Advancing Translation Preference Modeling with RLHF: A Step Towards Cost-Effective Solution

arXiv:2402.11525v1 Announce Type: cross
Abstract: Faithfulness, expressiveness, and elegance is the constant pursuit in machine translation. However, traditional metrics like \textit{BLEU} do not strictly align with human preference of translation quality. In this paper, we explore leveraging reinforcement learning with human feedback (\textit{RLHF}) to improve translation quality. It is non-trivial to collect a large high-quality dataset of human comparisons between translations, especially for low-resource languages. To address this issue, we propose a cost-effective preference learning strategy, optimizing reward models by …

abstract arxiv bleu cost cs.cl cs.lg explore feedback human human feedback machine machine translation metrics modeling paper quality reinforcement reinforcement learning rlhf solution translation type