Multitask Intermediate Task Transfer for Zero-Shot Cross-Lingual Performance

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: What is the effect of combining multiple text-only intermediate tasks (e.g., NLI, QA, classification) on zero-shot cross-lingual transfer compared to single-task transfer? Measure performance (e.g., exact match, F1) on XTREME and analyze task complementarity.

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