Robustness to Adversarial Examples in Zero-Shot Cross-Lingual Settings via 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: Does English intermediate-task training improve robustness to adversarial examples in zero-shot cross-lingual settings compared to multilingual intermediate-task training, as evaluated by accuracy on adversarially perturbed XTREME-R benchmarks?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 7.9/10.