Scaling Model Size Effects on Zero-Shot Cross-Lingual Performance in XTREME-R with Hybrid Batch 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: What is the effect of scaling model size on zero-shot cross-lingual performance in XTREME-R when using hybrid batch training (English + multilingual data) versus English-only intermediate-task fine-tuning, as measured by accuracy across high-, mid-, and low-resource languages?

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