How would you do few-shot text classification with huggingface models?

Question

I need some help regarding few-shot learning for text classification. I have a working code for zero-shot classification using the zero-shot-classification pipe