alibaba/TorchEasyRec

An easy-to-use framework for large scale recommendation algorithms.

What it solves

TorchEasyRec provides a production-ready framework for building deep learning recommendation systems. It simplifies the development of complex models for candidate generation (matching), scoring (ranking), multi-task learning, and generative recommendation, reducing the friction between research and production deployment.

How it works

Built on PyTorch and TorchRec, the framework allows users to define models through simple configurations. It handles the large-scale data processing and distributed training needs of recommendation systems, including hybrid data/model parallelism and sharding for large embeddings. It integrates natively with Alibaba Cloud's data warehouses (MaxCompute) and serving infrastructure (EAS), while also supporting standard formats like Parquet and CSV.

Who it’s for

ML engineers and data scientists specializing in recommendation systems who need to deploy high-performance, scalable models into production environments.

Highlights

  • Extensive Model Library: Includes over 20 battle-tested algorithms such as DSSM, DeepFM, DIN, MMoE, and DLRM-HSTU.
  • Scalable Infrastructure: Supports row-wise, column-wise, and table-wise sharding for large embeddings and zero-collision hashing with eviction policies.
  • Production-Ready: Offers consistent feature generation between training and serving, and acceleration via TensorRT and AOTInductor.
  • Flexible Data Integration: Native support for MaxCompute, Parquet, CSV, and Kafka streaming data.

Related

  • Project
    meta-pytorch/torchrec

    A PyTorch domain library providing sparsity and parallelism primitives for training and deploying large-scale recommender systems with massive embedding tables sharded across multiple GPUs.

  • Project
    datawhalechina/torch-rechub

    Torch-RecHub is a PyTorch framework for building recommender systems with 30+ models, modular design, ONNX export, and support for multiple hardware backends.

  • Project
    PersiaML/PERSIA

    A PyTorch-based parallel training system designed to scale deep learning recommendation models up to 100 trillion parameters on commodity hardware.

  • Project
    pytorch/pytorch

    A Python-based machine learning framework that provides GPU-accelerated tensor computation and a dynamic autograd system for building deep neural networks.

  • Project
    alibaba/RecIS

    A unified deep learning framework for ultra-large-scale recommendation systems that optimizes sparse and dense computing within the PyTorch ecosystem.