List your company

Mathematical Reasoning on MathVerse-Plus Vision Intensive 1.0 (test)

45AccuracySpatialMath
3.414.22535.8Jan 24, 2026

Evaluation results

MethodLinks
SpatialMathBase Model=Phi-42026.0145
SFTBase Model=Phi-42026.0133
ICL (1-shot)Base Model=Phi-42026.0131
Zero-ShotBase Model=Phi-42026.0130
CoT (1-shot)Base Model=Phi-42026.0130
SpatialMathBase Model=LLaVA-NeXT-34B2026.0128
SFTBase Model=LLaVA-NeXT-34B2026.0122
SFT + Data AugmentationBase Model=Phi-42026.0122
Zero-ShotBase Model=LLaVA-NeXT-34B2026.0121
SFT + Data AugmentationBase Model=LLaVA-NeXT-34B2026.0118
CoT (1-shot)Base Model=LLaVA-NeXT-34B2026.0115
ICL (1-shot)Base Model=LLaVA-NeXT-34B2026.015