Task-Specific Fine-Tuning Duration Effects on Zero-Shot Cross-Lingual Transfer Efficiency and Effectiveness in XTREME-R

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: To what extent does task-specific fine-tuning duration (e.g., 1 epoch vs. 10 epochs) on English intermediate tasks influence the efficiency and effectiveness of zero-shot cross-lingual transfer on XTREME-R tasks, evaluated by F1 score and inference latency?

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