Abstract: Knowledge-based visual question answering (VQA) requires external knowledge beyond the image to answer the question ... Recent works have resorted to using a powerful large language model ...

Read instructions and answer in context Both Chan and Lee ... This can appear in many forms, such as figurative language -similes and metaphors can reveal a person’s perspective.