Searching protocol for "non-reasoning"
Boost non-reasoning task performance with repetition.
Boost LLM accuracy with prompt repetition.