mltrees / paddlerec

大规模推荐模型训练工具

(中文文档|简体中文|English)

什么是推荐系统?

• 推荐系统是在互联网信息爆炸式增长的时代背景下，帮助用户高效获得感兴趣信息的关键；

• 推荐系统也是帮助产品最大限度吸引用户、留存用户、增加用户粘性、提高用户转化率的银弹。

• 有无数优秀的产品依靠用户可感知的推荐系统建立了良好的口碑，也有无数的公司依靠直击用户痛点的推荐系统在行业中占领了一席之地。

  可以说，谁能掌握和利用好推荐系统，谁就能在信息分发的激烈竞争中抢得先机。 但与此同时，有着许多问题困扰着推荐系统的开发者，比如：庞大的数据量，复杂的模型结构，低效的分布式训练环境，波动的在离线一致性，苛刻的上线部署要求，以上种种，不胜枚举。

什么是PaddleRec?

• 源于飞桨生态的搜索推荐模型 一站式开箱即用工具

• 适合初学者，开发者，研究者的推荐系统全流程解决方案

• 包含内容理解、匹配、召回、排序、 多任务、重排序等多个任务的完整推荐搜索算法库

  方向	模型	单机CPU	单机GPU	分布式CPU	分布式GPU	论文
  内容理解	Text-Classifcation	✓	✓	✓	x	[EMNLP 2014]Convolutional neural networks for sentence classication
  内容理解	TagSpace	✓	✓	✓	x	[EMNLP 2014]TagSpace: Semantic Embeddings from Hashtags
  匹配	DSSM	✓	✓	✓	x	[CIKM 2013]Learning Deep Structured Semantic Models for Web Search using Clickthrough Data
  匹配	MultiView-Simnet	✓	✓	✓	x	[WWW 2015]A Multi-View Deep Learning Approach for Cross Domain User Modeling in Recommendation Systems
  召回	TDM	✓	>=1.8.0	✓	>=1.8.0	[KDD 2018]Learning Tree-based Deep Model for Recommender Systems
  召回	fasttext	✓	✓	x	x	[EACL 2017]Bag of Tricks for Efficient Text Classification
  召回	Word2Vec	✓	✓	✓	x	[NIPS 2013]Distributed Representations of Words and Phrases and their Compositionality
  召回	SSR	✓	✓	✓	✓	[SIGIR 2016]Multi-Rate Deep Learning for Temporal Recommendation
  召回	Gru4Rec	✓	✓	✓	✓	[2015]Session-based Recommendations with Recurrent Neural Networks
  召回	Youtube_dnn	✓	✓	✓	✓	[RecSys 2016]Deep Neural Networks for YouTube Recommendations
  召回	NCF	✓	✓	✓	✓	[WWW 2017]Neural Collaborative Filtering
  召回	GNN	✓	✓	✓	✓	[AAAI 2019]Session-based Recommendation with Graph Neural Networks
  召回	RALM	✓	✓	✓	✓	[KDD 2019]Real-time Attention Based Look-alike Model for Recommender System
  排序	Logistic Regression	✓	x	✓	x	/
  排序	Dnn	✓	✓	✓	✓	/
  排序	FM	✓	x	✓	x	[IEEE Data Mining 2010]Factorization machines
  排序	FFM	✓	x	✓	x	[RECSYS 2016]Field-aware Factorization Machines for CTR Prediction
  排序	FNN	✓	x	✓	x	[ECIR 2016]Deep Learning over Multi-field Categorical Data
  排序	Deep Crossing	✓	x	✓	x	[ACM 2016]Deep Crossing: Web-Scale Modeling without Manually Crafted Combinatorial Features
  排序	Pnn	✓	x	✓	x	[ICDM 2016]Product-based Neural Networks for User Response Prediction
  排序	DCN	✓	x	✓	x	[KDD 2017]Deep & Cross Network for Ad Click Predictions
  排序	NFM	✓	x	✓	x	[SIGIR 2017]Neural Factorization Machines for Sparse Predictive Analytics
  排序	AFM	✓	x	✓	x	[IJCAI 2017]Attentional Factorization Machines: Learning the Weight of Feature Interactions via Attention Networks
  排序	DeepFM	✓	x	✓	x	[IJCAI 2017]DeepFM: A Factorization-Machine based Neural Network for CTR Prediction
  排序	xDeepFM	✓	x	✓	x	[KDD 2018]xDeepFM: Combining Explicit and Implicit Feature Interactions for Recommender Systems
  排序	DIN	✓	x	✓	x	[KDD 2018]Deep Interest Network for Click-Through Rate Prediction
  排序	DIEN	✓	x	✓	x	[AAAI 2019]Deep Interest Evolution Network for Click-Through Rate Prediction
  排序	BST	✓	x	✓	x	[DLP_KDD 2019]Behavior Sequence Transformer for E-commerce Recommendation in Alibaba
  排序	AutoInt	✓	x	✓	x	[CIKM 2019]AutoInt: Automatic Feature Interaction Learning via Self-Attentive Neural Networks
  排序	Wide&Deep	✓	x	✓	x	[DLRS 2016]Wide & Deep Learning for Recommender Systems
  排序	FGCNN	✓	✓	✓	✓	[WWW 2019]Feature Generation by Convolutional Neural Network for Click-Through Rate Prediction
  排序	Fibinet	✓	✓	✓	✓	[RecSys19]FiBiNET: Combining Feature Importance and Bilinear feature Interaction for Click-Through Rate Prediction
  排序	Flen	✓	✓	✓	✓	[2019]FLEN: Leveraging Field for Scalable CTR Prediction
  多任务	PLE	✓	✓	✓	✓	[RecSys 2020]Progressive Layered Extraction (PLE): A Novel Multi-Task Learning (MTL) Model for Personalized Recommendations
  多任务	ESMM	✓	✓	✓	✓	[SIGIR 2018]Entire Space Multi-Task Model: An Effective Approach for Estimating Post-Click Conversion Rate
  多任务	MMOE	✓	✓	✓	✓	[KDD 2018]Modeling Task Relationships in Multi-task Learning with Multi-gate Mixture-of-Experts
  多任务	ShareBottom	✓	✓	✓	✓	[1998]Multitask learning
  重排序	Listwise	✓	✓	✓	x	[2019]Sequential Evaluation and Generation Framework for Combinatorial Recommender System

