Performance of XLM-R and mT5 in Zero-Shot Cross-Lingual Logical Inference with Non-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: How does the performance of XLM-R and mT5 on zero-shot cross-lingual logical inference tasks in XTREME-R vary when intermediate-task training is conducted on non-English tasks instead of English tasks?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 9.0/10.