From LLM to NMT: Advancing Low-Resource Machine Translation with Claude
This work addresses low-resource machine translation for languages like Yoruba, though it is incremental as it builds on existing LLM and distillation methods.
The paper tackles low-resource machine translation by showing that Claude 3 Opus outperforms other LLMs in translation, with knowledge distillation from Claude advancing state-of-the-art in Yoruba-English translation, surpassing baselines like NLLB-54B and Google Translate.
We show that Claude 3 Opus, a large language model (LLM) released by Anthropic in March 2024, exhibits stronger machine translation competence than other LLMs. Though we find evidence of data contamination with Claude on FLORES-200, we curate new benchmarks that corroborate the effectiveness of Claude for low-resource machine translation into English. We find that Claude has remarkable \textit{resource efficiency} -- the degree to which the quality of the translation model depends on a language pair's resource level. Finally, we show that advancements in LLM translation can be compressed into traditional neural machine translation (NMT) models. Using Claude to generate synthetic data, we demonstrate that knowledge distillation advances the state-of-the-art in Yoruba-English translation, meeting or surpassing strong baselines like NLLB-54B and Google Translate.