Data Processing Matters: SRPH-Konvergen AI's Machine Translation System for WMT'21
This work addresses machine translation performance for multilingual tasks, showing that data preprocessing can be as effective as advanced architectures, though it is incremental in approach.
The paper tackled machine translation for the WMT'21 task by using a standard Transformer model with enhanced data preprocessing, achieving an average BLEU score of 22.97 on the test set and ranking first in Indonesian to Javanese translation.
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.