Focal Pruning (FoPru) retains as few as 25% of visual tokens, selected by vision-encoder attention, and keeps accuracy within about 1% on three LVLM benchmarks while cutting time-to-first-token by up to 2.5x.
Honeybee: Locality-enhanced projector for multimodal llm
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
FoPru: Focal Pruning for Efficient Large Vision-Language Models
Focal Pruning (FoPru) retains as few as 25% of visual tokens, selected by vision-encoder attention, and keeps accuracy within about 1% on three LVLM benchmarks while cutting time-to-first-token by up to 2.5x.