Scaling Intermediate-Task Data Size and Zero-Shot Cross-Lingual Performance in Quantized Language Models

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: What is the impact of scaling intermediate-task training data size on the zero-shot cross-lingual performance of language models, measured by accuracy on XTREME tasks, and how does it correlate with inference efficiency after quantization?

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