使用gpt-35-turbo运行AgentExecutor触发InvalidRequestError错误求助

运行从LangChain官方文档复制的OpenAIFunctionsAgent代码时，执行agent_executor.run()触发以下错误：

InvalidRequestError: Unrecognized request argument supplied: functions

相关代码 #

from langchain.agents import tool
from langchain.agents import OpenAIFunctionsAgent
from langchain.agents import AgentExecutor
from langchain.chat_models import ChatOpenAI
from langchain.schema import SystemMessage

# Create LLM
llm = ChatOpenAI(
    model_kwargs={"engine": deployment_name},
    temperature=0.2)

@tool
def get_word_length(word: str) -> int:
    """Returns the length of a word."""
    return len(word)

tools = [get_word_length]

# Prompt
system_message = SystemMessage(content="You are very powerful assistant, but bad at calculating lengths of words.")
prompt = OpenAIFunctionsAgent.create_prompt(system_message=system_message)

# Create Agent
agent = OpenAIFunctionsAgent(llm=llm, tools=tools, prompt=prompt)

# Create Agent Executor
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)

# Run it
agent_executor.run("how many letters in the word educa?")

错误栈信息 #

> Entering new AgentExecutor chain...
------
InvalidRequestError: Unrecognized request argument supplied: functions                        
Traceback (most recent call last)
Cell In[29], line 51
     48 agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
     50 # Run it
---> 51 agent_executor.run("how many letters in the word educa?")

错误原因 #

OpenAIFunctionsAgent依赖OpenAI的functions参数实现工具调用，但旧版本的gpt-35-turbo（如0301版本）不支持该参数，只有gpt-35-turbo-0613及以后的版本才支持工具调用功能。

修复方案 #

方案1：升级到支持工具调用的模型版本 #

修改ChatOpenAI的初始化代码，指定支持工具调用的模型版本：

llm = ChatOpenAI(
    model="gpt-35-turbo-0613",  # 指定兼容工具调用的模型版本
    model_kwargs={"engine": deployment_name},
    temperature=0.2)

方案2：改用兼容旧模型的Agent类型 #

如果无法升级模型，可使用StructuredChatAgent替代OpenAIFunctionsAgent，示例代码如下：

from langchain.agents import tool
from langchain.agents import AgentExecutor
from langchain.chat_models import ChatOpenAI
from langchain.schema import SystemMessage
from langchain.agents import StructuredChatAgent
from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder

# Create LLM
llm = ChatOpenAI(
    model_kwargs={"engine": deployment_name},
    temperature=0.2)

@tool
def get_word_length(word: str) -> int:
    """Returns the length of a word."""
    return len(word)

tools = [get_word_length]

# 构建适配StructuredChatAgent的Prompt
system_message = SystemMessage(content="You are very powerful assistant, but bad at calculating lengths of words.")
prompt = ChatPromptTemplate.from_messages([
    system_message,
    MessagesPlaceholder(variable_name="chat_history"),
    ("user", "{input}"),
    MessagesPlaceholder(variable_name="agent_scratchpad")
])

# 创建Agent和Executor
agent = StructuredChatAgent(llm=llm, tools=tools, prompt=prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)

# 运行
agent_executor.run("how many letters in the word educa?")

内容的提问来源于stack exchange，提问作者Andrii