Summary of When Is Tree Search Useful For Llm Planning? It Depends on the Discriminator, by Ziru Chen et al.
When is Tree Search Useful for LLM Planning? It Depends on the Discriminator
by Ziru Chen, Michael White, Raymond Mooney, Ali Payani, Yu Su, Huan Sun
First submitted to arxiv on: 16 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A novel framework for solving multi-step problems using large language models (LLMs) is proposed, consisting of a generator, discriminator, and planning method. The framework’s efficacy is evaluated through the application of two advanced planning methods: iterative correction and tree search. Comparative experiments on text-to-SQL parsing and mathematical reasoning tasks reveal that high-discrimination accuracy (>90%) is required for these methods to outperform simpler re-ranking approaches. Current LLMs’ discrimination capabilities fall short of meeting this requirement, limiting their potential in real-world applications. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Large language models are super smart computers that can solve problems step by step. This paper looks at how they work and tries to make them even better. They use a special system with three parts: one that creates ideas, one that checks if those ideas are good, and one that helps figure out the best plan. The researchers tested two fancy methods to see if they could help or hurt the language models’ ability to solve problems. They found that for these methods to work well, the language model needs to be super good at checking ideas (at least 90% accurate). Unfortunately, current language models aren’t quite there yet. |
Keywords
* Artificial intelligence * Language model * Parsing
