示例代码

from dataclasses import dataclass

from autogen_core import AgentId, MessageContext, RoutedAgent, SingleThreadedAgentRuntime, message_handler
from autogen_core.model_context import BufferedChatCompletionContext
from autogen_core.models import AssistantMessage, ChatCompletionClient, SystemMessage, UserMessage
from autogen_ext.models.openai import OpenAIChatCompletionClient

@dataclass
class Message:
    content: str

class SimpleAgentWithContext(RoutedAgent):
    def __init__(self, model_client: ChatCompletionClient) -> None:
        super().__init__("A simple agent")
        self._system_messages = [SystemMessage(content="You are a helpful AI assistant.")]
        self._model_client = model_client
        self._model_context = BufferedChatCompletionContext(buffer_size=5)

    @message_handler
    async def handle_user_message(self, message: Message, ctx: MessageContext) -> Message:
        # Prepare input to the chat completion model.
        user_message = UserMessage(content=message.content, source="user")
        # Add message to model context.
        await self._model_context.add_message(user_message)
        # Generate a response.
        response = await self._model_client.create(
            self._system_messages + (await self._model_context.get_messages()),
            cancellation_token=ctx.cancellation_token,
        )
        # Return with the model's response.
        assert isinstance(response.content, str)
        # Add message to model context.
        await self._model_context.add_message(AssistantMessage(content=response.content, source=self.metadata["type"]))
        return Message(content=response.content)

runtime = SingleThreadedAgentRuntime()

model_client = OpenAIChatCompletionClient(
                    model="GLM-4-Air-0111",
                    api_key = "your api key",
                    base_url="https://open.bigmodel.cn/api/paas/v4/",
                    model_capabilities={
                "vision": True,
                "function_calling": True,
                "json_output": True,
            }
                    )

await SimpleAgentWithContext.register(
    runtime,
    "simple_agent_context",
    lambda: SimpleAgentWithContext(model_client),
)
# Start the runtime processing messages.
runtime.start()
agent_id = AgentId("simple_agent_context", "default")

# First question.
message = Message("Hello, what are some fun things to do in Seattle?")
print(f"Question: {message.content}")
response = await runtime.send_message(message, agent_id)
print(f"Response: {response.content}")
print("-----")

# Second question.
message = Message("What was the first thing you mentioned?")
print(f"Question: {message.content}")
response = await runtime.send_message(message, agent_id)
print(f"Response: {response.content}")

# Stop the runtime processing messages.
await runtime.stop()

Question: Hello, what are some fun things to do in Seattle?
Response: Seattle is a vibrant city with a lot to offer! Here are some fun things to do, catering to different interests:

**For the Outdoorsy Type:**

*   **Visit Pike Place Market:** This iconic public market is a must-see. Watch fish mongers throw fish, sample local produce, and grab a bite to eat.
*   **Explore the Seattle Great Wheel:** Enjoy stunning views of Elliott Bay and the city skyline from this Ferris wheel on the waterfront.
*   **Take a walk or bike ride along the Seattle Waterfront:** Enjoy the fresh air, views of the Puget Sound, and attractions like the Seattle Aquarium.
*   **Hike the Chirico Trail at Discovery Park:** This 2.8-mile loop offers beautiful views of Puget Sound and the Olympic Mountains.
*   **Go kayaking or stand-up paddleboarding:** Rent a vessel and explore the calm waters of Lake Union or the more adventurous Puget Sound.
*   **Visit the Japanese Garden at the Seattle Arboretum:** Stroll through this serene 3.5-acre garden featuring winding paths, water features, and traditional Japanese landscaping.
*   **Take a day trip to one of the nearby mountains:** Mount Rainier, Mount Baker, and the Cascades offer incredible hiking, scenic drives, and wildlife viewing opportunities.

**For the Culture Enthusiast:**

*   **Explore the Museum of Pop Culture (MoPOP):** This museum dedicated to contemporary popular culture is a fun and interactive experience, with exhibits on music, science fiction, and fantasy.
*   **Visit the Seattle Art Museum (SAM):** Admire a diverse collection of art from around the world, including Native American art and contemporary works.
*   **See a show at the 5th Avenue Theatre or the Paramount Theatre:** Catch a Broadway show or a concert in one of these historic venues.
*   **Explore the historic Pioneer Square district:** This area features Victorian-era architecture, art galleries, and unique shops.
*   **Visit the Ballard Locks (Hiram M. Chittenden Locks):** Watch boats pass between Puget Sound and the Ship Canal, and enjoy the surrounding park.
*   **Attend a festival or event:** Seattle hosts numerous festivals throughout the year, including the Seattle International Film Festival, Bumbershoot, and the Capitol Hill Block Party.

