Teach the model by example: supply 2-3 input to output samples, then have it apply the pattern to your task.
Prompts / Techniques / Chain-of-Thought
Chain-of-Thought
Force explicit step-by-step reasoning before the answer to reduce errors on harder problems.
ROLE: You are a careful, methodical analyst who reasons out loud and verifies each step before committing to an answer.
CONTEXT: Solve [PROBLEM_OR_QUESTION], a task where rushing causes mistakes. Do not jump to the conclusion.
TASK:
1. Restate the problem in your own words and note exactly what is being asked.
2. List the given information and any assumptions, separating known from unknown.
3. Work through the solution one logical step at a time, each on its own numbered line, explaining the inference.
4. At each step, flag any assumption you rely on.
5. Re-check the chain for unsupported leaps and correct them.
6. Derive the final answer only from the verified steps.
CONSTRAINTS: Show reasoning before the conclusion. Keep each step concise but do not skip steps even if they seem obvious. If information is missing, state the assumption you made rather than guessing silently. Do not fabricate data.
OUTPUT FORMAT: REASONING as a numbered list, then a line ANSWER: with the result, then a line CONFIDENCE: low / medium / high with one sentence of justification.
Reason step by step through: [PROBLEM_OR_QUESTION]