Enhancing Zero-Shot Cross-Lingual Transfer via Task-Specific Adaptations in English Intermediate-Task Training

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: Can task-specific adaptations (e.g., modifying task formulations or adding multilingual prompts) improve the zero-shot cross-lingual transfer performance of English intermediate-task training beyond baseline XTREME metrics?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 9.2/10.