Visualizing attention zones in machine reading comprehension models

TL;DR

This paper presents a visualization pipeline for attention zones in machine reading comprehension models, enhancing explainability by showing how attention operates across different layers of pretrained language models.

Contribution

The paper introduces a generalizable protocol and code for visualizing attention zones in MRC models, aiding interpretability of attention mechanisms.

Findings

Visualizes attention zones across layers in MRC models

Provides a protocol applicable to various pretrained language models

Enhances understanding of model decision processes

Abstract

The attention mechanism plays an important role in the machine reading comprehension (MRC) model. Here, we describe a pipeline for building an MRC model with a pretrained language model and visualizing the effect of each attention zone in different layers, which can indicate the explainability of the model. With the presented protocol and accompanying code, researchers can easily visualize the relevance of each attention zone in the MRC model. This approach can be generalized to other pretrained language models.