Summary of A Question on the Explainability Of Large Language Models and the Word-level Univariate First-order Plausibility Assumption, by Jeremie Bogaert et al.
A Question on the Explainability of Large Language Models and the Word-Level Univariate First-Order Plausibility Assumption
by Jeremie Bogaert, Francois-Xavier Standaert
First submitted to arxiv on: 15 Mar 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
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 The proposed characterization questions the possibility of providing simple and informative explanations for large language models. The study defines statistical concepts such as signal, noise, and signal-to-noise ratio to analyze word-level univariate explanations. Results show that feature-based models carry more signal and less noise compared to transformer models. The paper discusses alternative definitions of signal and noise to capture complex explanations while considering plausibility for readers. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research investigates how well large language models can be explained. It creates new ways to measure the quality of these explanations by looking at the “signal” (important information) and “noise” (unimportant information). The study finds that some types of models are better at providing clear explanations than others. It also explores how we might improve explanation quality while making sure it’s easy for people to understand. |
Keywords
» Artificial intelligence » Transformer
