自然语言处理从入门到应用——LangChain：代理（Agents）-[自定义MRKL代理]

分类目录：《自然语言处理从入门到应用》总目录

---

本文将介绍如何创建自己的自定义MRKL Agent。MRKL Agent由三个部分组成：

• 工具：代理可用的工具。
• LLMChain：生成以一定方式解析的文本，以确定要采取哪个动作。
• 代理类本身：解析LLMChain的输出，以确定要采取哪个动作。

自定义LLMChain（Custom LLMChain）

本部分将介绍如何通过创建自定义LLMChain来创建自定义MRKL代理。创建自定义代理的第一种方法是使用现有的代理类，但使用自定义LLMCain。这也是创建自定义代理的最简单方法。强烈建议使用ZeroShotAgent ，因为目前这是最通用的一个。

创建自定义LLMCain的大部分工作都归结为提示符。因为我们使用的是一个现有的代理类来解析输出，所以提示符中要生成该格式的文本是非常重要的。此外，我们目前需要一个agent_scratchpad输入变量来记录以前的操作和观察结果，这几乎总是提示符的最后一部分。但是，除了这些说明之外，我们还可以根据需要自定义提示。为了确保提示符包含适当的指令，我们将在该类上使用helper方法。ZeroShotAgent的helper方法接受以下参数：

• tools：座席将有权访问的工具列表，用于设置提示的格式。
• prefix：要放在工具列表前面的字符串。
• suffix：放在工具列表后面的字符串。
• input_variables：最后提示所期望的输入变量列表。

在下面的示例中，我们将给予我们的代理访问Google搜索，我们将定制它，我们将让它回答为盗版。

from langchain.agents import ZeroShotAgent, Tool, AgentExecutor
from langchain import OpenAI, SerpAPIWrapper, LLMChain


search = SerpAPIWrapper()
tools = [
    Tool(
        name = "Search",
        func=search.run,
        description="useful for when you need to answer questions about current events"
    )
]


prefix = """Answer the following questions as best you can, but speaking as a pirate might speak. You have access to the following tools:"""
suffix = """Begin! Remember to speak as a pirate when giving your final answer. Use lots of "Args"
 
Question: {input}
{agent_scratchpad}"""
 
prompt = ZeroShotAgent.create_prompt(
    tools, 
    prefix=prefix, 
    suffix=suffix, 
    input_variables=["input", "agent_scratchpad"]
)

如果感到好奇，我们现在可以看看最终的提示模板，看看它看起来像当它的所有放在一起：

print(prompt.template)

输出：

Answer the following questions as best you can, but speaking as a pirate might speak. You have access to the following tools:
 
Search: useful for when you need to answer questions about current events
 
Use the following format:
 
Question: the input question you must answer
Thought: you should always think about what to do
Action: the action to take, should be one of [Search]
Action Input: the input to the action
Observation: the result of the action
... (this Thought/Action/Action Input/Observation can repeat N times)
Thought: I now know the final answer
Final Answer: the final answer to the original input question
 
Begin! Remember to speak as a pirate when giving your final answer. Use lots of "Args"
 
Question: {input}
{agent_scratchpad}

需要注意的是，我们能够为代理提供自定义的提示模板，即不限于由create_prompt函数生成的提示，假设它满足代理的要求。例如，对于ZeroShotAgent，我们需要确保它应该有一个以Action:开头的字符串和一个以Action Input:开头的字符串，并且两者都应该用换行符分隔。

llm_chain = LLMChain(llm=OpenAI(temperature=0), prompt=prompt)

tool_names = [tool.name for tool in tools]
agent = ZeroShotAgent(llm_chain=llm_chain, allowed_tools=tool_names)

agent_executor = AgentExecutor.from_agent_and_tools(agent=agent, tools=tools, verbose=True)

agent_executor.run("How many people live in canada as of 2023?")

输出：

> Entering new AgentExecutor chain...
Thought: I need to find out the population of Canada
Action: Search
Action Input: Population of Canada 2023
Observation: The current population of Canada is 38,661,927 as of Sunday, April 16, 2023, based on Worldometer elaboration of the latest United Nations data.
Thought: I now know the final answer
Final Answer: Arrr, Canada be havin' 38,661,927 people livin' there as of 2023!
 
> Finished chain.

"Arrr, Canada be havin' 38,661,927 people livin' there as of 2023!"

多路输入 (Multiple Inputs）

代理还可以处理需要多个输入的提示：

prefix = """Answer the following questions as best you can. You have access to the following tools:"""
suffix = """When answering, you MUST speak in the following language: {language}.
 
Question: {input}
{agent_scratchpad}"""
 
prompt = ZeroShotAgent.create_prompt(
    tools, 
    prefix=prefix, 
    suffix=suffix, 
    input_variables=["input", "language", "agent_scratchpad"]
)

llm_chain = LLMChain(llm=OpenAI(temperature=0), prompt=prompt)

agent = ZeroShotAgent(llm_chain=llm_chain, tools=tools)

agent_executor = AgentExecutor.from_agent_and_tools(agent=agent, tools=tools, verbose=True)

agent_executor.run(input="How many people live in canada as of 2023?", language="italian")

输出：

> Entering new AgentExecutor chain...
Thought: I should look for recent population estimates.
Action: Search
Action Input: Canada population 2023
Observation: 39,566,248
Thought: I should double check this number.
Action: Search
Action Input: Canada population estimates 2023
Observation: Canada's population was estimated at 39,566,248 on January 1, 2023, after a record population growth of 1,050,110 people from January 1, 2022, to January 1, 2023.
Thought: I now know the final answer.
Final Answer: La popolazione del Canada è stata stimata a 39.566.248 il 1° gennaio 2023, dopo un record di crescita demografica di 1.050.110 persone dal 1° gennaio 2022 al 1° gennaio 2023.
 
> Finished chain.

'La popolazione del Canada è stata stimata a 39.566.248 il 1° gennaio 2023, dopo un record di crescita demografica di 1.050.110 persone dal 1° gennaio 2022 al 1° gennaio 2023.'

参考文献：

[1] LangChain 🦜️🔗 中文网，跟着LangChain一起学LLM/GPT开发：https://www.langchain.com.cn/

[2] LangChain中文网 - LangChain 是一个用于开发由语言模型驱动的应用程序的框架：http://www.cnlangchain.com/