--- title: "对话系统构建指南"

description: "介绍如何使用大语言模型构建智能对话系统" tags: ["对话系统", "Chatbot", "LLM"] category: "llm" icon: "🧠"

对话系统构建指南

对话系统类型

• *任务型对?：完成特定任务（订餐、订票）
• *闲聊型对?：开放式聊天
• *检索增强对?：结合知识库回答问题
• 多轮对话：维持上下文的连续对?

使用API构建对话系统

from openai import OpenAI
client = OpenAI()

class SimpleChatbot:
    def __init__(self, system_prompt="你是一个有帮助的助手?):
        self.history = []
        self.system_prompt = system_prompt

    def chat(self, user_input):
        self.history.append({"role": "user", "content": user_input})
        messages = [{"role": "system", "content": self.system_prompt}] + self.history
        response = client.chat.completions.create(model="gpt-3.5-turbo", messages=messages)
        reply = response.choices[0].message.content
        self.history.append({"role": "assistant", "content": reply})
        return reply

    def reset(self):
        self.history = []

bot = SimpleChatbot()
print(bot.chat("你好?))
print(bot.chat("什么是机器学习?))

上下文管?

当对话历史过长时，需要进行上下文管理?

class ContextManager:
    def __init__(self, max_history=10):
        self.history = []
        self.max_history = max_history

    def add_message(self, role, content):
        self.history.append({"role": role, "content": content})
        if len(self.history) > self.max_history:
            self.history = self.history[-self.max_history:]

    def get_context(self):
        return self.history

    def summarize_and_compress(self, llm_func):
        full_text = "\n".join([f"{m['role']}: {m['content']}" for m in self.history])
        summary = llm_func(f"请总结以下对话要点：\n{full_text}")
        self.history = [{"role": "system", "content": f"对话历史摘要：{summary}"}]

ctx = ContextManager(max_history=20)
ctx.add_message("user", "我想学Python")
ctx.add_message("assistant", "建议从基础语法开始学?)
print(ctx.get_context())

对话模板格式

不同模型使用不同的对话模板：

• ChatML：OpenAI系列使用，格式为 <|im_start|>system...
• LLaMA：使?[INST]...[/INST] 标记
• Alpaca：使?### Instruction: 前缀

流式对话

def stream_chat(messages, model="gpt-3.5-turbo"):
    stream = client.chat.completions.create(
        model=model, messages=messages, stream=True
    )
    response = ""
    for chunk in stream:
        if chunk.choices[0].delta.content:
            content = chunk.choices[0].delta.content
            print(content, end="", flush=True)
            response += content
    return response

总结

构建对话系统需要考虑上下文管理、对话模板、历史压缩等关键问题。通过合理设计，可以创建流畅的多轮对话体验?