A cost-benefit analysis of cross-lingual transfer methods
This work addresses the problem of selecting cost-effective cross-lingual methods for NLP practitioners, but it is incremental as it builds on existing techniques.
The paper analyzed cross-lingual transfer methods by evaluating their effectiveness, costs, and latencies, finding that the best method is task-dependent and achieving state-of-the-art results on two out of three datasets by combining zero-shot and translation approaches.
An effective method for cross-lingual transfer is to fine-tune a bilingual or multilingual model on a supervised dataset in one language and evaluating it on another language in a zero-shot manner. Translating examples at training time or inference time are also viable alternatives. However, there are costs associated with these methods that are rarely addressed in the literature. In this work, we analyze cross-lingual methods in terms of their effectiveness (e.g., accuracy), development and deployment costs, as well as their latencies at inference time. Our experiments on three tasks indicate that the best cross-lingual method is highly task-dependent. Finally, by combining zero-shot and translation methods, we achieve the state-of-the-art in two of the three datasets used in this work. Based on these results, we question the need for manually labeled training data in a target language. Code and translated datasets are available at https://github.com/unicamp-dl/cross-lingual-analysis