Intermediate Code-Related Task Training for Zero-Shot Non-English Transfer in the XTREME Benchmark

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 intermediate-task training with code-related tasks (e.g., code-switching, programming language understanding) improve zero-shot transfer performance on non-English tasks in the XTREME benchmark?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.9/10.