Knowledge density in image captions, not task format diversity, is the primary driver of multimodal LLM scaling performance.
For page-like images, focus on the main text or core information rather than layout or decorative elements
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Caption First, VQA Second: Knowledge Density, Not Task Format, Drives Multimodal Scaling
Knowledge density in image captions, not task format diversity, is the primary driver of multimodal LLM scaling performance.