GreenPLM: Cross-Lingual Transfer of Monolingual Pre-Trained Language Models at Almost No Cost
This work addresses cross-linguistic access and energy sustainability in NLP, reducing inequalities for speakers of low-resource languages, though it is incremental as it builds on existing translation heuristics.
The study tackled the problem of high training costs and limited data for pre-trained language models in low-resource languages by proposing GreenPLM, a framework that uses bilingual lexicons to translate models between languages with minimal cost, achieving performance comparable to or better than expensive methods and outperforming original monolingual models in six out of seven languages with up to 200x less pre-training effort.
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}