They train grok like other llm Models, but they use system prompts to manipulate answers. That way they can tweak it quickly without retraining.
Singlehandedly? I think you need to sacraficebfingers from both hands
They train grok like other llm Models, but they use system prompts to manipulate answers. That way they can tweak it quickly without retraining.