WARP: Word-level Adversarial ReProgramming

TL;DR

This paper introduces WARP, an adversarial reprogramming method that learns task-specific word embeddings to adapt pretrained language models for various NLP tasks, outperforming existing transfer learning approaches especially in low-data scenarios.

Contribution

WARP presents a novel adversarial reprogramming technique that uses minimal trainable parameters to effectively repurpose pretrained language models for multiple NLP tasks.

Findings

Outperforms existing methods on GLUE benchmark with fewer parameters.

Effective in few-shot learning, surpassing GPT-3 on SuperGLUE with only 32 samples.

Uses up to 25K trainable parameters per task.

Abstract

Transfer learning from pretrained language models recently became the dominant approach for solving many NLP tasks. A common approach to transfer learning for multiple tasks that maximize parameter sharing trains one or more task-specific layers on top of the language model. In this paper, we present an alternative approach based on adversarial reprogramming, which extends earlier work on automatic prompt generation. Adversarial reprogramming attempts to learn task-specific word embeddings that, when concatenated to the input text, instruct the language model to solve the specified task. Using up to 25K trainable parameters per task, this approach outperforms all existing methods with up to 25M trainable parameters on the public leaderboard of the GLUE benchmark. Our method, initialized with task-specific human-readable prompts, also works in a few-shot setting, outperforming GPT-3 on two SuperGLUE tasks with just 32 training samples.