Cargando…
Beyond the Transformer: A Novel Polynomial Inherent Attention (PIA) Model and Its Great Impact on Neural Machine Translation
This paper describes a novel polynomial inherent attention (PIA) model that outperforms all state-of-the-art transformer models on neural machine translation (NMT) by a wide margin. PIA is based on the simple idea that natural language sentences can be transformed into a special type of binary atten...
| Autores principales: | , |
|---|---|
| Formato: | Online Artículo Texto |
| Lenguaje: | English |
| Publicado: |
Hindawi
2022
|
| Materias: | |
| Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9519290/ https://www.ncbi.nlm.nih.gov/pubmed/36188704 http://dx.doi.org/10.1155/2022/1912750 |