English

Improving Language Plasticity via Pretraining with Active Forgetting

Computation and Language 2024-01-15 v3 Machine Learning Neural and Evolutionary Computing

Abstract

Pretrained language models (PLMs) are today the primary model for natural language processing. Despite their impressive downstream performance, it can be difficult to apply PLMs to new languages, a barrier to making their capabilities universally accessible. While prior work has shown it possible to address this issue by learning a new embedding layer for the new language, doing so is both data and compute inefficient. We propose to use an active forgetting mechanism during pretraining, as a simple way of creating PLMs that can quickly adapt to new languages. Concretely, by resetting the embedding layer every K updates during pretraining, we encourage the PLM to improve its ability of learning new embeddings within a limited number of updates, similar to a meta-learning effect. Experiments with RoBERTa show that models pretrained with our forgetting mechanism not only demonstrate faster convergence during language adaptation but also outperform standard ones in a low-data regime, particularly for languages that are distant from English.

Keywords

Cite

@article{arxiv.2307.01163,
  title  = {Improving Language Plasticity via Pretraining with Active Forgetting},
  author = {Yihong Chen and Kelly Marchisio and Roberta Raileanu and David Ifeoluwa Adelani and Pontus Stenetorp and Sebastian Riedel and Mikel Artetxe},
  journal= {arXiv preprint arXiv:2307.01163},
  year   = {2024}
}

Comments

NeurIPS 2023 Final Version

R2 v1 2026-06-28T11:20:58.686Z