open-webui

Open WebUI 是一个可扩展、功能丰富且用户友好的自托管 WebUI,旨在完全离线运行。它支持各种 LLM 运行程序,包括 Ollama 和 OpenAI 兼容的 API。有关更多信息,请务必查看我们的 Open WebUI 文档。

安装

官方安装地址:https://docs.openwebui.com/getting-started/ 有两种安装方式,docker和源码。 熟悉docker的同学可以采用该方式。 本文以源码的形式进行安装。

环境要求

熟练使用python

🐰 Node.js >= 20.10

🐍 Python >= 3.11

步骤

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
克隆代码
git clone https://github.com/open-webui/open-webui.git

cd open-webui/

拷贝.env文件,可鼠标右键复制粘贴
# Copying required .env file
cp -RPp .env.example .env

构建前端页面
# Building Frontend Using Node
npm i
npm run build

构建后端
# Serving Frontend with the Backend
cd ./backend
pip install -r requirements.txt -U

启动服务
linux or mac:bash start.sh
windows:start_windows.bat

image

访问

打开地址:127.0.0.1:8000

第一次打开的时候,需要注册一个账号。(如admin/admin) 册的第一个用户默认为管理员账号,以后注册的账号角色均为pending(待分配权限)。

选择模型

image

开始对话

image

结束

收工