Retrieve, Caption, Generate: Visual Grounding For Enhancing Commonsense In Text Generation Models

Feng Steven Y., Lu Kevin, Tao Zhuofu, Alikhani Malihe, Mitamura Teruko, Hovy Eduard, Gangal Varun. Arxiv 2021

[Paper]    
Applications Language Modeling Model Architecture Multimodal Models Pretraining Methods Transformer

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.

Similar Work