Why multiple models in config_list?

[dusty hazel]

Hello! Newbie in the world of AI, LLMs and AutoGen here.

I've been poking around the documentations and notebooks and came across some examples (e.g. https://github.com/microsoft/autogen/blob/main/notebook/agentchat_groupchat_research.ipynb) with multiple models in the config_list. How does it work? What are the advantages in having multiple models? How do we select the models to be included into the list?

Additionally, would appreciate if anyone can point me in the right direction to learn more about the things I should know to become proficient at working with AutoGen! 😃

[indigo talon]

autogen will use the first model in the list that passes the filter (if provided) and is actually available. More info here: https://github.com/microsoft/autogen/blob/main/notebook/oai_openai_utils.ipynb

[hot canyon]

It's explained in https://microsoft.github.io/autogen/docs/Use-Cases/enhanced_inference#error-handling
This question is also asked in https://github.com/microsoft/autogen/issues/749