Сравнение методов

	Адаптация RoBERTa на основе домена для классификации	Многоязычная классификация на основе RoBERTa
Область	Глубокое обучение	Глубокое обучение
Семейство	Machine learning	Machine learning
Год появления	2019–2020	2020
Автор метода	Liu et al. (RoBERTa); Gururangan et al. (domain-adaptive pretraining)	Conneau, A. et al. (Facebook AI Research)
Тип	Pre-trained transformer with domain-adaptive pretraining and task fine-tuning	Pretrained multilingual transformer fine-tuned for classification
Основополагающий источник	Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., & Stoyanov, V. (2019). RoBERTa: A Robustly Optimized BERT Pretraining Approach. arXiv preprint arXiv:1907.11692.	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.
Другие названия	DA-RoBERTa, domain-adapted RoBERTa classifier, RoBERTa domain adaptation, domain-specific RoBERTa fine-tuning	XLM-RoBERTa classification, mRoBERTa, cross-lingual RoBERTa classifier, multilingual transformer classification
Связанные	5	4
Сводка	Domain-adaptive RoBERTa-based classification extends the RoBERTa transformer by first continuing its masked-language-model pretraining on a domain-specific corpus before fine-tuning for a classification task. This two-stage adaptation bridges the gap between general web-crawled training data and specialized fields such as biomedical, legal, or scientific text, consistently outperforming standard RoBERTa fine-tuning when target-domain text is available.	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.