The Fine-Tuning Paradox: Boosting Translation Quality Without Sacrificing LLM Abilities

TL;DR

This paper investigates how fine-tuning large language models for translation improves quality but can degrade other valuable abilities, and proposes strategies to preserve these skills while enhancing translation performance.

Contribution

It provides an extensive evaluation of fine-tuning effects on LLMs' translation and other behaviors, highlighting the importance of data strategies to maintain LLM capabilities.

Findings

Fine-tuning improves translation quality but reduces steerability and document-level translation abilities.

Including monolingual data in fine-tuning preserves LLM abilities while improving translation.

Fine-tuning results in less literal translations, indicating a shift in translation style.

Abstract

Fine-tuning large language models (LLMs) for machine translation has shown improvements in overall translation quality. However, it is unclear what is the impact of fine-tuning on desirable LLM behaviors that are not present in neural machine translation models, such as steerability, inherent document-level translation abilities, and the ability to produce less literal translations. We perform an extensive translation evaluation on the LLaMA and Falcon family of models with model size ranging from 7 billion up to 65 billion parameters. Our results show that while fine-tuning improves the general translation quality of LLMs, several abilities degrade. In particular, we observe a decline in the ability to perform formality steering, to produce technical translations through few-shot examples, and to perform document-level translation. On the other hand, we observe that the model produces less literal translations after fine-tuning on parallel data. We show that by including monolingual data as part of the fine-tuning data we can maintain the abilities while simultaneously enhancing overall translation quality. Our findings emphasize the need for fine-tuning strategies that preserve the benefits of LLMs for machine translation.