Data Processing Matters: SRPH-Konvergen AI's Machine Translation System for WMT'21

TL;DR

This paper demonstrates that effective data preprocessing can significantly enhance machine translation performance, achieving top rankings in specific language pairs without advanced model modifications.

Contribution

The study shows that simple Transformer models combined with strong data preprocessing can outperform more complex approaches in multilingual translation tasks.

Findings

Achieved 22.97 BLEU on the WMT'21 test set

Ranked first in Indonesian to Javanese translation

Outperformed models with advanced architectures using data preprocessing

Abstract

In this paper, we describe the submission of the joint Samsung Research Philippines-Konvergen AI team for the WMT'21 Large Scale Multilingual Translation Task - Small Track 2. We submit a standard Seq2Seq Transformer model to the shared task without any training or architecture tricks, relying mainly on the strength of our data preprocessing techniques to boost performance. Our final submission model scored 22.92 average BLEU on the FLORES-101 devtest set, and scored 22.97 average BLEU on the contest's hidden test set, ranking us sixth overall. Despite using only a standard Transformer, our model ranked first in Indonesian to Javanese, showing that data preprocessing matters equally, if not more, than cutting edge model architectures and training techniques.