Cross-lingual performance variation in zero-shot FLORES-200 with reasoning vs. code intermediate tasks

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 intermediate-task training on reasoning tasks (e.g., GSM8K) improve zero-shot cross-lingual performance on FLORES-200 when compared to code-based intermediate tasks (e.g., HumanEval), and how does this vary by language family?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 9.3/10.