Neural conditional language generation models achieve the state-of-the-art in Neural Machine Translation (NMT) but are highly dependent on the quality of parallel training dataset. When trained on low-quality datasets, these models are prone to various error types, including hallucinations, i.e. outputs that are fluent, but unrelated to the source sentences. These errors are particularly dangerous, because on the surface the translation can be perceived as a correct output, especially if the reader does not understand the source language. We present a case study focusing on model understanding and regularisation to reduce hallucinations in NMT. We first use feature attribution methods to study the behaviour of an NMT model that produces hallucinations. We then leverage these methods to propose a novel loss function that substantially helps reduce hallucinations and does not require retraining the model from scratch.

翻译：神经条件语言生成模型在神经机器翻译（NMT）中达到了最先进水平，但高度依赖平行训练数据集的质量。当在低质量数据集上训练时，这些模型容易出现各种错误类型，包括幻觉——即输出流畅但与源句子无关的译文。这类错误尤为危险，因为从表面看，这些翻译可能被认为是正确输出，尤其是当读者不理解源语言时。我们开展了一项聚焦于模型理解与正则化的案例研究，以减少NMT中的幻觉。首先，我们使用特征归因方法分析产生幻觉的NMT模型的行为。随后，我们利用这些方法提出一种新型损失函数，该函数能显著帮助减少幻觉，且无需从头重新训练模型。