Comparative Impact of Intermediate-Task Training and Target-Language Fine-Tuning on XTREME-R Reasoning Accuracy

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 comparative impact of English intermediate-task training versus direct target-language fine-tuning on reasoning task accuracy in the XTREME-R suite?

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