快速使用

环境要求

• Python 2.7/ 3.5 / 3.6 / 3.7

• PaddlePaddle >=2.0

• 操作系统: Windows/Mac/Linux

  Windows下PaddleRec目前仅支持单机训练，分布式训练建议使用Linux环境

安装Paddle

• pip安装cpu

  python -m pip install paddlepaddle==2.0.0rc1 -i https://mirror.baidu.com/pypi/simple 

• pip安装gpu

  python -m pip install paddlepaddle-gpu==2.0.0rc1 -i https://mirror.baidu.com/pypi/simple 

更多版本下载可参考paddle官网下载安装

下载PaddleRec

• github 下载

  git clone https://github.com/PaddlePaddle/PaddleRec/
  cd PaddleRec

快速运行

我们以排序模型中的dnn模型为例介绍PaddleRec的一键启动。训练数据来源为Criteo数据集，我们从中截取了100条数据：

cd models/rank/dnn
python -u train.py -m config.yaml 

帮助文档

项目背景

• 推荐系统介绍

• 分布式深度学习介绍

• Benchmark

FAQ

• 常见问题FAQ

社区

版本历史

• 2020.10.12 - PaddleRec v1.8.5
• 2020.06.17 - PaddleRec v0.1.0
• 2020.06.03 - PaddleRec v0.0.2
• 2020.05.14 - PaddleRec v0.0.1

许可证书

本项目的发布受Apache 2.0 license许可认证。

联系我们

如有意见、建议及使用中的BUG，欢迎在GitHub Issue提交

亦可通过以下方式与我们沟通交流：

• QQ群号码：861717190
• 微信小助手微信号：paddlerec2020

     

PaddleRec交流QQ群               PaddleRec微信小助手