[![GitHub Repo stars](https://img.shields.io/github/stars/intelligent-machine-learning/dlrover?style=social)](https://github.com/intelligent-machine-learning/dlrover/stargazers)
[![Build](https://github.com/intelligent-machine-learning/dlrover/actions/workflows/main.yml/badge.svg)](https://github.com/intelligent-machine-learning/dlrover/actions/workflows/main.yml)
[![CodeFactor](https://www.codefactor.io/repository/github/intelligent-machine-learning/dlrover/badge)](https://www.codefactor.io/repository/github/intelligent-machine-learning/dlrover)
| [English](README.md) | [中文](docs/README-zh-Hans.md) |
TFPlus is a high-performance TensorFlow extension library developed in-house by Ant Group. It supports core businesses such as search and promotion within the Ant Group, supporting ultra-large scale sparse training. TFPlus has accumulated essential functionalities and performance optimizations for core sparse scenarios. It has deeply optimized the performance in terms of IO, operators, graph optimization, distribution, and collective communication for sparse models. Also, it provides special optimizers, fault tolerance, elasticity, incremental updates, etc., unique to sparse scenarios.