Impact of Code-Related Intermediate-Task Training on Zero-Shot Cross-Lingual Code Generation Performance

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 intermediate-task training on code-related tasks (e.g., HumanEval, MBPP) improve the zero-shot cross-lingual code generation performance (measured by pass@1, pass@k) of multilingual models (e.g., CodeGen, StarCoder) compared to general-domain intermediate tasks?

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