feature:0.06 dense_feature:0.36 dense_feature:0.0 dense_feature:0.02 dense_feature:0.09 dense_feature:0.42 dense_feature:0.042 dense_feature:0.2 dense_feature:0.3 dense_feature:0.3 dense_feature:0.0 1:737395 2:288975 3:885137 4:368487 5:515218 6:906818 7:569753 8:799133 9:27346 10:635043 11:883202 12:780104 13:492605 14:873363 15:234451 16:94894 17:796504 18:653705 19:880474 20:984402 21:400692 22:26235 23:410878 24:767424 25:934391 26:958132
click:1 dense_feature:0.15 dense_feature:0.0398009950249 dense_feature:0.02 dense_feature:0.04 dense_feature:1.5625e-05 dense_feature:0.0 dense_feature:0.06 dense_feature:0.04 dense_feature:0.026 dense_feature:0.1 dense_feature:0.3 dense_feature:0.0 dense_feature:0.0 1:715353 2:532829 3:721632 4:377785 5:286835 6:553107 7:959856 8:718499 9:27346 10:737746 11:432444 12:706936 13:169268 14:873363 15:896219 16:461005 17:342789 18:286597 19:26230 20:26229 21:602049 22:26235 23:700326 24:510447 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00663349917081 dense_feature:0.05 dense_feature:0.08 dense_feature:0.155421875 dense_feature:0.55 dense_feature:0.08 dense_feature:0.24 dense_feature:1.73 dense_feature:0.0 dense_feature:0.3 dense_feature:0.0 dense_feature:0.08 1:737395 2:288975 3:385122 4:57409 5:286835 6:25207 7:339181 8:67180 9:27346 10:284863 11:531306 12:229544 13:32168 14:117289 15:632422 16:615549 17:342789 18:240865 19:880474 20:984402 21:253725 22:26235 23:410878 24:837371 25:934391 26:948190
click:0 dense_feature:0.0 dense_feature:0.0398009950249 dense_feature:0.06 dense_feature:0.12 dense_feature:0.11359375 dense_feature:0.55 dense_feature:0.03 dense_feature:0.12 dense_feature:0.186 dense_feature:0.0 dense_feature:0.2 dense_feature:0.0 dense_feature:0.12 1:737395 2:158107 3:738359 4:343895 5:286835 6:948614 7:513189 8:760861 9:27346 10:741641 11:214926 12:142871 13:753229 14:873363 15:502825 16:864586 17:681378 18:889198 19:26230 20:26229 21:368414 22:191474 23:410878 24:256238 25:26224 26:26223
click:1 dense_feature:0.25 dense_feature:0.00663349917081 dense_feature:0.03 dense_feature:0.04 dense_feature:7.8125e-05 dense_feature:0.0 dense_feature:0.48 dense_feature:0.06 dense_feature:0.004 dense_feature:0.2 dense_feature:1.3 dense_feature:0.0 dense_feature:0.0 1:737395 2:414770 3:100889 4:981572 5:286835 6:446063 7:600430 8:507110 9:27346 10:566014 11:40711 12:330691 13:361076 14:15048 15:176957 16:759140 17:342789 18:212244 19:26230 20:26229 21:688637 22:26235 23:634287 24:762432 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00663349917081 dense_feature:0.04 dense_feature:0.02 dense_feature:0.109765625 dense_feature:0.202 dense_feature:0.13 dense_feature:0.02 dense_feature:0.078 dense_feature:0.0 dense_feature:0.1 dense_feature:0.1 dense_feature:0.02 1:737395 2:7753 3:871178 4:183530 5:286835 6:948614 7:358953 8:718499 9:27346 10:837400 11:432444 12:775997 13:169268 14:255651 15:250644 16:639940 17:342789 18:421122 19:880474 20:984402 21:410471 22:26235 23:410878 24:228420 25:269955 26:870795
click:0 dense_feature:0.05 dense_feature:0.162520729685 dense_feature:0.28 dense_feature:0.16 dense_feature:0.001046875 dense_feature:0.028 dense_feature:1.03 dense_feature:0.84 dense_feature:0.534 dense_feature:0.1 dense_feature:2.3 dense_feature:0.0 dense_feature:0.28 1:737395 2:334074 3:108983 4:898979 5:286835 6:948614 7:600430 8:718499 9:27346 10:668726 11:40711 12:62821 13:361076 14:202774 15:722413 16:688170 17:342789 18:746785 19:957809 20:984402 21:96056 22:191474 23:410878 24:703372 25:129305 26:591537
