A capability-aligned SFT and necessity-gated RL recipe lets an 8B multimodal model use visual tools effectively, outperforming larger models on high-resolution perception benchmarks.
Title resolution pending
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
-
ToolVision: Learning When and How to Use Visual Tools with Capability-Aligned Supervision
A capability-aligned SFT and necessity-gated RL recipe lets an 8B multimodal model use visual tools effectively, outperforming larger models on high-resolution perception benchmarks.