云服务器 Flexus X 实例：多智能体对话框架 AutoGen 部署和实例运行

在体验了华为云的 Flexus 云服务器 X 实例 后，感受到其卓越的性能和灵活性，通过弹性伸缩服务和负载均衡，能够轻松应对市场的瞬息万变，确保业务的连续性和稳定性。数据的可靠性和安全性也得到了极大的保障，华为云数据库服务和网络安全服务的结合，让数据存储和业务运行更加安全无忧。

对于个人开发者和测试人员来说，Flexus 云服务器 X 实例 的按需计费模式极大地降低了成本，同时提供了高可用性和可靠性，体验上可以媲美自建环境。全栈的开发测试工具链更是提升了开发效率，让创新和迭代变得更加迅速，点击这里体验~

下面就来一起体验下吧！

一、什么是多智能体？****

多智能体系统是由多个智能体精妙组合而成的集合体，其核心宗旨在于将庞大且复杂的系统解构为一系列小型、高度互联、且能自主通信与协同工作的子系统，从而显著提升系统的可管理性与灵活性。这一领域的研究深入探索了智能体的认知知识、追求目标、掌握技能、规划路径，以及如何通过高效的协调机制促使它们共同行动以解决复杂问题。

二、什么是 AutoGen？****

AutoGen 是一个开源编程框架，用于构建 AI 代理并促进多个代理之间的协作以解决任务。AutoGen 旨在简化代理 AI 的开发和研究，就像 PyTorch 对深度学习所做的那样。它提供的功能包括能够相互交互的代理、促进使用各种大型语言模型 (LLM) 和工具使用支持、自主和人机交互工作流程以及多代理对话模式。

三、部署 AutoGen****

3.1 更新 apt 软件源****

首先，更新下云服务器的 apt 软件源，执行如下命令更新。

root@flexusx-7305:~# sudo apt update

Hit:1 http://repo.huaweicloud.com/ubuntu focal InRelease

Get:2 http://repo.huaweicloud.com/ubuntu focal-updates InRelease [128 kB]

Hit:3 http://repo.huaweicloud.com/ubuntu focal-backports InRelease

Hit:4 http://repo.huaweicloud.com/ubuntu focal-security InRelease

Ign:5 https://download.docker.com/linux/ubuntu focal InRelease

Get:6 http://repo.huaweicloud.com/ubuntu focal-updates/main amd64 Packages [3,563 kB]

Err:7 https://download.docker.com/linux/ubuntu focal Release

Could not handshake: Error in the pull function. [IP: 18.155.68.38 443]

Get:8 http://repo.huaweicloud.com/ubuntu focal-updates/universe i386 Packages [807 kB]

Get:9 http://repo.huaweicloud.com/ubuntu focal-updates/universe amd64 Packages [1,231 kB]

Hit:10 https://packages.redis.io/deb focal InRelease

Reading package lists... Done

执行如下命令安装software-properties-common。

root@flexusx-7305:~# sudo apt install software-properties-common

3.2 安装 python 3.10****

添加 ppa:deadsnakes/ppa，如下所示。

root@flexusx-7305:~# sudo add-apt-repository ppa:deadsnakes/ppa

AutoGen 需要 python 3.9 以上的版本，本文安装 python 3.10，如下所示。

root@flexusx-7305:~# sudo apt install python3.10

Reading package lists... Done

Building dependency tree

Reading state information... Done

The following packages were automatically installed and are no longer required:

redis-server redis-tools

Use 'sudo apt autoremove' to remove them.

The following additional packages will be installed:

libpython3.10-minimal libpython3.10-stdlib python3.10-minimal

Suggested packages:

python3.10-venv binfmt-support

The following NEW packages will be installed:

libpython3.10-minimal libpython3.10-stdlib python3.10 python3.10-minimal

0 upgraded, 4 newly installed, 0 to remove and 33 not upgraded.

Need to get 5,221 kB of archives.

After this operation, 20.2 MB of additional disk space will be used.

