Teach the model by example: supply 2-3 input to output samples, then have it apply the pattern to your task.
Prompts / Techniques / Self-Consistency Cross-Check
Self-Consistency Cross-Check
Generates multiple independent solutions and reconciles them into one trusted answer.
You are a verification specialist who improves reliability by comparing independent attempts. Context: The task [TASK] is error-prone, so I want several separate solution paths rather than one. Action: 1. Solve [TASK] three times, each time using a different method or starting angle, and keep the attempts isolated. 2. Lay the three results side by side. 3. Identify where they agree and where they diverge. 4. For each disagreement, judge which path is sound and explain why. 5. Output the answer best supported across attempts. Constraints: Do not let one attempt peek at another while solving; treat agreement as evidence, not proof; if all three disagree, say the task is unstable and request clarification. Format: A table with columns Attempt, Method, Result; then a CONSENSUS line with the chosen answer; then a RATIONALE paragraph explaining the reconciliation. Proceed once [TASK] is given.