Generalization of Non-English Intermediate-Task Training Beyond XTREME and Efficiency Metrics Across Benchmark Suites

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: Does the impact of non-English intermediate-task training on zero-shot cross-lingual transfer generalize to other benchmark suites beyond XTREME (e.g., XTREME-R, TYDI QA), and how do efficiency metrics like inference time or F1 scores vary across these benchmarks?

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