ChatImage renders LLM answers as images, then uses visual grounding to place clickable hotspots on rendered regions for interactive follow-up.
InternVL: Scaling up vision foundation models and aligning for generic visual-linguistic tasks
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
-
ChatImage: Navigating Long-Form LLM Answers through Interactive Images
ChatImage renders LLM answers as images, then uses visual grounding to place clickable hotspots on rendered regions for interactive follow-up.