Abstract: We introduce a novel actor-critic framework that utilizes vision-language models (VLMs) and large language models (LLMs) for design concept generation, particularly for producing a diverse array of innovative solutions to ...