Your question is Minimize Inference Latency On-Device. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
What strategies would you use to minimize inference latency for an on-device text prediction model?