How does the performance of mT5-3B on XTREME-R vary when fine-tuned with multilingual intermediate-task data (e.g., XNLI)

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 mT5-3B on XTREME-R vary when fine-tuned with multilingual intermediate-task data (e.g., XNLI) compared to monolingual English intermediate-task data, measured in accuracy and inference efficiency (tokens per second)?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 7.8/10.