Learning to Few-Shot Learn Across Diverse Natural Language Classification Tasks

TL;DR

This paper introduces LEOPARD, a meta-learning method that enables transformer models to generalize to diverse NLP classification tasks with minimal labeled data, significantly improving few-shot learning performance.

Contribution

LEOPARD is a novel optimization-based meta-learning approach that handles tasks with varying class numbers, enhancing few-shot NLP classification across diverse domains.

Findings

LEOPARD outperforms baselines on 17 NLP tasks.

Achieves 14.5% relative accuracy gain with 4 examples per label.

Generalizes well to unseen tasks with minimal data.

Abstract

Self-supervised pre-training of transformer models has shown enormous success in improving performance on a number of downstream tasks. However, fine-tuning on a new task still requires large amounts of task-specific labelled data to achieve good performance. We consider this problem of learning to generalize to new tasks with few examples as a meta-learning problem. While meta-learning has shown tremendous progress in recent years, its application is still limited to simulated problems or problems with limited diversity across tasks. We develop a novel method, LEOPARD, which enables optimization-based meta-learning across tasks with different number of classes, and evaluate different methods on generalization to diverse NLP classification tasks. LEOPARD is trained with the state-of-the-art transformer architecture and shows better generalization to tasks not seen at all during training, with as few as 4 examples per label. Across 17 NLP tasks, including diverse domains of entity typing, natural language inference, sentiment analysis, and several other text classification tasks, we show that LEOPARD learns better initial parameters for few-shot learning than self-supervised pre-training or multi-task training, outperforming many strong baselines, for example, yielding 14.5% average relative gain in accuracy on unseen tasks with only 4 examples per label.