Multimodal vs. Text-Only Pre-Training for Zero-Shot Cross-Lingual Transfer in XTREME-R Low-Resource Languages

This paper studies zero-shot cross-lingual transfer of vision-language models. Specifically, we focus on multilingual text-to-video search and propose a Transformer-based model that learns contextualized multilingual multimodal embeddings. Under a zero-shot setting, we empirically demonstrate that performance degrades significantly when we query the multilingual text-video model with non-English sentences. To address this problem, we introduce a multilingual multimodal pre-training strategy, and collect a new multilingual instructional video dataset (MultiHowTo100M) for pre-training. Experimen

Research goal: What is the impact of multimodal pre-training (e.g., using CLIP or LaMDA) on zero-shot cross-lingual transfer for low-resource languages in the XTREME-R benchmark compared to text-only pre-training?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.2/10.