Causes and Cures for Interference in Multilingual Translation

TL;DR

This paper investigates the factors influencing interference in multilingual machine translation, revealing how model size, data proportion, and sampling strategies affect performance and proposing methods to mitigate interference.

Contribution

It systematically identifies key factors affecting interference and demonstrates how adjusting model size and data sampling can reduce interference and improve translation quality.

Findings

Interference is mainly caused by small model size relative to data.

Transformer models under one billion parameters largely reduce interference.

Tuning sampling temperature balances interference across language pairs.

Abstract

Multilingual machine translation models can benefit from synergy between different language pairs, but also suffer from interference. While there is a growing number of sophisticated methods that aim to eliminate interference, our understanding of interference as a phenomenon is still limited. This work identifies the main factors that contribute to interference in multilingual machine translation. Through systematic experimentation, we find that interference (or synergy) are primarily determined by model size, data size, and the proportion of each language pair within the total dataset. We observe that substantial interference occurs mainly when the model is very small with respect to the available training data, and that using standard transformer configurations with less than one billion parameters largely alleviates interference and promotes synergy. Moreover, we show that tuning the sampling temperature to control the proportion of each language pair in the data is key to balancing the amount of interference between low and high resource language pairs effectively, and can lead to superior performance overall.