Impact of Non-English Intermediate-Task Training on Zero-Shot XTREME-R Accuracy

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: To what extent does adding intermediate-task training with non-English datasets (e.g., TyDiQA) improve zero-shot accuracy on XTREME-R compared to English-only intermediate training, as measured by per-language F1 scores?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.5/10.