click:0 dense_feature:0.2 dense_feature:0.0945273631841 dense_feature:0.02 dense_feature:0.18 dense_feature:0.021078125 dense_feature:0.046 dense_feature:0.52 dense_feature:0.44 dense_feature:0.18 dense_feature:0.1 dense_feature:0.8 dense_feature:0.0 dense_feature:0.22 1:663372 2:532829 3:714247 4:673800 5:286835 6:906818 7:219517 8:67180 9:27346 10:161916 11:40711 12:441505 13:361076 14:255651 15:992961 16:137571 17:796504 18:395194 19:26230 20:26229 21:800938 22:179327 23:410878 24:719782 25:26224 26:26223
click:1 dense_feature:0.15 dense_feature:0.24543946932 dense_feature:0.0 dense_feature:0.12 dense_feature:0.0001875 dense_feature:0.004 dense_feature:0.08 dense_feature:0.12 dense_feature:0.072 dense_feature:0.1 dense_feature:0.4 dense_feature:0.0 dense_feature:0.04 1:663372 2:70321 3:202829 4:415480 5:286835 6:553107 7:32934 8:67180 9:27346 10:1873 11:699999 12:55775 13:371214 14:873363 15:685332 16:719499 17:342789 18:135819 19:26230 20:26229 21:973542 22:852086 23:410878 24:635223 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.0679933665008 dense_feature:0.02 dense_feature:0.02 dense_feature:0.20015625 dense_feature:0.016 dense_feature:0.03 dense_feature:0.02 dense_feature:0.014 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.02 1:737395 2:229199 3:956202 4:475901 5:286835 6:948614 7:614385 8:718499 9:27346 10:171202 11:670646 12:566018 13:386065 14:873363 15:936716 16:825279 17:681378 18:758631 19:26230 20:26229 21:113534 22:26235 23:410878 24:551443 25:26224 26:26223
click:1 dense_feature:0.05 dense_feature:0.00497512437811 dense_feature:0.04 dense_feature:0.22 dense_feature:0.015921875 dense_feature:0.022 dense_feature:0.04 dense_feature:0.4 dense_feature:0.182 dense_feature:0.1 dense_feature:0.2 dense_feature:0.0 dense_feature:0.22 1:737395 2:64837 3:751736 4:291977 5:286835 6:25207 7:377931 8:718499 9:27346 10:724396 11:433484 12:517940 13:439712 14:201966 15:628624 16:780717 17:342789 18:895034 19:880474 20:31181 21:463725 22:26235 23:410878 24:787115 25:884722 26:164940
click:1 dense_feature:0.0 dense_feature:0.00995024875622 dense_feature:0.15 dense_feature:0.48 dense_feature:0.051375 dense_feature:0.0 dense_feature:0.0 dense_feature:0.06 dense_feature:0.556 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.5 1:737395 2:532829 3:158777 4:112926 5:286835 6:948614 7:764249 8:67180 9:27346 10:795273 11:330644 12:524443 13:78129 14:873363 15:127209 16:146094 17:342789 18:976129 19:26230 20:26229 21:901094 22:26235 23:410878 24:259263 25:26224 26:26223
click:1 dense_feature:0.0 dense_feature:0.00497512437811 dense_feature:1.75 dense_feature:0.0 dense_feature:0.922828125 dense_feature:1.078 dense_feature:0.0 dense_feature:0.0 dense_feature:0.112 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 1:26973 2:62956 3:428206 4:935291 5:286835 6:446063 7:360307 8:437731 9:502607 10:957425 11:626052 12:641189 13:988058 14:217110 15:637914 16:293992 17:342789 18:832710 19:774721 20:537425 21:516798 22:191474 23:700326 24:204648 25:884722 26:776972
