GreenPLM: Cross-Lingual Transfer of Monolingual Pre-Trained Language Models at Almost No Cost

TL;DR

GreenPLM introduces an energy-efficient method to transfer monolingual pre-trained language models across languages using bilingual lexicons, significantly reducing costs and energy consumption while improving performance in low-resource settings.

Contribution

The paper presents a novel framework that enables cross-lingual transfer of pre-trained models with minimal additional cost and energy, outperforming original models in several languages.

Findings

Comparable or better performance than high-cost heuristics

Outperforms original models with up to 200x less pre-training

Reduces language inequalities and energy consumption

Abstract

Large pre-trained models have revolutionized natural language processing (NLP) research and applications, but high training costs and limited data resources have prevented their benefits from being shared equally amongst speakers of all the world's languages. To address issues of cross-linguistic access to such models and reduce energy consumption for sustainability during large-scale model training, this study proposes an effective and energy-efficient framework called GreenPLM that uses bilingual lexicons to directly "translate" pre-trained language models of one language into another at almost no additional cost. We validate this approach in 18 languages' BERT models and show that this framework is comparable to, if not better than, other heuristics with high training costs. In addition, given lightweight continued pre-training on limited data where available, this framework outperforms the original monolingual language models in six out of seven tested languages with up to 200x less pre-training efforts. Aiming at the Leave No One Behind Principle (LNOB), our approach manages to reduce inequalities between languages and energy consumption greatly. We make our codes and models publicly available here: \url{https://github.com/qcznlp/GreenPLMs}