背景
在完整的AI Agent中,工具的调用这里采用ChatGPT的Function Calling的思想,通过定义“Tools Schema”来连接LLM和Tools的调用.
架构结构
User Query
↓
Agent (LLM with tool access)
↓
Decide:
- Call Tool?
- Or Final Answer?
↓
Tool Result
↓
Append to Context
↓
Repeat
1,先定义Tools Schema
这里用到的工具: Search Database, Send Mail, Get Weather等等. 这些工具也就是自定义工具方法的名字.
tools = [
{
"type": "function",
"function": {
"name": "Search Database",
"description": "Retrive profile from database",
"parameters":{
"type": "object",
"properties": {
"name": {"type": "string"}
},
"required": ["name"]
}
}
},
{
"type": "function",
"function": {
"name": "Send Mail",
"description": "Send Mail to Gmail",
"parameters":{
"to": {"type": "string"},
"subject": {"type": "string"},
"body": {"type": "string"}
}
}
},
{
"type": "function",
"function": {
"name": "Get Weather",
"description": "Get Weather from location",
"parameters":{
"type": "object",
"properties": {
"location": {"type": "string"}
},
"required": ["location"]
}
}
},
......
]
2, 由LLM决定下一步的行动 (ReAct框架)
Agent Prompt:
prompt = ‘
You are an intelligent task agent.
User query: {Query}
At each step:
- Decide whether to call a tool.
- If needed, call exactly one tool.
- If enough information is gathered, provide final answer.
- Always reason step-by-step.
Available tools:
1. Search Database
2. Send Mail
3. Get Weather
’
messages = [{"role": "user", "content": user_query}]
while True:
response = client.chat.completions.create(
model="gpt-4o",
messages=messages,
tools=tools
)
message = response.choices[0].message
if message.tool_calls:
tool_call = message.tool_calls[0]
result = execute_tool(tool_call)
messages.append(message)
messages.append({
"role": "tool",
"tool_call_id": tool_call.id,
"content": json.dumps(result)
})
else:
final_answer = message.content
break
评价: (ReAct框架 + Function Calling 推荐指数:🌟🌟🌟🌟)
优点:
1, 动态规划
2, 自动条件判断
3, 支持分支逻辑
4, 自我修正
5, 更符合人类思维:ReAct
