It is all in the [MASK]: Simple instruction-tuning enables BERT-like masked language models as generative classifiers

While encoder-only models such as BERT and ModernBERT are ubiquitous in real-world NLP applications, their conventional reliance on task-specific classification heads can limit their applicability compared to decoder-based large language models (LLMs). In this work, we introduce ModernBERT-Large-Ins...

ver descrição completa

Detalhes bibliográficos

Principais autores:	Benjamin Clavié, Nathan Cooper, Benjamin Warner
Formato:	Artigo
Idioma:	Inglês
Publicado em:	Elsevier 2025-06-01
coleção:	Natural Language Processing Journal
Assuntos:	Zero-shot classification Multiple-choice question answering Encoder models BERT ModernBERT Masked language modeling
Acesso em linha:	http://www.sciencedirect.com/science/article/pii/S2949719125000263
Tags:	Adicionar Tag Sem tags, seja o primeiro a adicionar uma tag!