Parameter- and data-efficient fine-tuning of large language models (keywords: few-shot learning, meta-learning, active-learning).