MORE: Multi-mOdal REtrieval Augmented Generative Commonsense Reasoning
This addresses the challenge of enhancing commonsense reasoning in AI systems, though it appears incremental by extending existing retrieval methods to include images.
The paper tackles the problem of language models lacking sufficient commonsense knowledge by proposing a multi-modal retrieval augmentation framework that uses both text and images, achieving improved performance on the Common-Gen task.
Since commonsense information has been recorded significantly less frequently than its existence, language models pre-trained by text generation have difficulty to learn sufficient commonsense knowledge. Several studies have leveraged text retrieval to augment the models' commonsense ability. Unlike text, images capture commonsense information inherently but little effort has been paid to effectively utilize them. In this work, we propose a novel Multi-mOdal REtrieval (MORE) augmentation framework, to leverage both text and images to enhance the commonsense ability of language models. Extensive experiments on the Common-Gen task have demonstrated the efficacy of MORE based on the pre-trained models of both single and multiple modalities.