BJTU-WeChat's Systems for the WMT22 Chat Translation Task

TL;DR

This paper presents a high-performing chat translation system for English-German using Transformer variants, pre-training, data augmentation, speaker adaptation, and ensemble techniques, achieving top COMET scores at WMT'22.

Contribution

It introduces a comprehensive system combining multiple advanced techniques for chat translation, achieving the highest COMET scores among WMT'22 submissions.

Findings

Achieved top COMET scores of 0.810 and 0.946 for English-German and German-English.

Effective use of data filtering, synthetic data generation, and speaker adaptation.

Ensemble methods significantly improved translation quality.

Abstract

This paper introduces the joint submission of the Beijing Jiaotong University and WeChat AI to the WMT'22 chat translation task for English-German. Based on the Transformer, we apply several effective variants. In our experiments, we utilize the pre-training-then-fine-tuning paradigm. In the first pre-training stage, we employ data filtering and synthetic data generation (i.e., back-translation, forward-translation, and knowledge distillation). In the second fine-tuning stage, we investigate speaker-aware in-domain data generation, speaker adaptation, prompt-based context modeling, target denoising fine-tuning, and boosted self-COMET-based model ensemble. Our systems achieve 0.810 and 0.946 COMET scores. The COMET scores of English-German and German-English are the highest among all submissions.