V-PRUNE: Semantic-Aware Patch Pruning Before Tokenization in Vision–Language Model Inference
Recent vision–language models (VLMs) achieve strong performance across multimodal benchmarks but suffer from high inference costs due to the large number of visual tokens. Prior studies have shown that many image tokens receive consistently low attention scores during inference, indicating that a su...
Na minha lista:
| Principais autores: | , |
|---|---|
| Formato: | Artigo |
| Idioma: | Inglês |
| Publicado em: |
MDPI AG
2025-08-01
|
| Colecção: | Applied Sciences |
| Assuntos: | |
| Acesso em linha: | https://www.mdpi.com/2076-3417/15/17/9463 |
| Tags: |
Sem tags, seja o primeiro a adicionar uma tag!
|