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 tasResearch goal: How does the number of intermediate language-understanding tasks (5 vs. 15) affect the zero-shot cross-lingual transfer performance of PaLI on XTREME-R compared to text-only models like mT5 and XLM-R, when evaluated using the XTREME-R leaderboard metrics?Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 9.2/10.