Allocating visual tokens by the cost of getting a question wrong reduces cost-weighted error of VLMs under a fixed total token budget, with controlled evidence that the gain comes from error cost, not content sensitivity.
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year =
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Not All Visual Tokens Are Equally Safe to Remove:Consequence-Sensitive Visual Token Compression
Allocating visual tokens by the cost of getting a question wrong reduces cost-weighted error of VLMs under a fixed total token budget, with controlled evidence that the gain comes from error cost, not content sensitivity.