Multimodal Benchmark Training for Zero-Shot Cross-Lingual Transfer in Vision-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: Does intermediate-task training on multimodal benchmarks (e.g., VQA or OK-VQA) improve zero-shot cross-lingual transfer in vision-language models like CLIP or BLIP-2, as measured by accuracy on XTREME-R or MMMU?

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