What is chain-of-thought (CoT) prompting?
mediumAnswer
- Instruct the model to 'think step by step' or provide few-shot examples that show intermediate reasoning before the answer.
- On multi-step tasks (math, logic, complex QA), CoT can improve accuracy 10-40% because it exploits the LLM's ability to condition on its own intermediate tokens.
- Introduced by Wei et al. (2022).
- Zero-shot CoT ('Let's think step by step') works well; few-shot CoT with worked examples works better on hardest problems.
Check yourself — multiple choice
- CoT hurts reasoning
- Ask for step-by-step intermediate reasoning → 10-40% gains on multi-step math/logic (Wei 2022)
- Only for classification
- CoT is deprecated
CoT: expose intermediate reasoning → the model conditions on it → better multi-step accuracy.
#prompting#chain-of-thought
Practise LLMs & GenAI
214 interview questions in this topic.