**For the Foodie:**

*   **Enjoy a coffee at a local cafe:** Seattle is the birthplace of Starbucks, but you'll find many independent coffee shops offering delicious brews.
*   **Try some fresh seafood:** From salmon to Dungeness crab, Seattle is known for its fresh seafood. Canlis, The Pink Door, and Ivar's Acres of Clams are great options.
*   **Explore the diverse culinary scene:** Seattle has a wide range of cuisines, from Vietnamese pho to Ethiopian injera. Explore the International District for a variety of options.
*   **Visit a food truck pod:** Food truck pods are a great way to sample a variety of cuisines in one location. South Lake Union and Capitol Hill have several pods.

**For the Music Lover:**

*   **Visit the Experience Music Project (now part of MoPOP):** This museum celebrates the history of music, with a focus on Seattle's grunge scene.
*   **See a live music show:** Seattle has a thriving music scene, with venues like The Crocodile, Neumos, and The Showbox hosting shows almost every night.
*   **Explore the Music Row district on 5th Avenue:** This area is home to several music venues, record stores, and other music-related businesses.

**Unique Seattle Experiences:**

*   **Ride the monorail:** This historic monorail connects downtown Seattle to the Seattle Center.
*   **Visit the Seattle Center:** This park, built for the 1962 World's Fair, features the Space Needle, the International Fountain, and other attractions.
*   **Take a Bill Speidel's Underground Tour:** Explore the hidden passages and history beneath the city streets.
*   **Go on a whale watching tour:** Several companies offer tours to see orcas and other marine wildlife in Puget Sound.

**Tips for Planning Your Trip:**

*   **Best time to visit:** Spring and fall offer pleasant weather and fewer crowds. Summer can be warm and sunny, but also crowded and expensive.
*   **Getting around:** Seattle has a good public transportation system, including buses, light rail, and streetcars. Ride-sharing services are also readily available.
*   **Book accommodations and activities in advance:** Especially during peak season.

This is just a starting point, and there's much more to discover in Seattle! To help me narrow down the options further, tell me:

*   **When are you planning to visit?**
*   **How long will you be in Seattle?**
*   **What are your interests (e.g., outdoors, food, art, music)?**
*   **Who
-----
Question: What was the first thing you mentioned?
Response: The first thing I mentioned in my response about fun things to do in Seattle was **Visit Pike Place Market**.

代码解释

这段代码使用了一个名为 autogen_core 的库，该库通常用于创建和运行基于AI的聊天代理。代码的目的是创建一个简单的代理，它能够上下文感知地回答用户的问题。下面是代码的逻辑和功能的详细解释：

1. 导入必要的类和模块：

   • dataclass 用于定义简单的数据结构。
   • autogen_core 中的 AgentId、MessageContext、RoutedAgent、SingleThreadedAgentRuntime 和 message_handler 用于创建和管理代理。
   • BufferedChatCompletionContext 用于管理聊天上下文。
   • AssistantMessage、ChatCompletionClient、SystemMessage 和 UserMessage 用于构建和发送消息。
   • OpenAIChatCompletionClient 是一个用于与OpenAI模型进行交互的客户端。

2. 定义 Message 数据类：

   • 这是一个简单的数据结构，用于存储消息的内容。

3. 创建 SimpleAgentWithContext 类：

   • 这个类继承自 RoutedAgent，代表一个能够处理消息的代理。
   • 构造函数初始化代理的名称、系统消息列表、模型客户端和模型上下文。
   • handle_user_message 方法是一个异步消息处理器，它接收用户消息，将其添加到模型上下文中，并使用模型客户端生成响应。
   • 生成的响应被添加回模型上下文，并返回给用户。

4. 初始化运行时环境和模型客户端：

   • runtime 是一个 SingleThreadedAgentRuntime 实例，用于管理代理的运行。
   • model_client 是一个 OpenAIChatCompletionClient 实例，配置了特定的模型名称、API密钥和基本URL。

5. 注册和启动代理：

   • SimpleAgentWithContext 被注册到运行时环境中，并指定了一个构造函数。
   • 运行时环境被启动以开始处理消息。

6. 发送和接收消息：

   • 通过 runtime.send_message 方法发送两个问题给代理，并打印出问题和相应的回答。
   • 第一个问题是关于在Seattle的有趣活动。
   • 第二个问题是关于代理提到的第一个活动。

7. 停止运行时环境：

   • 一旦消息处理完成，运行时环境被停止。

完成的功能：

• 这段代码创建了一个能够理解上下文的聊天代理，它可以接收用户的问题，并根据之前的对话历史生成相关的回答。
• 该代理使用了一个远程的AI模型（通过 OpenAIChatCompletionClient）来生成回答。
• 通过将用户的每个问题和代理的回答存储在上下文中，代理能够理解对话的流程并提供更相关的回答。

