Choose the Final Translation from NMT and LLM hypotheses Using MBR Decoding: HW-TSC's Submission to the WMT24 General MT Shared Task

TL;DR

By using Minimum Bayesian risk (MBR) decoding to select the final translation from multiple hypotheses for NMT and LLM-based MT models, the submission of Huawei Translate Services Center (HW-TSC) receives competitive results in the final evaluation.

Abstract

This paper presents the submission of Huawei Translate Services Center (HW-TSC) to the WMT24 general machine translation (MT) shared task, where we participate in the English to Chinese (en2zh) language pair. Similar to previous years' work, we use training strategies such as regularized dropout, bidirectional training, data diversification, forward translation, back translation, alternated training, curriculum learning, and transductive ensemble learning to train the neural machine translation (NMT) model based on the deep Transformer-big architecture. The difference is that we also use continue pre-training, supervised fine-tuning, and contrastive preference optimization to train the large language model (LLM) based MT model. By using Minimum Bayesian risk (MBR) decoding to select the final translation from multiple hypotheses for NMT and LLM-based MT models, our submission receives competitive results in the final evaluation.

Figures (4)

• Figure 1: CPT, SFT and CPO data templates used for LLM-based MT training.
• Figure 2: The overall training flow of NMT system.
• Figure 3: The training flow of LLM-based MT system.
• Figure 4: Choose the Final Translation from NMT and LLM hypotheses Using MBR Decoding.