Cross-Lingual Reasoning Performance of LLMs via Intermediate-Task 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: How does intermediate-task training on English language understanding tasks affect the zero-shot cross-lingual reasoning performance of LLMs (e.g., LLaMA, Mistral) on XTREME-R, as measured by F1 score and latency trade-offs?

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