Domain-Specific Translation with Open-Source Large Language Models: Resource-Oriented Analysis
This work addresses the problem of achieving high-quality domain-specific translation for researchers and practitioners, especially in medium- and low-resource settings, by highlighting the limitations of current LLMs and the need for specialized models.
The study compared open-source large language models (LLMs) to specialized machine translation models for domain-specific translation in the medical field, finding that NLLB-200 3.3B outperformed LLMs in the 7-8B parameter range across most language directions, with fine-tuning not closing the gap.
In this work, we compare the domain-specific translation performance of open-source autoregressive decoder-only large language models (LLMs) with task-oriented machine translation (MT) models. Our experiments focus on the medical domain and cover four language directions with varied resource availability: English-to-French, English-to-Portuguese, English-to-Swahili, and Swahili-to-English. Despite recent advancements, LLMs demonstrate a significant quality gap in specialized translation compared to multilingual encoder-decoder MT models such as NLLB-200. Our results indicate that NLLB-200 3.3B outperforms all evaluated LLMs in the 7-8B parameter range across three out of the four language directions. While fine-tuning improves the performance of LLMs such as Mistral and Llama, these models still underperform compared to fine-tuned NLLB-200 3.3B models. Our findings highlight the ongoing need for specialized MT models to achieve high-quality domain-specific translation, especially in medium-resource and low-resource settings. Moreover, the superior performance of larger LLMs over their 8B variants suggests potential value in pre-training domain-specific medium-sized language models, employing targeted data selection and knowledge distillation approaches to enhance both quality and efficiency in specialized translation tasks.