lauraflorentin Reasoning Techniques
Reasoning techniques (like Chain-of-Thought, Tree-of-Thought) force the LLM to show its work. Large Language Models are statistical, not logical. By making them output a step-by-step reasoning path before the final answer, you allow the model to provide context to itself, significantly reducing logic errors and "hallucinations of calculation".
When to Use
- Math & Logic: Word problems, puzzles, navigation tasks.
- Complex Planning: "How do I move this couch through this door?"
- Legal/Medical reasoning: Deriving a conclusion from a set of rules and facts.
- Debugging: Asking the model to explain why code is failing before fixing it.
Use Cases
- Zero-Shot CoT: Simply adding "Let's think step by step" to the prompt.
- Few-Shot CoT: Providing examples of [Question -> Reasoning -> Answer] to guide the model.
- Self-Consistency: Generating 5 different Chain-of-Thought paths and picking the answer that appears most frequently (Majority Voting).
- Tree of Thoughts: Exploring multiple possible reasoning branches and backtracking if one leads to a dead end.
Implementation Pattern
def chain_of_thought_prompt(question):
prompt = f"""
Question: {question}
Instruction: Answer the question by reasoning step-by-step.
Format your answer as:
Reasoning:
1. [First Step]
2. [Second Step]
...
Final Answer: [Answer]
"""
return llm.generate(prompt)