Does adding code-generation tasks to the intermediate-task training pipeline enhance the zero-shot cross-lingual code

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: Does adding code-generation tasks to the intermediate-task training pipeline enhance the zero-shot cross-lingual code understanding capabilities on the MBPP benchmark, measured by pass@1 scores?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.0/10.