이 논문은 과거 추론 궤적을 활용하여 문제 해결 능력을 향상시키는 State Machine of Thoughts (SMoT) 패러다임을 소개합니다.
The author argues that by utilizing a state machine to record past reasoning trajectories, the proposed State Machine of Thoughts (SMoT) can significantly enhance problem-solving abilities. SMoT selects optimal sub-solutions based on past experiences, improving efficiency and accuracy in problem-solving.