Cold-Start Data Selection for Few-shot Language Model Fine-tuning: A Prompt-Based Uncertainty Propagation Approach
This addresses the challenge of efficient data selection for few-shot learning in NLP, which is incremental as it builds on existing methods for uncertainty estimation and diversity promotion.
The paper tackles the problem of selecting data for few-shot fine-tuning of language models in cold-start scenarios without initial labeled data, proposing PATRON which uses prompt-based uncertainty estimation and a partition-then-rewrite strategy, achieving up to 6.9% improvement over baselines and reaching 91.0-92.1% of fully supervised performance with only 128 labels.
Large Language Models have demonstrated remarkable few-shot performance, but the performance can be sensitive to the selection of few-shot instances. We propose PATRON, a new method that uses prompt-based uncertainty estimation for data selection for pre-trained language model fine-tuning under cold-start scenarios, i.e., no initial labeled data are available. In PATRON, we design (1) a prompt-based uncertainty propagation approach to estimate the importance of data points and (2) a partition-then-rewrite (PTR) strategy to promote sample diversity when querying for annotations. Experiments on six text classification datasets show that PATRON outperforms the strongest cold-start data selection baselines by up to 6.9%. Besides, with 128 labels only, PATRON achieves 91.0% and 92.1% of the fully supervised performance based on vanilla fine-tuning and prompt-based learning respectively. Our implementation of PATRON is available at \url{https://github.com/yueyu1030/Patron}.