方法对比

	多语言视觉Transformer	基于多语言 RoBERTa 的分类
领域	深度学习	深度学习
方法族	Machine learning	Machine learning
起源年份	2021–2023	2020
提出者	Dosovitskiy et al. (ViT base); multilingual extension by multiple groups (2021–2023)	Conneau, A. et al. (Facebook AI Research)
类型	Transformer-based vision model with multilingual capabilities	Pretrained multilingual transformer fine-tuned for classification
开创性文献	Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., & Houlsby, N. (2021). An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. International Conference on Learning Representations (ICLR 2021).	Conneau, A., Khandelwal, K., Goyal, N., Chaudhary, V., Wenzek, G., Guzman, F., Grave, E., Ott, M., Zettlemoyer, L., & Stoyanov, V. (2020). Unsupervised Cross-lingual Representation Learning at Scale. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (ACL 2020), pp. 8440–8451.
别名	Multilingual ViT, Cross-lingual Vision Transformer, Multilingual Visual Transformer, ML-ViT	XLM-RoBERTa classification, mRoBERTa, cross-lingual RoBERTa classifier, multilingual transformer classification
相关	4	4
摘要	Multilingual Vision Transformer (Multilingual ViT) extends the Vision Transformer architecture to operate across multiple languages, enabling image understanding and image-text reasoning in multilingual or cross-lingual settings. It combines patch-based image encoding with multilingual text representations, allowing a single model to serve diverse linguistic communities for tasks such as image captioning, visual question answering, and cross-lingual image retrieval.	Multilingual RoBERTa-based classification uses XLM-RoBERTa — a transformer pretrained on 100+ languages via masked language modeling — and fine-tunes it on labeled text to assign categories across multiple languages. By sharing a single model across languages, it enables robust cross-lingual and zero-shot text classification without needing separate per-language classifiers.