(논문 요약) OpenAI o1 System Card (Paper)

핵심 내용

• 학습 관련 내용
  • trained with large-scale reinforcement learning to reason using chain of thought
  • the models learn to refine their thinking process, try different strategies, and recognize their mistakes
• 데이터
  • Public Data: reasoning data, scientific literature
  • Proprietary Data from Data Partnerships: paywalled content, specialized archives, other domain-specific datasets