Summary of Latent Logic Tree Extraction For Event Sequence Explanation From Llms, by Zitao Song et al.
Latent Logic Tree Extraction for Event Sequence Explanation from LLMs
by Zitao Song, Chao Yang, Chaojie Wang, Bo An, Shuang Li
First submitted to arxiv on: 3 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computation and Language (cs.CL)
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 This paper proposes an efficient method to elicit logic tree-based explanations from Large Language Models (LLMs) for high-stakes systems like healthcare or robotics. The goal is to provide customized insights into event sequences generated by these systems. The approach uses a temporal point process model and the likelihood function as a score to evaluate generated logic trees. An amortized Expectation-Maximization (EM) learning framework is designed, treating the logic tree as latent variables. The method employs an LLM prior and generates logic tree samples from the posterior using a learnable GFlowNet. The M-step approximates marginalization over the posterior, facilitating the learning of model parameters and refining the tunable LLM prior parameters. In the online setting, the lightweight model iteratively extracts relevant rules from LLMs for each sequence. The framework shows promising performance and adaptability. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about a new way to use special language models to understand complex event sequences in high-stakes systems like healthcare or robotics. These events are important because they can help make better decisions or improve the efficiency of these systems. The researchers developed a method to extract the most relevant information from these events and explain why certain things happened. They used a combination of old and new ideas to create an efficient way to do this, which is important because it can be applied to many different areas where event sequences are generated. |
Keywords
» Artificial intelligence » Likelihood
