179 questions
No questions match those filters.
← All questions
System DesignHard· Latency· RAG· Optimization
The p99 budget just dropped from 800ms to 250ms. One en...
This is one of the questions in the full AI/ML interview bank. Pro unlocks all 1789 questions; Premium includes the same bank plus the highest daily Practice limit.
See plansThe decision depends on the measured value of the selector. If the selector’s contribution to accuracy is near zero, zero-shot is the correct choice. If it is high, reducing the number of chunks may hurt recall. A third option is to replace the heavy selector with a distilled 110M-parameter scorer, which can maintain performance while meeting the tighter latency requirements.