Inducing high energy-latency of large vision-language models with verbose images
Large vision-language models (VLMs) such as GPT-4 have achieved exceptional performance across various multi-modal tasks. However, the deployment of VLMs necessitates substantial energy consumption and computational resources. Once attackers maliciously induce high energy consumption and latency tim...
Главные авторы: | Gao, K, Bai, Y, Gu, J, Xia, ST, Torr, P, Li, Z, Liu, W |
---|---|
Формат: | Conference item |
Язык: | English |
Опубликовано: |
OpenReview
2024
|
Схожие документы
-
Energy-latency manipulation of multi-modal large language models via verbose samples
по: Gao, K, и др.
Опубликовано: (2024) -
The verbosity epidemic.
по: Grais, R, и др.
Опубликовано: (2008) -
Evaluation of Rust code verbosity, understandability and complexity
по: Luca Ardito, и др.
Опубликовано: (2021-02-01) -
Head Concepts Selection for Verbose Medical Queries Expansion
по: Mohammed Maree, и др.
Опубликовано: (2020-01-01) -
The Role of Inhibition in Age-Related Off-Topic Verbosity: Not Access but Deletion and Restraint Functions
по: Shufei eYin, и др.
Опубликовано: (2016-04-01)