Chain-of-Thought (CoT) Prompting
A prompting technique where intermediate reasoning steps are shown in few-shot examples, causing language models to generate their own step-by-step reasoning before producing a final answer.
A prompting technique where intermediate reasoning steps are shown in few-shot examples, causing language models to generate their own step-by-step reasoning before producing a final answer. Instead of providing just (question, answer) pairs, CoT examples show (question, reasoning steps, answer) triples. This technique is most effective for large models (100B+ parameters) on multi-step reasoning tasks like math and logic.
Solving problems by writing out reasoning step-by-step in natural language. Works well but requires humans to verify correctness.
Getting the model to reason step by step before answering, improving accuracy.
Asking the model to reason step by step before giving its final answer, which improves accuracy on harder problems.