VERBA: Verbalizing Model Differences Using Large Language Models

TL;DR

VERBA uses large language models to generate verbal descriptions of differences between machine learning models, aiding in model comparison and understanding.

Contribution

This paper introduces VERBA, a novel LLM-based method for verbalizing pairwise model differences, including a protocol for evaluating verbalization informativeness and a diverse benchmark suite.

Findings

VERBA achieves up to 80% accuracy in verbalizing model differences.

Including structural information increases verbalization accuracy to 90%.

VERBA facilitates fine-grained, post-hoc model comparison and transparency.

Abstract

In the current machine learning landscape, we face a "model lake" phenomenon: Given a task, there is a proliferation of trained models with similar performances despite different behavior. For model users attempting to navigate and select from the models, documentation comparing model pairs is helpful. However, for every $N$ models there could be $O(N^2)$ pairwise comparisons, a number prohibitive for the model developers to manually perform pairwise comparisons and prepare documentations. To facilitate fine-grained pairwise comparisons among models, we introduced $\textbf{VERBA}$. Our approach leverages a large language model (LLM) to generate verbalizations of model differences by sampling from the two models. We established a protocol that evaluates the informativeness of the verbalizations via simulation. We also assembled a suite with a diverse set of commonly used machine learning models as a benchmark. For a pair of decision tree models with up to 5% performance difference but 20-25% behavioral differences, $\textbf{VERBA}$ effectively verbalizes their variations with up to 80% overall accuracy. When we included the models' structural information, the verbalization's accuracy further improved to 90%. $\textbf{VERBA}$ opens up new research avenues for improving the transparency and comparability of machine learning models in a post-hoc manner.