This is how DeepSeek r1 reasoning model works behind the scenes. A 3-turn chain of thought reflecting on 3 questions.

→ View original post on X — @godofprompt