[chheplo]

DALL-E is for text-to-image, not text-to-art. This experiment is valid in terms of benchmarking the extent to which LLM understands a series of texts. For an AI researcher, this gives more fuel for the next iteration.