Z-LaVI: Zero-Shot Language Solver Fueled by Visual Imagination

Abstract: Large-scale pretrained LLMs have made significant advances in solving downstream language understanding tasks. However, they generally suffer from reporting bias, the phenomenon describing the lack of explicit commonsense knowledge in written text, e.g., ''an orange is orange''. To overcome this limitation, we develop a novel approach, Z-LaVI, to endow LLMs with visual imagination capabilities. Specifically, we leverage two complementary types of ''imaginations'': (i) recalling existing images through retrieval and (ii) synthesizing nonexistent images via text-to-image generation. Jointly exploiting the language inputs and the imagination, a pretrained vision-LLM (e.g., CLIP) eventually composes a zero-shot solution to the original language tasks. Notably, fueling LLMs with imagination can effectively leverage visual knowledge to solve plain language tasks. In consequence, Z-LaVI consistently improves the zero-shot performance of existing LLMs across a diverse set of language tasks.