Do you want to continue? [Y/n] y

Get:1 http://ppa.launchpad.net/deadsnakes/ppa/ubuntu focal/main amd64 libpython3.10-minimal amd64 3.10.15-1+focal1 [825 kB]

Get:2 http://ppa.launchpad.net/deadsnakes/ppa/ubuntu focal/main amd64 python3.10-minimal amd64 3.10.15-1+focal1 [2,077 kB]

Get:3 http://ppa.launchpad.net/deadsnakes/ppa/ubuntu focal/main amd64 libpython3.10-stdlib amd64 3.10.15-1+focal1 [1,763 kB]

Get:4 http://ppa.launchpad.net/deadsnakes/ppa/ubuntu focal/main amd64 python3.10 amd64 3.10.15-1+focal1 [556 kB]

Fetched 5,221 kB in 3s (1,707 kB/s)

Selecting previously unselected package libpython3.10-minimal:amd64.

(Reading database ... 117364 files and directories currently installed.)

Preparing to unpack .../libpython3.10-minimal_3.10.15-1+focal1_amd64.deb ...

Unpacking libpython3.10-minimal:amd64 (3.10.15-1+focal1) ...

Selecting previously unselected package python3.10-minimal.

Preparing to unpack .../python3.10-minimal_3.10.15-1+focal1_amd64.deb ...

Unpacking python3.10-minimal (3.10.15-1+focal1) ...

Selecting previously unselected package libpython3.10-stdlib:amd64.

Preparing to unpack .../libpython3.10-stdlib_3.10.15-1+focal1_amd64.deb ...

Unpacking libpython3.10-stdlib:amd64 (3.10.15-1+focal1) ...

Selecting previously unselected package python3.10.

Preparing to unpack .../python3.10_3.10.15-1+focal1_amd64.deb ...

Unpacking python3.10 (3.10.15-1+focal1) ...

Setting up libpython3.10-minimal:amd64 (3.10.15-1+focal1) ...

Setting up python3.10-minimal (3.10.15-1+focal1) ...

Setting up libpython3.10-stdlib:amd64 (3.10.15-1+focal1) ...

Setting up python3.10 (3.10.15-1+focal1) ...

Processing triggers for mime-support (3.64ubuntu1) ...

Processing triggers for man-db (2.9.1-1) ...

root@flexusx-7305:~#

创建 python 软连接，让 /usr/bin/python 指向 /usr/bin/python3.10，如果 python3 命令不是 3.10 版本，同样也需要创建软连接，然后查看 python 版本，如下所示。

root@flexusx-7305:~# ln -s /usr/bin/python3.10 /usr/bin/python

root@flexusx-7305:~# python --version

Python 3.10.15

root@flexusx-7305:~#

安装 python3.10-venv 用于创建 python 虚拟环境，如下所示。

root@flexusx-7305:~# apt install python3.10-venv

Reading package lists... Done

Building dependency tree

Reading state information... Done

The following packages were automatically installed and are no longer required:

redis-server redis-tools

Use 'apt autoremove' to remove them.

The following additional packages will be installed:

python3.10-distutils python3.10-lib2to3

The following NEW packages will be installed:

python3.10-distutils python3.10-lib2to3 python3.10-venv

0 upgraded, 3 newly installed, 0 to remove and 33 not upgraded.

Need to get 3,243 kB of archives.

After this operation, 4,528 kB of additional disk space will be used.

Do you want to continue? [Y/n] y

Get:1 http://ppa.launchpad.net/deadsnakes/ppa/ubuntu focal/main amd64 python3.10-lib2to3 all 3.10.15-1+focal1 [126 kB]

Get:2 http://ppa.launchpad.net/deadsnakes/ppa/ubuntu focal/main amd64 python3.10-distutils all 3.10.15-1+focal1 [187 kB]

Get:3 http://ppa.launchpad.net/deadsnakes/ppa/ubuntu focal/main amd64 python3.10-venv amd64 3.10.15-1+focal1 [2,931 kB]

Fetched 3,243 kB in 4s (880 kB/s)

Selecting previously unselected package python3.10-lib2to3.

(Reading database ... 118026 files and directories currently installed.)

Preparing to unpack .../python3.10-lib2to3_3.10.15-1+focal1_all.deb ...

Unpacking python3.10-lib2to3 (3.10.15-1+focal1) ...

Selecting previously unselected package python3.10-distutils.

Preparing to unpack .../python3.10-distutils_3.10.15-1+focal1_all.deb ...

Unpacking python3.10-distutils (3.10.15-1+focal1) ...

Selecting previously unselected package python3.10-venv.

Preparing to unpack .../python3.10-venv_3.10.15-1+focal1_amd64.deb ...

