📄 中文摘要
在AI代理中，任务的上下文通常由消息承载。通过将消息作为有效负载发送到模型服务，LLM能够“看到”当前任务的状态。这一过程涉及将用户输入和系统响应整合为一系列消息，以便模型能够理解并生成相应的输出。示例代码展示了如何使用OpenAI的API进行消息传递，确保模型在处理请求时具备必要的上下文信息。通过这种方式，AI能够更有效地执行任务，提升用户体验。
聊天模板：从消息到令牌
出处: Chat Template: From Messages To Tokens
发布: 2026年3月10日
📄 English Summary
Chat Template: From Messages To Tokens
In an AI agent, context is typically conveyed through messages. By sending messages as payloads to the model service, the LLM can 'see' the current state of the task. This process involves integrating user inputs and system responses into a series of messages, allowing the model to understand and generate appropriate outputs. Example code illustrates how to use OpenAI's API for message passing, ensuring that the model has the necessary contextual information when processing requests. This approach enables AI to perform tasks more effectively, enhancing user experience.