The Unreasonable Volatility of Neural Machine Translation Models

TL;DR

This paper investigates the unexpected and volatile translation behaviors of neural machine translation models, revealing that minor input modifications can cause significant translation changes, exposing generalization issues.

Contribution

It uncovers the extent of translation volatility in NMT models and highlights the need to understand and address their stability problems.

Findings

RNN models show 26% volatility in translations.

Transformer models show 19% volatility.

Minor input changes can cause mistranslations.

Abstract

Recent works have shown that Neural Machine Translation (NMT) models achieve impressive performance, however, questions about understanding the behavior of these models remain unanswered. We investigate the unexpected volatility of NMT models where the input is semantically and syntactically correct. We discover that with trivial modifications of source sentences, we can identify cases where \textit{unexpected changes} happen in the translation and in the worst case lead to mistranslations. This volatile behavior of translating extremely similar sentences in surprisingly different ways highlights the underlying generalization problem of current NMT models. We find that both RNN and Transformer models display volatile behavior in 26% and 19% of sentence variations, respectively.