CSV 代理

本文档展示了如何使用代理与 CSV 进行交互。它主要针对问题回答进行了优化。

注意：该代理在底层调用了 Pandas DataFrame 代理，而 Pandas DataFrame 代理又调用了 Python 代理，后者执行了由 LLM 生成的 Python 代码 - 如果 LLM 生成的 Python 代码具有恶意性，这可能是危险的。请谨慎使用。

from langchain.agents import create_csv_agent

from langchain.llms import OpenAI
from langchain.chat_models import ChatOpenAI
from langchain.agents.agent_types import AgentType

使用 ZERO_SHOT_REACT_DESCRIPTION

这展示了如何使用 ZERO_SHOT_REACT_DESCRIPTION 代理类型初始化代理。请注意，这是上述方法的一种替代方式。

agent = create_csv_agent(
    OpenAI(temperature=0),
    'titanic.csv',
    verbose=True,
    agent_type=AgentType.ZERO_SHOT_REACT_DESCRIPTION
)

使用 OpenAI 函数

这展示了如何使用 OPENAI_FUNCTIONS 代理类型初始化代理。请注意，这是上述方法的一种替代方式。

agent = create_csv_agent(
    ChatOpenAI(temperature=0, model="gpt-3.5-turbo-0613"), 
    'titanic.csv', 
    verbose=True, 
    agent_type=AgentType.OPENAI_FUNCTIONS
)

agent.run("how many rows are there?")

Error in on_chain_start callback: 'name'



Invoking: `python_repl_ast` with `df.shape[0]`


891There are 891 rows in the dataframe.

Finished chain.

'There are 891 rows in the dataframe.'

agent.run("how many people have more than 3 siblings")

Error in on_chain_start callback: 'name'



Invoking: `python_repl_ast` with `df[df['SibSp'] > 3]['PassengerId'].count()`


30There are 30 people in the dataframe who have more than 3 siblings.

Finished chain.

'There are 30 people in the dataframe who have more than 3 siblings.'

agent.run("whats the square root of the average age?")

Error in on_chain_start callback: 'name'



Invoking: `python_repl_ast` with `import pandas as pd
import math

# Create a dataframe
data = {'Age': [22, 38, 26, 35, 35]}
df = pd.DataFrame(data)

# Calculate the average age
average_age = df['Age'].mean()

# Calculate the square root of the average age
square_root = math.sqrt(average_age)

square_root`


5.585696017507576The square root of the average age is approximately 5.59.

Finished chain.

'The square root of the average age is approximately 5.59.'

多个 CSV 示例

下面的部分展示了代理如何与作为列表传入的多个 CSV 文件进行交互。

agent = create_csv_agent(ChatOpenAI(temperature=0, model="gpt-3.5-turbo-0613"), ['titanic.csv', 'titanic_age_fillna.csv'], verbose=True, agent_type=AgentType.OPENAI_FUNCTIONS)
agent.run("how many rows in the age column are different between the two dfs?")

Error in on_chain_start callback: 'name'



Invoking: `python_repl_ast` with `df1['Age'].nunique() - df2['Age'].nunique()`


-1There is 1 row in the age column that is different between the two dataframes.

Finished chain.

'There is 1 row in the age column that is different between the two dataframes.'