'One size doesn't fit all': Learning how many Examples to use for In-Context Learning for Improved Text Classification

arXiv:2403.06402v1 Announce Type: cross
Abstract: Predictive models in natural language processing (NLP) have evolved from training models from scratch to fine-tuning pre-trained models with labelled data. An extreme form of this fine-tuning involves in-context learning (ICL), where the output of a pre-trained generative model (frozen decoder parameters) is controlled only with variations in the input strings (called instructions or prompts). An important component of ICL is the use of a small number of labelled data instances as examples in the …

abstract arxiv classification context cs.cl cs.lg data examples fine-tuning form in-context learning language language processing natural natural language natural language processing nlp predictive predictive models pre-trained models processing scratch text text classification training training models type