We investigate the use of multimodal information contained in images as an effective method for enhancing the commonsense of Transformer models for text generation. We perform experiments using BART and T5 on concept-to-text generation, specifically the task of generative commonsense reasoning, or CommonGen. We call our approach VisCTG: Visually Grounded Concept-to-Text Generation. VisCTG involves captioning images representing appropriate everyday scenarios, and using these captions to enrich and steer the generation process. Comprehensive evaluation and analysis demonstrate that VisCTG noticeably improves model performance while successfully addressing several issues of the baseline generations, including poor commonsense, fluency, and specificity.
@article{arxiv.2109.03892,
title = {Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models},
author = {Steven Y. Feng and Kevin Lu and Zhuofu Tao and Malihe Alikhani and Teruko Mitamura and Eduard Hovy and Varun Gangal},
journal= {arXiv preprint arXiv:2109.03892},
year = {2022}
}
Comments
Accepted to AAAI 2022. Code at https://github.com/styfeng/VisCTG