Minimal example for LiteLLM with using GPT-4[T] vs. a model that doesn't support functions calling and needs prompt wrapping?

[twardoch]

Would anyone care to contribute a minimal Python example for LiteLLM with using GPT-4 or GPT-4Turbo (that supports function calling) vs. a model that doesn't support functions calling and needs prompt wrapping?

Ideally starting from something like a pydantic model.

[jxnl]

https://jxnl.github.io/instructor/concepts/patching/