Investigating Multi-Pivot Ensembling with Massively Multilingual Machine Translation Models
This work addresses performance issues in low-resource machine translation for users of multilingual models, but it is incremental as it builds on existing pivoting strategies with a new combination method.
The paper tackled the problem of low performance in low-resource machine translation using massively multilingual models by investigating multi-pivot ensembling strategies, finding that a novel method called MaxEns improved translation quality and reduced hallucination compared to direct translation and averaging approaches, though it still lagged behind using English as a single pivot on average.
Massively multilingual machine translation models allow for the translation of a large number of languages with a single model, but have limited performance on low- and very-low-resource translation directions. Pivoting via high-resource languages remains a strong strategy for low-resource directions, and in this paper we revisit ways of pivoting through multiple languages. Previous work has used a simple averaging of probability distributions from multiple paths, but we find that this performs worse than using a single pivot, and exacerbates the hallucination problem because the same hallucinations can be probable across different paths. We also propose MaxEns, a novel combination strategy that makes the output biased towards the most confident predictions, hypothesising that confident predictions are less prone to be hallucinations. We evaluate different strategies on the FLORES benchmark for 20 low-resource language directions, demonstrating that MaxEns improves translation quality for low-resource languages while reducing hallucination in translations, compared to both direct translation and an averaging approach. On average, multi-pivot strategies still lag behind using English as a single pivot language, raising the question of how to identify the best pivoting strategy for a given translation direction.