Scaling Multilingual Intermediate-Task Training for Zero-Shot Cross-Lingual Transfer in 10B-Parameter Models on XTREME-R NLI

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 effect of multilingual intermediate-task training on the zero-shot cross-lingual transfer performance of 10B-parameter models compared to English-only intermediate tasks, and how does this trade-off scale with the number of languages in the intermediate training set on XTREME-R NLI tasks?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.7/10.