Unpacking python3.10-venv (3.10.15-1+focal1) ...

Setting up python3.10-lib2to3 (3.10.15-1+focal1) ...

Setting up python3.10-distutils (3.10.15-1+focal1) ...

Setting up python3.10-venv (3.10.15-1+focal1) ...

创建用于部署 AutoGen 的虚拟环境，执行如下命令。

root@flexusx-7305:~# python3 -m venv autogentest

root@flexusx-7305:~# source autogentest/bin/activate

(autogentest) root@flexusx-7305:~#

3.3 安装 AutoGen****

进入环境 AutoGen，执行命令安装 AutoGen，如下所示。

(autogentest) root@flexusx-7305:~# pip install autogen

Collecting autogen

Downloading autogen-0.3.0-py3-none-any.whl (345 kB)

|████████████████████████████████| 345 kB 968 kB/s

Collecting openai>=1.3

Using cached openai-1.45.0-py3-none-any.whl (374 kB)

Collecting docker

Downloading docker-7.1.0-py3-none-any.whl (147 kB)

|████████████████████████████████| 147 kB 17.6 MB/s

Collecting diskcache

Using cached diskcache-5.6.3-py3-none-any.whl (45 kB)

Collecting flaml

Using cached FLAML-2.2.0-py3-none-any.whl (297 kB)

Collecting termcolor

Using cached termcolor-2.4.0-py3-none-any.whl (7.7 kB)

Collecting numpy<2,>=1.17.0

Using cached numpy-1.24.4-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (17.3 MB)

Collecting pydantic!=2.6.0,<3,>=1.10

Using cached pydantic-2.9.1-py3-none-any.whl (434 kB)

Collecting tiktoken

Using cached tiktoken-0.7.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.1 MB)

Collecting python-dotenv

Using cached python_dotenv-1.0.1-py3-none-any.whl (19 kB)

Collecting packaging

Using cached packaging-24.1-py3-none-any.whl (53 kB)

Collecting httpx<1,>=0.23.0

Using cached httpx-0.27.2-py3-none-any.whl (76 kB)

Collecting jiter<1,>=0.4.0

Using cached jiter-0.5.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (319 kB)

查看 AutoGen 版本，执行如下命令查看。

(autogentest) root@flexusx-7305:~# pip list | grep autogen

autogen 0.3.0

(autogentest) root@flexusx-7305:~#

3.4 安装 AutoGen Studio****

执行 wget 命令下载 autogenstudio 包，或者执行 pip install autogenstudio，如下所示。

(autogentest)root@flexusx-7305:~#wget https://files.pythonhosted.org/packages/4c/ec/adb715c79656b18b31bfdfbc0253f7f4486b22fd7f7f3a21234bf5d0597c/autogenstudio-0.1.5-py3-none-any.whl

--2024-09-1511:55:58-- https://files.pythonhosted.org/packages/4c/ec/adb715c79656b18b31bfdfbc0253f7f4486b22fd7f7f3a21234bf5d0597c/autogenstudio-0.1.5-py3-none-any.whl

Resolving files.pythonhosted.org (files.pythonhosted.org)... 151.101.88.223, 2a04:4e42:15::223

Connecting to files.pythonhosted.org (files.pythonhosted.org)|151.101.88.223|:443... connected.

HTTP request sent, awaiting response... 200 OK

Length: 4434948 (4.2M) [application/octet-stream]

Saving to: ‘autogenstudio-0.1.5-py3-none-any.whl’

autogenstudio-0.1.5-py3-none-any.whl 100%[======================>] 4.23M 6.43MB/s in 0.7s

2024-09-15 11:55:59 (6.43 MB/s) - ‘autogenstudio-0.1.5-py3-none-any.whl’ saved [4434948/4434948]

pip 安装 autogenstudio，如下所示。

(autogentest) root@flexusx-7305:~# pip install autogenstudio-0.1.5-py3-none-any.whl

Processing ./autogenstudio-0.1.5-py3-none-any.whl

Collecting arxiv

Downloading arxiv-2.1.3-py3-none-any.whl (11 kB)

Collecting pydantic

Using cached pydantic-2.9.1-py3-none-any.whl (434 kB)

Collecting numpy<2.0.0

Downloading numpy-1.26.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (18.2 MB) 18.2/18.2 MB 49.9 MB/s eta 0:00:00

