A LAnguage Model INterface Enginee as god intended, because the rest of the standards to inference LM APIs are 💩.

I despise with pasion:

• The fact that each provider uses a different spec for LLM requests
• The OpenAI spec in particular
• LiteLLM wacky codebase

So this is how it should actually be done. You are welcome.

For user manual see HOW-TO.ipynb