Chain-of-thought is a technical term in LLMs — not literally “what it’s thinking.”
As far as I understand it, it’s a generated narration conditioned by the prompt, not direct access to internal reasoning.
Chain-of-thought is a technical term in LLMs — not literally “what it’s thinking.”
As far as I understand it, it’s a generated narration conditioned by the prompt, not direct access to internal reasoning.