click:1 dense_feature:1.95 dense_feature:0.00829187396352 dense_feature:0.08 dense_feature:0.1 dense_feature:0.01878125 dense_feature:0.044 dense_feature:0.42 dense_feature:0.24 dense_feature:0.358 dense_feature:0.1 dense_feature:0.2 dense_feature:0.1 dense_feature:0.26 1:737395 2:638265 3:526671 4:362576 5:938478 6:948614 7:999918 8:67180 9:27346 10:806276 11:181589 12:688684 13:367155 14:255651 15:709602 16:386859 17:228085 18:204112 19:668832 20:537425 21:541553 22:191474 23:410878 24:606704 25:49230 26:68113
click:0 dense_feature:0.0 dense_feature:0.00331674958541 dense_feature:0.0 dense_feature:0.0 dense_feature:0.38159375 dense_feature:0.022 dense_feature:0.18 dense_feature:0.0 dense_feature:0.016 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.0 1:737395 2:841163 3:284187 4:385559 5:286835 6:446063 7:311604 8:67180 9:27346 10:38910 11:76230 12:520869 13:429321 14:255651 15:296507 16:542357 17:342789 18:377250 19:880474 20:31181 21:325494 22:26235 23:410878 24:26284 25:499868 26:467348
click:0 dense_feature:0.0 dense_feature:0.00663349917081 dense_feature:0.08 dense_feature:0.0 dense_feature:0.077125 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.03 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 1:737395 2:238813 3:821667 4:209184 5:286835 6:906818 7:261420 8:67180 9:27346 10:748867 11:277196 12:790086 13:495408 14:873363 15:572266 16:281532 17:342789 18:99340 19:880474 20:537425 21:815896 22:26235 23:669531 24:17430 25:734238 26:251811
click:0 dense_feature:0.0 dense_feature:0.210613598673 dense_feature:0.01 dense_feature:0.0 dense_feature:0.041375 dense_feature:0.0 dense_feature:0.0 dense_feature:0.08 dense_feature:0.026 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 1:737395 2:532829 3:559456 4:565823 5:286835 6:948614 7:48897 8:67180 9:27346 10:708281 11:214000 12:431427 13:477774 14:873363 15:637383 16:678446 17:276641 18:849284 19:26230 20:26229 21:758879 22:26235 23:410878 24:399458 25:26224 26:26223
click:1 dense_feature:0.2 dense_feature:0.00331674958541 dense_feature:0.0 dense_feature:0.0 dense_feature:0.00440625 dense_feature:0.036 dense_feature:0.04 dense_feature:0.3 dense_feature:0.03 dense_feature:0.1 dense_feature:0.1 dense_feature:0.0 dense_feature:0.0 1:715353 2:532829 3:967094 4:707735 5:286835 6:948614 7:555710 8:154004 9:27346 10:708281 11:514992 12:158604 13:780149 14:255651 15:285282 16:149708 17:342789 18:553067 19:26230 20:26229 21:229985 22:26235 23:700326 24:777746 25:26224 26:26223
click:1 dense_feature:0.0 dense_feature:0.00331674958541 dense_feature:0.0 dense_feature:0.0 dense_feature:0.23178125 dense_feature:0.222 dense_feature:0.06 dense_feature:0.0 dense_feature:0.408 dense_feature:0.0 dense_feature:0.2 dense_feature:0.0 dense_feature:0.0 1:715353 2:227084 3:456811 4:828682 5:286835 6:948614 7:406567 8:67180 9:27346 10:66123 11:598531 12:527138 13:731439 14:813514 15:35257 16:43339 17:342789 18:918487 19:26230 20:26229 21:580653 22:26235 23:410878 24:495283 25:26224 26:26223
click:0 dense_feature:0.15 dense_feature:0.462686567164 dense_feature:0.08 dense_feature:0.22 dense_feature:0.00015625 dense_feature:0.022 dense_feature:0.03 dense_feature:0.52 dense_feature:0.022 dense_feature:0.1 dense_feature:0.1 dense_feature:0.0 dense_feature:0.22 1:576931 2:99294 3:263211 4:501662 5:938478 6:154084 7:128918 8:67180 9:27346 10:912799 11:801006 12:506258 13:378182 14:201966 15:150934 16:240427 17:681378 18:393279 19:26230 20:26229 21:152038 22:26235 23:700326 24:551443 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00331674958541 dense_feature:0.0 dense_feature:0.0 dense_feature:0.181484375 dense_feature:0.06 dense_feature:0.01 dense_feature:0.0 dense_feature:0.056 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.0 1:230803 2:283434 3:367596 4:197992 5:938478 6:948614 7:268098 8:67180 9:27346 10:870993 11:632267 12:139817 13:718764 14:255651 15:884839 16:80117 17:276641 18:556463 19:880474 20:537425 21:271358 22:26235 23:410878 24:488077 25:253255 26:584828
