Active Finetuning: Exploiting Annotation Budget in the Pretraining-Finetuning Paradigm

Open in new window