Summary of Multi-level Explanations For Generative Language Models, by Lucas Monteiro Paes et al.
Multi-Level Explanations for Generative Language Models
by Lucas Monteiro Paes, Dennis Wei, Hyo Jin Do, Hendrik Strobelt, Ronny Luss, Amit Dhurandhar, Manish Nagireddy, Karthikeyan Natesan Ramamurthy, Prasanna Sattigeri, Werner Geyer, Soumya Ghosh
First submitted to arxiv on: 21 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 This work extends perturbation-based explanation methods, such as LIME and SHAP, to generative language models. To address the challenges of text output and long input texts, a general framework called MExGen is proposed, which can be instantiated with different attribution algorithms. The framework introduces scalarizers for mapping text to real numbers to handle text output and takes a multi-level approach to handle long inputs. A systematic evaluation of perturbation-based attribution methods is conducted for summarization and context-grounded question answering tasks. The results show that MExGen can provide more locally faithful explanations of generated outputs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us understand how artificial intelligence models, like language generators, make predictions. Right now, we don’t fully know why these models generate certain texts or answers. This research proposes a new way to explain the decisions made by these models. It’s like trying to figure out why someone wrote a specific sentence or answered a question in a particular way. The method is designed to work with different types of language generation tasks and can help us better understand how these models make predictions. |
Keywords
» Artificial intelligence » Question answering » Summarization