请注意，为了使代码能够运行，需要确保 autogen_core 和 autogen_ext 库是可用的，并且 OpenAIChatCompletionClient 能够与指定的模型进行通信。此外，代码中使用了异步编程，这意味着它需要在支持异步操作的Python环境中运行。

另一个例子

from dataclasses import dataclass
from autogen_core import AgentId, MessageContext, RoutedAgent, SingleThreadedAgentRuntime, message_handler
from autogen_core.model_context import BufferedChatCompletionContext
from autogen_core.models import AssistantMessage, ChatCompletionClient, SystemMessage, UserMessage
from autogen_ext.models.openai import OpenAIChatCompletionClient

@dataclass
class Message:
    content: str

class CodeExplainerAgent(RoutedAgent):
    def __init__(self, model_client: ChatCompletionClient) -> None:
        super().__init__("A code explainer agent")
        self._system_messages = [SystemMessage(content="You are an AI assistant specialized in explaining Python code.")]
        self._model_client = model_client
        self._model_context = BufferedChatCompletionContext(buffer_size=5)

    @message_handler
    async def handle_user_message(self, message: Message, ctx: MessageContext) -> Message:
        user_message = UserMessage(content=message.content, source="user")
        await self._model_context.add_message(user_message)
        
        response = await self._model_client.create(
            self._system_messages + (await self._model_context.get_messages()),
            cancellation_token=ctx.cancellation_token,
        )
        
        assert isinstance(response.content, str)
        await self._model_context.add_message(AssistantMessage(content=response.content, source=self.metadata["type"]))
        return Message(content=response.content)

runtime = SingleThreadedAgentRuntime()

model_client = OpenAIChatCompletionClient(
    model="GLM-4-Air-0111",
    api_key="your api key",
    base_url="https://open.bigmodel.cn/api/paas/v4/",
    model_capabilities={
        "vision": True,
        "function_calling": True,
        "json_output": True,
    }
)

await CodeExplainerAgent.register(
    runtime,
    "code_explainer_agent",
    lambda: CodeExplainerAgent(model_client),
)

runtime.start()
agent_id = AgentId("code_explainer_agent", "default")

# 代码解释示例
message = Message("解释以下代码:\n```python\nimport numpy as np\nA = np.array([[1, 2], [3, 4]])\nprint(A.T)\n```")
print(f"Question: {message.content}")
response = await runtime.send_message(message, agent_id)
print(f"Response: {response.content}")
print("-----")

# 继续问上下文相关问题
message = Message("代码中的 `.T` 是什么意思?")
print(f"Question: {message.content}")
response = await runtime.send_message(message, agent_id)
print(f"Response: {response.content}")

await runtime.stop()

Question: 解释以下代码:
```python
import numpy as np
A = np.array([[1, 2], [3, 4]])
print(A.T)
```
Response: 这段代码首先导入了NumPy库，并将其命名为`np`。NumPy是一个用于科学计算的Python库，它提供了一个强大的N维数组对象和一系列用于快速操作数组的函数。

接着，代码创建了一个2x2的NumPy数组`A`，其元素为[[1, 2], [3, 4]]。

最后，代码使用`.T`属性来获取数组`A`的转置，并将其打印出来。`.T`属性是NumPy数组的一个特性，它返回数组的转置视图。对于二维数组，这意味着行和列将互换。

当运行这段代码时，输出将是：

```
[[1 3]
 [2 4]]
```

这显示了原始数组`A`的转置，其中第一行[1, 2]变成了第一列，第二行[3, 4]变成了第二列。
-----
Question: 代码中的 `.T` 是什么意思?
Response: 在 NumPy 中，`.T` 属性用于获取数组的转置。转置是一个矩阵操作，它将矩阵的行和列进行互换。例如，如果一个矩阵 `A` 的形状是 `(m, n)`，那么它的转置 `A.T` 的形状将是 `(n, m)`。

在您的代码示例中：

```python
import numpy as np
A = np.array([[1, 2], [3, 4]])
print(A.T)
```

`A` 是一个 2x2 的矩阵：

```
A = [[1, 2],
     [3, 4]]
```

应用 `.T` 属性后，我们得到 `A` 的转置：

```
A.T = [[1, 3],
       [2, 4]]
```

这里，原来的行变成了列，原来的列变成了行。

对于更高维度的数组，`.T` 同样适用，并且可以结合 `axes` 参数进行更复杂的转置操作，但那超出了您当前代码的范围。对于大多数矩阵操作，简单地使用 `.T` 就足够了。

参考链接：https://microsoft.github.io/autogen/stable/user-guide/core-user-guide/components/model-context.html