Teaching Large Language Models an Unseen Language on the Fly
This addresses the challenge of linguistic diversity preservation by allowing LLMs to learn unseen languages on the fly, though it is incremental as it builds on existing in-context learning methods.
The paper tackles the problem of enabling large language models to support low-resource languages with minimal data by proposing DiPMT++, a framework that uses in-context learning with a dictionary and 5K parallel sentences. It significantly improves GPT-4's performance from 0 to 16 BLEU for Chinese-to-Zhuang translation and achieves 32 BLEU for Zhuang-to-Chinese translation, validated on another unseen language.
Existing large language models struggle to support numerous low-resource languages, particularly the extremely low-resource ones, for which there is minimal training data available for effective parameter updating. We thus investigate whether LLMs can learn a new language on the fly solely through prompting. To study this question, we collect a research suite for Zhuang, a language supported by no LLMs currently. We introduce DiPMT++, a framework for adapting LLMs to unseen languages by in-context learning. Using a dictionary and 5K parallel sentences only, DiPMT++ significantly enhances the performance of GPT-4 from 0 to 16 BLEU for Chinese-to-Zhuang translation and achieves 32 BLEU for Zhuang-to-Chinese translation. We also validate the effectiveness of our framework on Kalamang, another unseen language. Furthermore, we demonstrate the practical utility of DiPMT++ in aiding humans in translating completely unseen languages, which could contribute to the preservation of linguistic diversity.