Tree of Thoughts
شجرة الأفكار
أسلوب استدلال للنماذج اللغوية الكبيرة اقترحه ياو وزملاؤه في برينستون وديب مايند، يُوسِّع سلسلة الأفكار إلى شجرة يُستكشف فيها عدة مسارات فكرية مع تقييم ذاتي وتراجع، فيرفع أداء المهام التي تتطلب بحثاً وتخطيطاً كلعبة 24 وحل الكلمات المتقاطعة.
A reasoning framework for large language models proposed by Yao and colleagues at Princeton and DeepMind that generalizes chain-of-thought into a tree where multiple thought paths are explored with self-evaluation and backtracking, improving performance on tasks requiring search and planning such as Game of 24 and mini crosswords.
Also translated asTree of Thoughts، ToT، أسلوب شجرة الأفكار، الاستدلال بشجرة الأفكار
First appears in this corpus in: Mastering the Game of Go with Deep Neural Networks and Tree Search (2016)
Appears in these papers
- Mastering the Game of Go with Deep Neural Networks and Tree Search2016in the sky ✦
- Chain-of-Thought Prompting Elicits Reasoning in Large Language Models2022in the sky ✦
- Tree of Thoughts: Deliberate Problem Solving with Large Language Models2023in the sky ✦
- Tree of Thoughts: Deliberate Problem Solving with Large Language Models2023in the sky ✦