**Cross-lingual Performance of Mixed-Language Intermediate-Task Fine-Tuning on XTREME-R**

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 impact of mixed-language intermediate-task fine-tuning (e.g., 50% English, 50% non-English data) on zero-shot cross-lingual evaluation on XTREME-R compared to English-only intermediate-task fine-tuning, measured by macro-averaged F1 scores across all languages?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 9.2/10.