ERASE prunes 85% of vision tokens in Qwen2.5-VL-7B while retaining 89.46% accuracy, outperforming prior methods that retain only 78.1%.
Accelerating multimodal large language models via dynamic visual-token exit and the empirical findings
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
ERASE: Eliminating Redundant Visual Tokens via Adaptive Two-Stage Token Pruning
ERASE prunes 85% of vision tokens in Qwen2.5-VL-7B while retaining 89.46% accuracy, outperforming prior methods that retain only 78.1%.