AmericasNLI: Evaluating Zero-shot Natural Language Understanding of Pretrained Multilingual Models in Truly Low-resource Languages
This addresses the challenge of AI accessibility for indigenous language communities, but it is incremental as it extends existing benchmarks and methods.
The paper tackled the problem of evaluating zero-shot natural language understanding for high-level semantic tasks in truly low-resource indigenous languages, finding that pretrained multilingual models like XLM-R performed poorly with an average accuracy of 38.62%, but translation-based methods improved this to 48.72%.
Pretrained multilingual models are able to perform cross-lingual transfer in a zero-shot setting, even for languages unseen during pretraining. However, prior work evaluating performance on unseen languages has largely been limited to low-level, syntactic tasks, and it remains unclear if zero-shot learning of high-level, semantic tasks is possible for unseen languages. To explore this question, we present AmericasNLI, an extension of XNLI (Conneau et al., 2018) to 10 indigenous languages of the Americas. We conduct experiments with XLM-R, testing multiple zero-shot and translation-based approaches. Additionally, we explore model adaptation via continued pretraining and provide an analysis of the dataset by considering hypothesis-only models. We find that XLM-R's zero-shot performance is poor for all 10 languages, with an average performance of 38.62%. Continued pretraining offers improvements, with an average accuracy of 44.05%. Surprisingly, training on poorly translated data by far outperforms all other methods with an accuracy of 48.72%.