DiffPrune reformulates visual token pruning as continuous control of token information using an Information Throttler with importance-conditioned variance-preserving noise, enabling fully differentiable learning of scores that are hard-thresholded at inference.
Zerosense: How vision matters in long context compression.arXiv preprint arXiv:2603.11846
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
LensVLM trains VLMs to scan compressed rendered text images and selectively expand task-relevant regions, achieving 4.3x compression with near full-text accuracy and outperforming baselines up to 10.1x on text QA benchmarks.
citing papers explorer
-
Beyond Surrogate Gradients: Fully Differentiable Token Pruning for Vision-Language Models
DiffPrune reformulates visual token pruning as continuous control of token information using an Information Throttler with importance-conditioned variance-preserving noise, enabling fully differentiable learning of scores that are hard-thresholded at inference.
-
LensVLM: Selective Context Expansion for Compressed Visual Representation of Text
LensVLM trains VLMs to scan compressed rendered text images and selectively expand task-relevant regions, achieving 4.3x compression with near full-text accuracy and outperforming baselines up to 10.1x on text QA benchmarks.