The VolcTrans System for WMT22 Multilingual Machine Translation Task

TL;DR

The paper presents VolcTrans, a transformer-based multilingual machine translation system for WMT22, utilizing diverse external data sources and heuristic cleaning to achieve competitive translation quality.

Contribution

Introduces a large-scale multilingual transformer model trained on multiple data sources, including external corpora and pseudo bitext, with heuristic cleaning for improved translation performance.

Findings

Achieved an average BLEU of 17.3 across language pairs.

Maintained an inference speed of 11.5 sentences/sec on a single GPU.

Demonstrated effectiveness of external data and data cleaning in multilingual translation.

Abstract

This report describes our VolcTrans system for the WMT22 shared task on large-scale multilingual machine translation. We participated in the unconstrained track which allows the use of external resources. Our system is a transformerbased multilingual model trained on data from multiple sources including the public training set from the data track, NLLB data provided by Meta AI, self-collected parallel corpora, and pseudo bitext from back-translation. A series of heuristic rules clean both bilingual and monolingual texts. On the official test set, our system achieves 17.3 BLEU, 21.9 spBLEU, and 41.9 chrF2++ on average over all language pairs. The average inference speed is 11.5 sentences per second using a single Nvidia Tesla V100 GPU. Our code and trained models are available at https://github.com/xian8/wmt22