List your company

Visual Evidence Quality Evaluation on M3CoT Reasoning (subset of 500 samples)

76.4AIM-CoT Win RateAIM-CoT
72.5874.4976.478.31Sep 30, 2025

Evaluation results

MethodLinks
AIM-CoTSelection Strategy=Info. Gain-driven AVP, Judge=GPT-4v2025.0976.416.27.4
ICoTSelection Strategy=Attention-driven Top-K, Judge=GPT-4v2025.0976.416.27.4