Summary of Pointwise Mutual Information As a Performance Gauge For Retrieval-augmented Generation, by Tianyu Liu et al.
Pointwise Mutual Information as a Performance Gauge for Retrieval-Augmented Generation
by Tianyu Liu, Jirui Qi, Paul He, Arianna Bisazza, Mrinmaya Sachan, Ryan Cotterell
First submitted to arxiv on: 12 Nov 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 Medium Difficulty summary: Recent research has shown that large language models can be influenced by the order in which retrieved documents are presented when solving tasks like question answering (QA). To address this, we propose a method that uses pointwise mutual information (PMI) as an effective gauge for language model performance. Our experiments on two QA datasets and various large language models demonstrate a correlation between answer accuracy and PMI. We also develop two methods that leverage PMI to select and construct prompts, resulting in improved performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Low Difficulty summary: Have you ever wondered how big AI models can get tricked into giving wrong answers? It turns out that the order of information they receive affects their responses! To help these models give better answers, we developed a new way to measure how well they understand questions. We tested our method on two different datasets and found it works great! We also created ways to use this measurement to guide AI models towards giving more accurate answers. |
Keywords
» Artificial intelligence » Language model » Question answering
