Prompting or training technique where an LLM produces intermediate reasoning steps before the final answer, dramatically improving accuracy on multi-step problems.
"Adding ‘think step by step' enables chain of thought and fixed the math errors."
No comments yet — say something.
Add your own interpretation of "chain of thought".