click:0 dense_feature:0.0 dense_feature:0.00497512437811 dense_feature:0.0 dense_feature:0.16 dense_feature:4.790078125 dense_feature:0.0 dense_feature:0.0 dense_feature:0.28 dense_feature:0.016 dense_feature:0.0 dense_feature:0.0 dense_feature:0.0 dense_feature:0.2 1:737395 2:532829 3:158777 4:112926 5:286835 6:948614 7:277312 8:67180 9:502607 10:708281 11:755513 12:524443 13:4029 14:873363 15:503814 16:146094 17:644532 18:121590 19:26230 20:26229 21:901094 22:191474 23:744389 24:259263 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:3.30845771144 dense_feature:0.0 dense_feature:0.04 dense_feature:0.022671875 dense_feature:0.062 dense_feature:0.01 dense_feature:0.4 dense_feature:0.062 dense_feature:0.0 dense_feature:0.1 dense_feature:0.0 dense_feature:0.04 1:663372 2:529436 3:511823 4:942782 5:286835 6:906818 7:190054 8:67180 9:27346 10:708281 11:32527 12:494263 13:652478 14:873363 15:616057 16:17325 17:342789 18:325238 19:26230 20:26229 21:256747 22:179327 23:410878 24:169709 25:26224 26:26223
click:0 dense_feature:0.0 dense_feature:0.00829187396352 dense_feature:0.01 dense_feature:0.16 dense_feature:0.206765625 dense_feature:0.328 dense_feature:0.13 dense_feature:0.16 dense_feature:0.176 dense_feature:0.0 dense_feature:0.7 dense_feature:0.0 dense_feature:0.16 1:737395 2:552854 3:606082 4:267619 5:286835 6:948614 7:918889 8:67180 9:27346 10:708281 11:400024 12:972010 13:66330 14:255651 15:432931 16:650209 17:506108 18:212910 19:26230 20:26229 21:107726 22:26235 23:410878 24:718419 25:26224 26:26223
# 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 paddle.fluid as fluid
import itertools
from paddlerec.core.utils import envs
from paddlerec.core.model import ModelBase
class Model(ModelBase):
def __init__(self, config):
ModelBase.__init__(self, config)
def _init_hyper_parameters(self):
self.is_distributed = True if envs.get_fleet_mode().upper(
) == "PSLIB" else False
self.sparse_feature_number = envs.get_global_env(
"hyper_parameters.sparse_feature_number")
self.sparse_feature_dim = envs.get_global_env(
"hyper_parameters.sparse_feature_dim")
self.learning_rate = envs.get_global_env(
"hyper_parameters.optimizer.learning_rate")
def _SENETLayer(self, inputs, filed_size, reduction_ratio=3):
reduction_size = max(1, filed_size // reduction_ratio)
Z = fluid.layers.reduce_mean(inputs, dim=-1)
A_1 = fluid.layers.fc(
input=Z,
size=reduction_size,
param_attr=fluid.initializer.Xavier(uniform=False),
act='relu',
name='W_1')
A_2 = fluid.layers.fc(
input=A_1,
size=filed_size,
param_attr=fluid.initializer.Xavier(uniform=False),
act='relu',
name='W_2')
V = fluid.layers.elementwise_mul(
inputs, y=fluid.layers.unsqueeze(
input=A_2, axes=[2]))
return fluid.layers.split(V, num_or_sections=filed_size, dim=1)
def _BilinearInteraction(self,
inputs,
filed_size,
embedding_size,
bilinear_type="interaction"):
if bilinear_type == "all":
p = [
