Multilingual Intermediate-Task Training for Zero-Shot Cross-Lingual Accuracy in 1B--10B Parameter Models on 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: What is the effect of incorporating multilingual intermediate-task training (e.g., using XNLI or MLQA) on zero-shot cross-lingual accuracy for models between 1B and 10B parameters on XTREME-R compared to English-only intermediate tasks?

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