Abstract: Prompt-based learning has demonstrated remarkable success in few-shot text classification, outperforming the traditional fine-tuning approach. This method transforms a text input into a ...

Abstract: Generating photo-realistic images from labels (e.g., semantic labels or sketch labels) is much more challenging than the general image-to-image translation task, mainly due to the large ...