Large Language Models With Contrastive Decoding Algorithm for Hallucination Mitigation in Low‐Resource Languages

ABSTRACT Neural machine translation (NMT) has advanced with deep learning and large‐scale multilingual models, yet translating low‐resource languages often lacks sufficient training data and leads to hallucinations. This often results in translated content that diverges significantly from the source...

Descripción completa

Detalles Bibliográficos

Autores principales:	Zan Hongying, Arifa Javed, Muhammad Abdullah, Javed Rashid, Muhammad Faheem
Formato:	Artigo
Lenguaje:	Inglês
Publicado:	Wiley 2025-08-01
Colección:	CAAI Transactions on Intelligence Technology
Materias:	ChatGLM contrastive decoding hallucination LLAMA LLM low resource NMT
Acceso en línea:	https://doi.org/10.1049/cit2.70004
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!