179 questions
No questions match those filters.
← All questions
RAGHard· System Design· Latency
You have an 8k context window, a 2-second p95 budget, a...
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 plansWhen choosing between IRCoT and Self-Ask, one must derive the prefill costs for each. IRCoT involves a cumulative loop that increases prefill tokens, while Self-Ask is generally lighter. The decision should be based on whether the intermediate sub-answers can be compressed into a string that retains the required citation data for the final answer.