geogubd / EasyRec

A framework for large scale recommendation algorithms.

EasyRec简介

What is EasyRec?

EasyRec is an easy to use framework for Recommendation

EasyRec致力于成为容易上手的工业界深度学习推荐算法框架，支持大规模训练、评估、导出和部署。EasyRec实现了业界领先的模型，包含排序、召回、多目标等模型，支持超参搜索，显著降低了建模的复杂度和工作量。

Why EasyRec?

Run everywhere

• Local / MaxCompute / DataScience / DLC
• TF1.12-1.15 / TF2.x / PAI-TF

Diversified input data

• MaxCompute Table
• HDFS files
• OSS files
• Kafka Streams
• Local CSV

Simple to config

• Flexible feature config and simple model config
• Efficient and robust feature generation[used in taobao]
• Nice web interface in development

It is smart

• EarlyStop / Best Checkpoint Saver
• Hyper Parameter Search / AutoFeatureCross
• In development: NAS, Knowledge Distillation, MultiModal

Large scale and easy deployment

• Support large scale embedding, incremental saving
• Many parallel strategies: ParameterServer, Mirrored, MultiWorker
• Easy deployment to EAS: automatic scaling, easy monitoring
• Consistency guarantee: train and serving

A variety of models

• DeepFM / MultiTower / Deep Interest Network / DSSM / MMoE / ESMM
• More models in development

Easy to customize

• Easy to implement customized models
• Not need to care about data pipelines

Get Started

• Download

    git clone https://github.com/AlibabaPAI/EasyRec.git
    wget https://easyrec.oss-cn-beijing.aliyuncs.com/data/easyrec_data_20210818.tar.gz
    sh scripts/gen_proto.sh

• EasyRec Framework

• Run

• PAI-DSW DEMO (Remember to select Python 3 kernel)

• Develop

• Doc