Scaling Model Size and Cross-Lingual Transfer in XTREME-R: Efficiency Trade-offs

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 scaling model size (e.g., from 7B to 70B parameters) amplify or diminish the benefits of English intermediate-task training on cross-lingual understanding tasks in XTREME-R, and what is the efficiency trade-off in terms of inference latency?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.7/10.