Comparison of Intermediate-Task Training Approaches for Zero-Shot Transfer on XTREME-R Benchmark

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 intermediate-task training on non-English high-resource languages compare to multilingual intermediate-task training (e.g., XLM-R or mT5) in zero-shot transfer performance on the XTREME-R benchmark for classification and QA tasks? Measured by average accuracy and per-task F1 scores across languages.

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