SkillAgentSearch skills...

EasyDeepRecommand

A user-friendly open-source project for recommendation systems.

Install / Use

/learn @Iamctb/EasyDeepRecommand
About this skill

Quality Score

0/100

Supported Platforms

Universal

README

<div align="center"> <img src="https://github.com/Iamctb/EasyDeepRecommand/blob/main/ResyDeepRecommand_logo.png" alt="Logo" width="360"/> </div> <div align="center"> <a href="https://www.python.org/"><img src="https://img.shields.io/badge/python-3.8%2B-blue" style="max-width: 100%;" alt="python version"></a > <a href="https://pytorch.org/"><img src="https://img.shields.io/badge/pytorch-1.13%2B-blue" style="max-width: 100%;" alt="pytorch version"></a > <a href="https://www.tensorflow.org/?hl=zh-cn"><img src="https://img.shields.io/badge/tensorflow-2.1+-blue" style="max-width: 100%;" alt="tensorflow version"></a > <a href="https://blog.csdn.net/qq_41915623/article/details/138839827?fromshare=blogdetail&sharetype=blogdetail&sharerId=138839827&sharerefer=PC&sharesource=qq_41915623&sharefrom=from_link"><img src="https://img.shields.io/badge/CSDN-Blog-red" style="max-width: 100%;" alt="Downloads"></a > <a href="https://juejin.cn/post/7424903278063140898"><img src="https://img.shields.io/badge/JueJin-Blog-red" style="max-width: 100%;" alt="Downloads"></a > <a href="https://www.apache.org/licenses/LICENSE-2.0.html"><img src="https://img.shields.io/badge/lisence-Apache_2.0-green" style="max-width: 100%;" alt="License"></a > </div> <hr/> <div align="center"> <a href="https://github.com/Iamctb/EasyDeepRecommand/stargazers"><img src="http://bytecrank.com/nastyox/reporoster/php/stargazersSVG.php?user=Iamctb&repo=EasyDeepRecommand" width="600"/><a/> </div>

一个通俗易懂的开源推荐系统(A user-friendly open-source project for recommendation systems).

本项目将结合:代码、数据流转图、博客、模型发展史 等多个方面通俗易懂地讲解经典推荐模型,让读者通过一个项目了解推荐系统概况! 非常适合"新手入门"、"阅读推荐相关经典论文"

持续更新中..., 欢迎star🌟, 第一时间获取更新!!!

Features

1️⃣ 分类解析推荐模型:特征交叉模型、多任务模型、行为序列模型等

2️⃣ 通过blog详细解释模型/论文

3️⃣ 提供模型发展图:介绍模型的优缺点,解决了什么问题,以及前后因果关系

4️⃣ 提供详细的代码注释,并包含详细的数据处理模块

Development of model

graph LR
    A[推荐模型] --> B[特征交叉模型]
    A --> C[多任务模型]

    B --> B1[Wide and Deep DLRS16 谷歌]
    B1 --> B1a[优点: Wide增强记忆,Deep增强泛化; 结构简单高效]
    B1 --> B1b[缺点: 需手工特征交互; 维度上升模型变复杂]

    B --> B2[DCN V1 AKDD17 谷歌]
    B2 --> B2a[优点: 适合小规模; 逐层特征交叉]
    B2 --> B2b[缺点: 大规模表达能力有限]

    B --> B3[DCN V2 WWW21 谷歌]

    C --> C1[Shared Bottom 结构]
    C1 --> C1a[优点: 共享参数降低复杂度; 多任务互相促进]
    C1 --> C1b[缺点: 负迁移; 跨域表现差]

    C --> C2[MMoE]
    C --> C3[PLE]

和很多朋友交流发现,我们在读很多论文时,都聚焦于论文中提出的模型本身,而没有关心模型间的因果关系,所以这个 板块用来介绍模型优缺点和模型间的前后因果关系。

由于很多论文中都没有显式介绍自己模型的优缺点和前因后果,所以很多观点都是本人结合网上资料加上个人理解作出的,如果有不对的地方,欢迎在issue中交流讨论。

Dataset

| Name | Preprocess_url | Download | Progress | | ------ | ------------------------------------------------------------ | ------------------------------------------------------------ | -------- | | Criteo | criteo_preprocess.py: 预处理源代码 | Download_URL | ✅ | | | 预处理说明 | | |

Model_Zoo

| No. | Publication | Model | Blog | Paper | Version | | ---- | ----------- | -------- | ------------------------------------------------------------ | ------------------------------------------------------------ | -------------------------------------------------------- | | 1 | DLRS'16 | WideDeep | 📝 WideDeep | Wide & Deep Learning for Recommender Systems, Google | ✅torch | | 2 | ADKDD'17 | DCN | 📝 DCN | Deep & Cross Network for Ad Click Predictions, Google | ✅torch | | 3 | WWW'21 | DCV-v2 | 📝 DCN-v2 | DCN V2: Improved Deep & Cross Network and Practical Lessons for Web-scale Learning to Rank Systems, Google | ✅torch | | 4 | NeurIPS'23 | TIGER | 📝 Tiger | Recommender Systems with Generative Retrieval, Google | Unofficial Code |

Dependencies

本项目环境主要有:

  • python=3.8.20
  • pytorch=1.13.0

其余安装包可以使用下面命令安装:

pip install -r requirements.txt

Quick Start

以Criteo数据集和WideDeep举例:

Step1: 数据预处理

cd DataProcess/criteo
python criteo_preprocess.py

样本数据是使用的Criteo一万条数据作为示例,在执行命令过程中,需要注意 数据集的路径

Step2: 训练模型

data_config.json 中配置数据集路径;

model_config.json 中配置模型信息;

然后运行下面命令即可:

cd ModelZoo/WideDeep/WideDeep_torch
python train.py

最后

开源项目的一个很大特点就是:共创!

欢迎各位在issue中交流讨论。

如果你觉得还不错的话,请帮忙点个star🌟吧,非常感谢!!! If you think it's good, please help out with a star🌟, thank you !!!

View on GitHub
GitHub Stars52
CategoryDevelopment
Updated19d ago
Forks6

Languages

Python

Security Score

100/100

Audited on Mar 9, 2026

No findings