Your Pretrained Model Tells the Difficulty Itself: A Self-Adaptive Curriculum Learning Paradigm for Natural Language Understanding

TL;DR

This paper introduces a self-adaptive curriculum learning method for NLP that uses pre-trained language models to predict example difficulty, enabling more effective training strategies and improving model performance.

Contribution

It proposes a novel curriculum learning paradigm where difficulty scores are predicted by the model itself, replacing manual difficulty metrics, and explores various training orderings.

Findings

Faster convergence in training.

Improved accuracy on NLU tasks.

Effective across multiple datasets.

Abstract

Curriculum learning is a widely adopted training strategy in natural language processing (NLP), where models are exposed to examples organized by increasing difficulty to enhance learning efficiency and performance. However, most existing approaches rely on manually defined difficulty metrics -- such as text length -- which may not accurately reflect the model's own perspective. To overcome this limitation, we present a self-adaptive curriculum learning paradigm that prioritizes fine-tuning examples based on difficulty scores predicted by pre-trained language models (PLMs) themselves. Building on these scores, we explore various training strategies that differ in the ordering of examples for the fine-tuning: from easy-to-hard, hard-to-easy, to mixed sampling. We evaluate our method on four natural language understanding (NLU) datasets covering both binary and multi-class classification tasks. Experimental results show that our approach leads to faster convergence and improved performance compared to standard random sampling.