Continual Pre-Training Mitigates Forgetting in Language and Vision

TL;DR

This paper explores continual pre-training in language and vision models, demonstrating that it reduces forgetting and that self-supervised methods outperform supervised ones in knowledge retention.

Contribution

It formalizes the continual pre-training scenario and provides empirical evidence that self-supervised pre-training enhances knowledge retention during continual learning.

Findings

Continual pre-training reduces catastrophic forgetting.

Self-supervised pre-training outperforms supervised pre-training in retaining knowledge.

Empirical results support the effectiveness of continual pre-training in language and vision models.

Abstract

Pre-trained models are nowadays a fundamental component of machine learning research. In continual learning, they are commonly used to initialize the model before training on the stream of non-stationary data. However, pre-training is rarely applied during continual learning. We formalize and investigate the characteristics of the continual pre-training scenario in both language and vision environments, where a model is continually pre-trained on a stream of incoming data and only later fine-tuned to different downstream tasks. We show that continually pre-trained models are robust against catastrophic forgetting and we provide strong empirical evidence supporting the fact that self-supervised pre-training is more effective in retaining previous knowledge than supervised protocols. Code is provided at https://github.com/AndreaCossu/continual-pretraining-nlp-vision .