Continued Pretraining for Better Zero- and Few-Shot Promptability

Zhaofeng Wu, Robert L Logan IV, Pete Walsh, Akshita Bhagia, Dirk Groeneveld, Sameer Singh, Iz Beltagy


Abstract
Recently introduced language model prompting methods can achieve high accuracy in zero- and few-shot settings while requiring few to no learned task-specific parameters. Nevertheless, these methods still often trail behind full model finetuning. In this work, we investigate if a dedicated continued pretraining stage could improve “promptability”, i.e., zero-shot performance with natural language prompts or few-shot performance with prompt tuning. We reveal settings where existing continued pretraining methods lack promptability. We also identify current methodological gaps, which we fill with thorough large-scale experiments. We demonstrate that a simple recipe, continued pretraining that incorporates a trainable prompt during multi-task learning, leads to improved promptability in both zero- and few-shot settings compared to existing methods, up to 31% relative. On the other hand, we find that continued pretraining using MAML-style meta-learning, a method that directly optimizes few-shot promptability, yields subpar performance. We validate our findings with two prompt tuning methods, and, based on our results, we provide concrete recommendations to optimize promptability for different use cases.
Anthology ID:
2022.emnlp-main.300
Volume:
Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing
Month:
December
Year:
2022
Address:
Abu Dhabi, United Arab Emirates
Editors:
Yoav Goldberg, Zornitsa Kozareva, Yue Zhang
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
4517–4531
Language:
URL:
https://aclanthology.org/2022.emnlp-main.300
DOI:
10.18653/v1/2022.emnlp-main.300
Bibkey:
Cite (ACL):
Zhaofeng Wu, Robert L Logan IV, Pete Walsh, Akshita Bhagia, Dirk Groeneveld, Sameer Singh, and Iz Beltagy. 2022. Continued Pretraining for Better Zero- and Few-Shot Promptability. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 4517–4531, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
Cite (Informal):
Continued Pretraining for Better Zero- and Few-Shot Promptability (Wu et al., EMNLP 2022)
Copy Citation:
PDF:
https://preview.aclanthology.org/nschneid-patch-2/2022.emnlp-main.300.pdf