Intermediate-Task Fine-Tuning Effects on XLM-R Base Latency and Throughput in XTREME-R Zero-Shot Evaluation

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 intermediate-task fine-tuning on the inference latency and throughput of XLM-R Base during zero-shot cross-lingual evaluation on XTREME-R, and how does this compare to direct fine-tuning on the target task in terms of efficiency-performance trade-offs?

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