Chain of Thought
Domain: LLM Concepts
🎯 Motivation
For complex reasoning, asking the model to think step by step produces better answers than asking for a direct answer.
📋 Concrete Example
Direct: ‘What is 15% of 240?’ might give wrong answer. CoT: ‘Let me calculate: 10% of 240 = 24, 5% = 12, so 15% = 36.’ Breaking it into steps leads to the correct answer.
🔗 Analogy
Showing your work on a math exam - instead of writing just the final answer, you show each step. This helps the teacher (and you) catch mistakes. CoT makes the model show its work.