Context-Aware or Context-Insensitive? Assessing LLMs' Performance in Document-Level Translation
This work addresses the reliability of LLMs for document-level translation, which is crucial for applications requiring accurate context handling, but it is incremental as it builds on existing evaluation methods.
The study assessed large language models (LLMs) in document-level translation, finding that while LLMs outperform encoder-decoder models overall, they do not show improved performance in pronoun translation, highlighting a need for context-aware finetuning.
Large language models (LLMs) are increasingly strong contenders in machine translation. In this work, we focus on document-level translation, where some words cannot be translated without context from outside the sentence. Specifically, we investigate the ability of prominent LLMs to utilize the document context during translation through a perturbation analysis (analyzing models' robustness to perturbed and randomized document context) and an attribution analysis (examining the contribution of relevant context to the translation). We conduct an extensive evaluation across nine LLMs from diverse model families and training paradigms, including translation-specialized LLMs, alongside two encoder-decoder transformer baselines. We find that LLMs' improved document-translation performance compared to encoder-decoder models is not reflected in pronoun translation performance. Our analysis highlight the need for context-aware finetuning of LLMs with a focus on relevant parts of the context to improve their reliability for document-level translation.