Impact of Model Scaling on Zero-Shot Cross-Lingual NER Performance 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: What is the impact of scaling model size (e.g., from 1B to 175B parameters) on zero-shot cross-lingual named entity recognition performance (measured in macro-F1) when using English intermediate-task training versus multilingual intermediate-task training on the XTREME-R benchmark?

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