My Agent is not stopping even i use TERMINATE

[sunilsvsnlr]

I'm using RoundRobinGroupChat, here I'm using two agents QueryAssistant and CodeExectuor. Its working fine but after successful execution of code it should terminate and ask for new question, if code execution is failed it should again call QueryAssistant to generate new version of code to execute properly. Once successful execution it should terminate and ask new question. Below is my code for reference. Please suggest me.

Below is the code
https://github.com/sunilsvsnlr/autogendemos/blob/main/roundrobingroupchat_demo.py

Console output:
---------- assistant ----------
To find the maximum project cost from the merged data, we need to follow these steps:

1. Load the data from the JSON file "C:/projects/proarch/genai/merged/merged_data.json".
2. Convert the cost column to a numeric format if it's not already.
3. Find and return the maximum value in the cost column.

Here's how you can implement this using Python:

import pandas as pd

# Load data from JSON file
file_path = "C:/projects/proarch/genai/merged/merged_data.json"
data = pd.read_json(file_path)

# Convert 'cost' column to numeric, handling any errors
data['cost'] = pd.to_numeric(data['cost'], errors='coerce')

# Find the maximum project cost
max_cost = data['cost'].max()

print(f"The maximum project cost till now is: {max_cost}")

This script will load the data, convert the cost column to a numeric format, and then find and print the maximum value in the cost column. If there are any errors converting the values, they will be handled by setting them to NaN and ignored during calculations.
---------- code_executor ----------
The maximum project cost till now is: 990639.66

---------- assistant ----------
Great! It looks like the script successfully loaded the data from the JSON file and found the maximum project cost, which is $990,639.66.

If you have any more questions or need further analysis, feel free to ask! If you're done with this task, I'll terminate.
---------- code_executor ----------
No code blocks found in the thread. Please provide at least one markdown-encoded code block to execute (i.e., quoting code in python or sh code blocks).
---------- assistant ----------
I apologize for any confusion earlier. Here's a simple Python script that finds the maximum project cost from the merged data:

import pandas as pd

# Load data from JSON file
file_path = "C:/projects/proarch/genai/merged/merged_data.json"
data = pd.read_json(file_path)

# Convert 'cost' column to numeric, handling any errors
data['cost'] = pd.to_numeric(data['cost'], errors='coerce')

# Find the maximum project cost
max_cost = data['cost'].max()

print(f"The maximum project cost till now is: {max_cost}")

To execute this script, you can copy and paste it into a Python environment like Jupyter Notebook, PyCharm, or any other Python IDE. Make sure you have Pandas installed (pip install pandas) before running the code.
---------- code_executor ----------
The maximum project cost till now is: 990639.66

[ekzhu]

Please read the section in tutorial about termination, and consider creating your own custom termination condition.

https://microsoft.github.io/autogen/stable/user-guide/agentchat-user-guide/tutorial/termination.html

Most likely, each application scenario requires a different type of termination setting, relying on model to generate TERMINATE is not always the best approach.

[sunilsvsnlr]

Hi @ekzhu
Could you please provide any example link to create custom agent for termination?

[ekzhu]

Please use the implementations of builtin termination conditions as reference. E.g., https://microsoft.github.io/autogen/stable/_modules/autogen_agentchat/conditions/_terminations.html#SourceMatchTermination

For custom agent, it is part of the documentation: https://microsoft.github.io/autogen/stable/user-guide/agentchat-user-guide/custom-agents.html