Cross-task Consistency Effects on Zero-shot Cross-lingual Transfer Performance in XTREME

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 cross-task consistency in intermediate-task selection (e.g., using only question-answering vs. mixed tasks) affect zero-shot cross-lingual transfer performance on XTREME, measured by average F1 score deviations across languages?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.4/10.