Impact of Multi-Task versus Single-Task Intermediate Training on Zero-Shot Cross-Lingual Transfer in Small and Large Language

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 combining multiple English intermediate tasks versus a single task on zero-shot cross-lingual transfer performance in XTREME-R for models under 1B parameters compared to those over 10B parameters?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.9/10.