Performance variation of intermediate-task training across language model architectures on XTREME-R benchmarks

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 the performance of intermediate-task training on non-English tasks vary across different language model architectures (e.g., Transformers vs. other architectures) when evaluated on XTREME-R benchmarks?

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