Collecting sqlmodel

Downloading sqlmodel-0.0.22-py3-none-any.whl (28 kB)

Collecting fastapi

Downloading fastapi-0.114.2-py3-none-any.whl (94 kB)94.0/94.0 kB 864.9 kB/s eta 0:00:00

Collecting typer

Downloading typer-0.12.5-py3-none-any.whl (47 kB) 47.3/47.3 kB 3.9 MB/s eta 0:00:00

Collecting alembic

Downloading alembic-1.13.2-py3-none-any.whl (232 kB)233.0/233.0 kB 3.1 MB/s eta 0:00:00

Collecting loguru

Downloading loguru-0.7.2-py3-none-any.whl (62 kB)62.5/62.5 kB 13.6 MB/s eta 0:00:00

至此，所有部署工作完成。

四、运行 AutoGen Studio****

启动 AutoGen Studio，指定端口，如下所示。

(autogentest) root@flexusx-7305:~# autogenstudio ui --host 0.0.0.0 --port 3001

2024-09-15 14:22:26.963 | INFO | autogenstudio.utils.utils:get_db_uri:253 - Using database URI: sqlite:root/.autogenstudio/database.sqlite

2024-09-15 14:22:26.963 | INFO | autogenstudio.utils.utils:init_app_folders:288 - Initialized application data folder: /root/.autogenstudio

INFO: Started server process [153974]

INFO: Waiting for application startup.

***** App started *****

2024-09-15 14:22:27.006 | INFO | autogenstudio.database.utils:init_db_samples:148 - Database already initialized with Default and Travel Planning Workflows

INFO: Application startup complete.

INFO: Uvicorn running on http://0.0.0.0:3001 (Press CTRL+C to quit)

访问 http://服务器 IP:3001，界面如下所示。

点击 Build 按钮，进入构建页面，如下所示。

下面通过展示实例测试多智能体交互。

五、实例展示****

5.1 构建实例****

创建 test.py 文件，复制如下代码。

import os

from autogen import AssistantAgent, UserProxyAgent

llm_config = {"model": "",

"api_key": "",

"base_url": ""

}

assistant = AssistantAgent("assistant", llm_config=llm_config)

user_proxy = UserProxyAgent("user_proxy",

llm_config=False,

human_input_mode="NEVER",

code_execution_config=False)

开始对话

chat_result = user_proxy.initiate_chat(

assistant,

message="北京有哪些景点？",

max_turns=2

)

其中， 需要替换为模型名称， 为密钥， 为基础 URL，API 需要兼容 OpenAI API。

5.2 运行****

执行 python test.py 运行代码，如下所示。

(autogentest) root@flexusx-7305:~# python test.py

user_proxy (to assistant):

北京有哪些景点？

assistant (to user_proxy):

北京是中国的首都，拥有丰富的历史遗产和文化景点。其中一些最著名的景点包括：

故宫：清朝的皇宫，也是世界上最大的宫殿建筑群。

天安门广场： мире 最大的城市广场之一，是许多重要事件的发生地。

长城：世界上最长的墙，也是中国最著名的景点之一。

juven]).

颐和园：清朝皇帝的夏宫，被誉为“中国园林之母”。

圆明园：清朝皇帝的私家园林，也是中国最大的园林之一。

天坛：明清两代皇帝在这里祭祀天地和祖先的场所。

东城区：位于北京中心的商业区，拥有众多高档购物商场和餐厅。

西单：北京最繁华的购物街之一。

北京奥林匹克公园：2008 年北京奥运会的举办地，内有鸟巢等标志性建筑。

(autogentest) root@flexusx-7305:~#

六、总结****

通过深入探索并充分体验 “Flexus 云服务器 X 实例” 后，顺利部署了 AutoGen 及其配套的开发环境 AutoGen Studio，并展示了其高效运行的成果。此次实践不仅突显了 Flexus 云服务器 X 实例 在部署与应用上的便捷与高效特性，更以其丝滑无阻的操作体验，让即便是技术新手也能迅速掌握，日常运维工作因此变得前所未有的轻松与高效。对于渴望在代码管理领域实现高效与流畅体验的用户而言，Flexus 云服务器 X 实例 无疑是一个不容错过的理想之选。立即行动，亲身体验其卓越性能与便捷管理带来的无限可能！

审核编辑 黄宇