CLOct 19, 2022

Continued Pretraining for Better Zero- and Few-Shot Promptability

AI2MIT
arXiv:2210.10258v2294 citationsh-index: 31
Originality Incremental advance
AI Analysis

This work addresses the gap in promptability for natural language processing tasks, offering incremental improvements for researchers and practitioners using prompt-based methods.

The authors tackled the problem of improving language model performance in zero- and few-shot settings through continued pretraining, finding that a simple multi-task learning approach with trainable prompts increased promptability by up to 31% relative compared to existing methods.

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.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes