AI RESEARCH
LatentGeo: Learnable Auxiliary Constructions in Latent Space for Multimodal Geometric Reasoning
arXiv CS.CV
•
ArXi:2603.12166v1 Announce Type: new Despite recent advances in multimodal reasoning, representing auxiliary geometric constructions remains a fundamental challenge for multimodal large language models (MLLMs). Such constructions are absent from the original diagram and must be