LLM not following description in config

[credelosa2022]

System Info

Os version: Google Colab
Python version: Google Colab
Pandasai version: 2.0.34

🐛 Describe the bug

from pandasai.llm import GooglePalm

llm = GooglePalm(api_key="xxxx")

data_df = pd.read_csv("Loan payments data.csv") #This is from your example in github

sdf = SmartDataframe(data_df, config={"llm": llm,
                                      "description":"You are a data analysis agent. Your main goal is to help non-technical users to analyze data. When you provide visualizations, make sure you include the axis labels.",
                                      "seed": 2024})
                                      
sdf.chat("In the same graph, show the age distribution where each histogram corresponds to a gender category. ")

Is there a better way to do this?

[seanshanker]

i'm a newbie in this area myself...so take what i say with a pinch of salt..i didnt realize you could tell it to plot with axes labels in the description. i've usually done it in the sdf.chat with decent success.

[ArslanSaleem]

credelosa2022 instead of config pass description to Agent class like:

agent = Agent([data_df], config=config, description="You are a data analysis agent. Your main goal is to help non-technical users to analyze data. When you provide visualizations, make sure you include the axis labels.")

agent.chat...

Let me know if it helps

Thanks

[adri0]

I noticed that since the version 2.0.34 my tests using the and Agent with FakeLLM started breaking.