Results data for the paper Clustering and Token Denoising for Faster and More Robust VLMs

Published: 25 September 2025| Version 1 | DOI: 10.17632/87fcmvmjrw.1
Contributor:
Baptiste Rossigneux

Description

Data obtained through the evaluation of LLaVA-1.5 with visual tokens pruned with our method ClustRS on MM-VET or ScienceQA-IMG. This data is thoroughly reported in the paper.

Files

Institutions

  • Commissariat a l'Energie Atomique

Categories

Machine Learning, Computer Vision Algorithms, Large Language Model

Licence