An instruction-guided aggregator that dynamically weights visual features from different encoder layers improves LLaVA-1.5 on most tested multimodal benchmarks.
Instructblip: Towards general-purpose vision-language models with instruction tuning
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
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Instruction-Guided Fusion of Multi-Layer Visual Features in Large Vision-Language Models
An instruction-guided aggregator that dynamically weights visual features from different encoder layers improves LLaVA-1.5 on most tested multimodal benchmarks.