Tree of Thoughts - Deliberate Problem Solving with Large Language Models
Offered By: Yannic Kilcher via YouTube
Course Description
Overview
Explore a comprehensive review of the "Tree of Thoughts" paper, which introduces a novel framework for improving large language model problem-solving capabilities. Learn how this approach generalizes the Chain-of-Thought prompting method by incorporating tree search, state evaluation, and backtracking. Discover the significant improvements demonstrated in experiments on toy tasks compared to classic and Chain-of-Thought prompting. Delve into the formalization of the algorithm, its applications in games like Game of 24 and creative writing tasks, and its potential as a general problem solver. Examine the ablation studies and draw insights from the conclusion of this groundbreaking research in AI and prompt engineering.
Syllabus
- Introduction
- From Chain-of-Thought to Tree-of-Thought
- Formalizing the algorithm
- Game of 24 & Creative writing
- Crosswords
- Is this a general problem solver?
- Ablation studies
- Conclusion
Taught by
Yannic Kilcher
Related Courses
Fantasy and Science Fiction: The Human Mind, Our Modern WorldUniversity of Michigan via Coursera Songwriting: Writing the Lyrics
Berklee College of Music via Coursera Plagues, Witches, and War: The Worlds of Historical Fiction
University of Virginia via Coursera Art & Activity: Interactive Strategies for Engaging with Art
The Museum of Modern Art via Coursera The Rise of Superheroes and Their Impact On Pop Culture
Smithsonian Institution via edX