ImaginE: An Imagination-Based Automatic Evaluation Metric for Natural Language Generation

TL;DR

ImaginE introduces a novel evaluation metric for natural language generation that leverages image generation to incorporate visual imagination, enhancing correlation with human judgments.

Contribution

This work presents a new multi-modal evaluation approach using StableDiffusion to generate images from text, improving automatic NLG evaluation metrics.

Findings

ImaginE improves correlation with human judgments in NLG evaluation.

Adding generated images enhances existing automatic metrics.

The method works for both reference-based and reference-free scenarios.

Abstract

Automatic evaluations for natural language generation (NLG) conventionally rely on token-level or embedding-level comparisons with text references. This differs from human language processing, for which visual imagination often improves comprehension. In this work, we propose ImaginE, an imagination-based automatic evaluation metric for natural language generation. With the help of StableDiffusion, a state-of-the-art text-to-image generator, we automatically generate an image as the embodied imagination for the text snippet and compute the imagination similarity using contextual embeddings. Experiments spanning several text generation tasks demonstrate that adding machine-generated images with our ImaginE displays great potential in introducing multi-modal information into NLG evaluation, and improves existing automatic metrics' correlations with human similarity judgments in both reference-based and reference-free evaluation scenarios.