Qwen-Agent: Qwen-based framework for intelligent agent applications, including tool calls, code interpreters, RAGs and Chrome extensions.

General Introduction

Qwen-Agent is an intelligent agent application framework developed based on Qwen 2.0 and above with command following, tool usage, planning and memorization capabilities. The framework provides a variety of sample applications, such as browser helpers, code interpreters, and custom helpers, to help developers quickly build and deploy intelligent agent applications.Qwen-Agent supports a variety of functional modules, including function calls, code interpretation, RAG (Retrieval Augmentation Generation), and Chrome extensions, for a variety of complex application scenarios.

BrowserQwen, a browser-based extension for smart bodies

Function List

• function call: Supports automation of complex tasks through function calls.
• code interpreter: Provide code interpretation features to help users understand and execute code.
• RAG (Retrieval Augmentation Generation): Supports large-scale document retrieval and generation for long-document quizzing tasks.
• Chrome Extension: Provides browser extensions to enhance the user's browsing experience.
• Customization Tools: Allow users to add custom tools to extend the functionality of the framework.

Using Help

Installation process

1. Installing the stable version from PyPI::

   pip install -U "qwen-agent[gui,rag,code_interpreter,python_executor]"
   

   Or install the least dependent version:

   pip install -U qwen-agent
   

   Optional dependencies include:

   • [gui]: Support for Gradio-based GUIs
   • [rag]: Support for RAG
   • [code_interpreter]: Support for code interpreters
   • [python_executor]: Tool-integrated reasoning with support for Qwen2.5-Math
2. Install the latest development version from source::

   git clone https://github.com/QwenLM/Qwen-Agent.git
   cd Qwen-Agent
   pip install -e ./[gui,rag,code_interpreter,python_executor]
   

   Or install the least dependent version:

   pip install -e ./
   

Functional operation flow

1. Configuration Modeling Service::
   • Use the modeling service provided by AliCloud DashScope to set environment variables DASHSCOPE_API_KEY for your DashScope API key.
   • Alternatively, deploy and use your own modeling service, following the instructions in the README for Qwen2.
2. Developing Custom Agents::
   Qwen-Agent provides base components, such as the components inherited from the BaseChatModel and the LLMs inherited from BaseTool tools, as well as high-level components such as those inherited from Agent The following is an example of creating an agent that can read a PDF file and use the tool. The following is an example of creating an agent that can read PDF files and use tools:

   from qwen_agent.agents import Assistant
   from qwen_agent.tools.base import BaseTool, register_tool
   @register_tool('my_image_gen')
   class MyImageGen(BaseTool):
   description = 'AI 绘画服务，输入文本描述，返回基于文本信息绘制的图像 URL。'
   parameters = [{'name': 'prompt', 'type': 'string', 'description': '所需图像内容的详细描述', 'required': True}]
   def call(self, params: str, **kwargs) -> str:
   prompt = json5.loads(params)['prompt']
   prompt = urllib.parse.quote(prompt)
   return json5.dumps({'image_url': f'https://image.pollinations.ai/prompt/{prompt}'}, ensure_ascii=False)
   llm_cfg = {
   'model': 'qwen-max',
   'model_server': 'dashscope',
   'generate_cfg': {'top_p': 0.8}
   }
   tools = ['my_image_gen', 'code_interpreter']
   files = ['./examples/resource/doc.pdf']
   bot = Assistant(llm=llm_cfg, system_message='You are a helpful assistant.', function_list=tools, files=files)
   messages = []
   while True:
   query = input('user query: ')
   messages.append({'role': 'user', 'content': query})
   response = bot.run(messages=messages)
   for res in response:
   print('bot response:', res)
   messages.extend(res)
   

3. Quick Start Gradio Demo::

   from qwen_agent.gui import WebUI
   WebUI(bot).run()