fluid.layers.elementwise_mul(
fluid.layers.fc(
input=v_i,
size=embedding_size,
param_attr=fluid.initializer.Xavier(uniform=False),
act=None,
name=None),
fluid.layers.squeeze(
input=v_j, axes=[1]))
for v_i, v_j in itertools.combinations(inputs, 2)
]
else:
raise NotImplementedError
return fluid.layers.concat(input=p, axis=1)
def _DNNLayer(self, inputs, dropout_rate=0.5):
deep_input = inputs
for i, hidden_unit in enumerate([400, 400, 400]):
fc_out = fluid.layers.fc(
input=deep_input,
size=hidden_unit,
param_attr=fluid.initializer.Xavier(uniform=False),
act='relu',
name='d_' + str(i))
fc_out = fluid.layers.dropout(fc_out, dropout_prob=dropout_rate)
deep_input = fc_out
return deep_input
def net(self, input, is_infer=False):
self.sparse_inputs = self._sparse_data_var[1:]
self.dense_input = self._dense_data_var[0]
self.label_input = self._sparse_data_var[0]
emb = []
for data in self.sparse_inputs:
feat_emb = fluid.embedding(
input=data,
size=[self.sparse_feature_number, self.sparse_feature_dim],
param_attr=fluid.ParamAttr(
name='dis_emb',
learning_rate=5,
initializer=fluid.initializer.Xavier(
fan_in=self.sparse_feature_dim,
fan_out=self.sparse_feature_dim)),
is_sparse=True)
emb.append(feat_emb)
concat_emb = fluid.layers.concat(emb, axis=1)
filed_size = len(self.sparse_inputs)
bilinear_type = envs.get_global_env("hyper_parameters.bilinear_type")
reduction_ratio = envs.get_global_env(
"hyper_parameters.reduction_ratio")
dropout_rate = envs.get_global_env("hyper_parameters.dropout_rate")
senet_output = self._SENETLayer(concat_emb, filed_size,
reduction_ratio)
senet_bilinear_out = self._BilinearInteraction(
senet_output, filed_size, self.sparse_feature_dim, bilinear_type)
concat_emb = fluid.layers.split(
concat_emb, num_or_sections=filed_size, dim=1)
bilinear_out = self._BilinearInteraction(
concat_emb, filed_size, self.sparse_feature_dim, bilinear_type)
dnn_input = fluid.layers.concat(
input=[senet_bilinear_out, bilinear_out, self.dense_input], axis=1)
dnn_output = self._DNNLayer(dnn_input, dropout_rate)
y_pred = fluid.layers.fc(
input=dnn_output,
size=1,
param_attr=fluid.initializer.Xavier(uniform=False),
act='sigmoid',
name='logit')
self.predict = y_pred
auc, batch_auc, _ = fluid.layers.auc(input=self.predict,
label=self.label_input,
num_thresholds=2**12,
slide_steps=20)
if is_infer:
self._infer_results["AUC"] = auc
self._infer_results["BATCH_AUC"] = batch_auc
return
self._metrics["AUC"] = auc
self._metrics["BATCH_AUC"] = batch_auc
cost = fluid.layers.log_loss(
input=self.predict,
label=fluid.layers.cast(
x=self.label_input, dtype='float32'))
avg_cost = fluid.layers.reduce_mean(cost)
self._cost = avg_cost
...@@ -37,35 +37,43 @@ ...@@ -37,35 +37,43 @@
| xDeepFM | xDeepFM | [xDeepFM: Combining Explicit and Implicit Feature Interactions for Recommender Systems](https://dl.acm.org/doi/pdf/10.1145/3219819.3220023)(2018) | | xDeepFM | xDeepFM | [xDeepFM: Combining Explicit and Implicit Feature Interactions for Recommender Systems](https://dl.acm.org/doi/pdf/10.1145/3219819.3220023)(2018) |
| DIN | Deep Interest Network | [Deep Interest Network for Click-Through Rate Prediction](https://dl.acm.org/doi/pdf/10.1145/3219819.3219823)(2018) | | DIN | Deep Interest Network | [Deep Interest Network for Click-Through Rate Prediction](https://dl.acm.org/doi/pdf/10.1145/3219819.3219823)(2018) |
