Home
Softono
torch-yun

torch-yun

Open source Apache-2.0 Java
26
Stars
12
Forks
17
Issues
1
Watchers
1 week
Last Commit

About torch-yun

至数云 (torch-yun) is an ultra-lightweight, enterprise-grade AI application platform designed for one-click deployment and immediate use. It enables rapid AI application building, large model deployment, model training, API exposure, and visual workflow orchestration. The platform is compatible with Tongyi Qianwen and DeepSeek models, helping enterprises adopt AI efficiently. Key features include an AI chat homepage with history and application listings, resource management with compute clusters, a model hub offering a model marketplace, model repository, intelligent agents, and model applications, plus an admin backend with user management, tenant lists, tenant members, and certificate installation. The platform supports technologies such as large model training, model orchestration, PyTorch, Docker, and HuggingFace. It can be quickly deployed via Docker and built from source on Mac, Linux, or Windows systems, with an installation package ready for production use.

Platforms

Web Self-hosted Docker

Languages

Java

至数云-超轻量级人工智能应用平台

注意:大版本升级维护中,切到稳定版本commit

git clone https://github.com/isxcode/torch-yun.git
cd spark-yun
git checkout b4a0b0e5f0334b17ef3b5f7f23eaafff187f771f

Docker Pulls build GitHub Repo stars GitHub forks FOSSA Status GitHub License

产品官网 https://zhishuyun.isxcode.com
源码仓库 https://github.com/isxcode/torch-yun
码云仓库 https://gitee.com/isxcode/torch-yun
演示环境 https://zhishuyun-demo.isxcode.com
部署文档 https://zhishuyun.isxcode.com/zh/docs/zh/1/2
安装包下载 https://zhishuyun-demo.isxcode.com/tools/open/file/zhishuyun.tar.gz
产品推荐 [至轻云] - 超轻量级智能化大数据中心
关键词 大模型训练, 模型中心, 模型调用, 模型编排, 人工智能, Pytorch, Docker, HuggingFace

产品介绍

        至数云是一款超轻量、企业级人工智能应用平台。一键部署,开箱即用。可快速实现AI应用构建、大模型部署、模型训练、接口开放、可视化流程等场景。兼容通义千问和DeepSeek模型,助力企业拥抱人工智能新时代。

功能列表

模块 功能(包含商业版)
首页 AI对话、历史记录、应用列表
资源管理 计算集群、资源中心
模型中心 模型广场、模型仓库、智能体、模型应用
后台管理 用户中心、租户列表、租户成员、证书安装

立即体验

演示地址:https://zhishuyun-demo.isxcode.com
体验账号:zhiyao
账号密码:zhiyao123 [已禁用,联系管理员]
添加微信:wx_ispong

相关文档

快速部署

# 访问地址:http://localhost:8080
# 管理员账号:admin 
# 管理员密码:admin123
docker run -p 8080:8080 -d isxcode/zhishuyun

源码构建

# 系统环境: Mac或Linux
# 安装包路径: /tmp/torch-yun/torch-yun-dist/build/distributions/zhishuyun.tar.gz
cd /tmp
git clone https://github.com/isxcode/torch-yun.git
docker run --rm \
  -v /tmp/torch-yun:/torch-yun \
  -it isxcode/zhishuyun-build \
  gradle package
# 系统环境: Windows
# 安装包路径: C:\Users\isxcode\Downloads\torch-yun\torch-yun-dist\build\distributions\zhishuyun.tar.gz
cd Downloads
git clone https://github.com/isxcode/torch-yun.git
docker run --rm ^
  -v C:\Users\isxcode\Downloads\torch-yun:/torch-yun ^
  -it isxcode/zhishuyun-build ^
  gradle package

产品展示

登录页 首页
数据源 作业流
pySpark 运行结果
作业配置 任务调度
数据分层 数据地图