Inconsistent Token Usage Reporting Between logs and get_actual_usage() Methods

[thrishathi02]

I've noticed a discrepancy in token usage reporting between the two supported methods in AutoGen:

logs method: https://microsoft.github.io/autogen/0.2/docs/notebooks/agentchat_logging/

get_actual_usage() method: https://autogenhub.github.io/autogen/docs/notebooks/agentchat_cost_token_tracking/

Both methods report the same total token usage, but they show a different pattern of token consumption across agents.

For example, in my case:
The logs method shows the user_proxy agent using around 100K tokens.

The get_actual_usage() method shows the user_proxy using no tokens, which seems correct since it doesn’t interact with an LLM.

This discrepancy makes it hard to determine which breakdown is accurate. Could you clarify which method gives the correct token usage per agent?

Thanks!

[Nikhil-Origamis]

HELP!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻

[Nikhil-Origamis]

please