| FGCNN | Feature Generation by CNN | [Feature Generation by Convolutional Neural Network for Click-Through Rate Prediction](https://arxiv.org/pdf/1904.04447.pdf)(2019) | | FGCNN | Feature Generation by CNN | [Feature Generation by Convolutional Neural Network for Click-Through Rate Prediction](https://arxiv.org/pdf/1904.04447.pdf)(2019) |
| FIBINET | Combining Feature Importance and Bilinear feature Interaction | [《FiBiNET: Combining Feature Importance and Bilinear feature Interaction for Click-Through Rate Prediction》]( https://arxiv.org/pdf/1905.09433.pdf)(2019) |
下面是每个模型的简介(注:图片引用自链接中的论文) 下面是每个模型的简介(注:图片引用自链接中的论文)
[wide&deep](https://dl.acm.org/doi/pdf/10.1145/2988450.2988454): [wide&deep](https://dl.acm.org/doi/pdf/10.1145/2988450.2988454):
<p align="center"> <p align="center">
<img align="center" src="../../doc/imgs/wide&deep.png"> <img align="center" src="../../doc/imgs/wide&deep.png">
<p> <p>
[DeepFM](https://arxiv.org/pdf/1703.04247.pdf): [DeepFM](https://arxiv.org/pdf/1703.04247.pdf):
<p align="center"> <p align="center">
<img align="center" src="../../doc/imgs/deepfm.png"> <img align="center" src="../../doc/imgs/deepfm.png">
<p> <p>
[XDeepFM](https://dl.acm.org/doi/pdf/10.1145/3219819.3220023): [XDeepFM](https://dl.acm.org/doi/pdf/10.1145/3219819.3220023):
<p align="center"> <p align="center">
<img align="center" src="../../doc/imgs/xdeepfm.png"> <img align="center" src="../../doc/imgs/xdeepfm.png">
<p> <p>
[DCN](https://dl.acm.org/doi/pdf/10.1145/3124749.3124754): [DCN](https://dl.acm.org/doi/pdf/10.1145/3124749.3124754):
<p align="center"> <p align="center">
<img align="center" src="../../doc/imgs/dcn.png"> <img align="center" src="../../doc/imgs/dcn.png">
<p> <p>
[DIN](https://dl.acm.org/doi/pdf/10.1145/3219819.3219823): [DIN](https://dl.acm.org/doi/pdf/10.1145/3219819.3219823):
<p align="center"> <p align="center">
<img align="center" src="../../doc/imgs/din.png"> <img align="center" src="../../doc/imgs/din.png">
<p> <p>
[FIBINET](https://arxiv.org/pdf/1905.09433.pdf):
<p align="center">
<img align="center" src="../../doc/imgs/fibinet.png">
<p>
## 使用教程(快速开始) ## 使用教程(快速开始)
使用样例数据快速开始,参考[训练](###训练) & [预测](###预测) 使用样例数据快速开始,参考[训练](###训练) & [预测](###预测)
## 使用教程(复现论文) ## 使用教程(复现论文)
为了方便使用者能够快速的跑通每一个模型,我们在每个模型下都提供了样例数据,并且调整了batch_size等超参以便在样例数据上更加友好的显示训练&测试日志。如果需要复现readme中的效果请按照如下表格调整batch_size等超参,并使用提供的脚本下载对应数据集以及数据预处理。 为了方便使用者能够快速的跑通每一个模型,我们在每个模型下都提供了样例数据,并且调整了batch_size等超参以便在样例数据上更加友好的显示训练&测试日志。如果需要复现readme中的效果请按照如下表格调整batch_size等超参,并使用提供的脚本下载对应数据集以及数据预处理。
...@@ -77,6 +85,7 @@ ...@@ -77,6 +85,7 @@
| DIN | 32 | 10 | 100 | | DIN | 32 | 10 | 100 |
| Wide&Deep | 40 | 1 | 40 | | Wide&Deep | 40 | 1 | 40 |
| xDeepFM | 100 | 1 | 10 | | xDeepFM | 100 | 1 | 10 |
| Fibinet | 1000 | 8 | 4 |
### 数据处理 ### 数据处理
参考每个模型目录数据下载&预处理脚本 参考每个模型目录数据下载&预处理脚本
...@@ -116,6 +125,7 @@ python -m paddlerec.run -m ./config.yaml # 以DNN为例 ...@@ -116,6 +125,7 @@ python -m paddlerec.run -m ./config.yaml # 以DNN为例
| Criteo | xDeepFM | 0.48657 | -- | -- | -- | | Criteo | xDeepFM | 0.48657 | -- | -- | -- |
| Census-income Data | Wide&Deep | 0.76195 | 0.90577 | -- | -- | | Census-income Data | Wide&Deep | 0.76195 | 0.90577 | -- | -- |
| Amazon Product | DIN | 0.47005 | 0.86379 | -- | -- | | Amazon Product | DIN | 0.47005 | 0.86379 | -- | -- |
| Criteo | Fibinet | -- | 0.86662 | -- | -- |
## 分布式 ## 分布式
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册