Hey, I'm wondering does the model recognize stuff like a conversation flow context?
Such as: User: hi Assistant: Not available ... etc so it can adopt a certain character or rules based on the context. Or maybe "system: don't do that" as of how each model is
