Zero-shot cross-lingual performance of multilingual models on XTREME-M with code generation versus NLU intermediate tasks

Zero-shot cross-lingual knowledge transfer enables the multilingual pretrained language model (mPLM), finetuned on a task in one language, make predictions for this task in other languages. While being broadly studied for natural language understanding tasks, the described setting is understudied for generation. Previous works notice a frequent problem of generation in a wrong language and propose approaches to address it, usually using mT5 as a backbone model. In this work, we test alternative mPLMs, such as mBART and NLLB-200, considering full finetuning and parameter-efficient finetuning wi

Research goal: How does the zero-shot cross-lingual performance (F1 score) of multilingual models on XTREME-M vary when using code generation tasks (e.g., MBPP) as intermediate tasks compared to using natural language understanding tasks (e.g., SuperGLUE)?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.5/10.