SOURCE-LINKED INTELLIGENCE

Learning Where Outcomes Change:Credit-Addressable Reasoning for Multimodal Geometry

arXiv · AI, language, vision and robotics · article · Aug 31, 2026 · UTC

Multimodal geometry reasoning requires VLMs to extract precise visual relations and preserve them through multi-step deduction. Existing free-form traces obscure the decisions that determine the answer, and trajectory-level reinforcement learning distributes a single terminal signal across the entire response. We introduce credit-addressable reasoning, in which the semantic units exposed during inference also define where learning compares alternatives and assigns credit. We instantiate this principle with Code-CoT, which retains the diagram, represents visual relations as line-addressable exe

Read original source ↗ Open in workspace

recordType

paper

region

Global