Scalability of Intermediate-Task Training for Zero-Shot Cross-Lingual Transfer in Code Generation

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 scalable is intermediate-task training for zero-shot cross-lingual transfer when applied to code generation tasks (e.g., CodeXGLUE) compared to natural language understanding tasks, measured by accuracy gains and training time per additional language?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 9.0/10.