Cross-lingual Transfer Performance of mT5 with Multilingual Intermediate Tasks

Intermediate-task training---fine-tuning a pretrained model on an intermediate task before fine-tuning again on the target task---often improves model performance substantially on language understanding tasks in monolingual English settings. We investigate whether English intermediate-task training is still helpful on non-English target tasks. Using nine intermediate language-understanding tasks, we evaluate intermediate-task transfer in a zero-shot cross-lingual setting on the XTREME benchmark. We see large improvements from intermediate training on the BUCC and Tatoeba sentence retrieval tas

Research goal: How does incorporating intermediate language-understanding tasks from multiple high-resource languages (e.g., English, Spanish, Chinese, French) during pretraining affect the zero-shot cross-lingual transfer performance of mT5 on the XTREME benchmark compared to using only English intermediate tasks, measured by average mAUROC scores?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 9.3/10.