List your company

Image Captioning on IIW-400

84.2PrecisionCapMAS
48.00857.40466.876.196Apr 14, 2026

Evaluation results

MethodLinks
CapMASModel=GPT-4.1-mini2026.0484.267.775.10
ReflectCAPModel=GPT-4.1-mini2026.0483.87277.52.4
Zero-shotModel=GPT-4.1-mini2026.0483.668.175.1—
CapMASModel=GPT-4.1-nano2026.0483.157.267.82
Self-Corr.Model=GPT-4.1-mini2026.0482.669.375.40.3
Few-shotModel=GPT-4.1-mini2026.0481.971.576.31.2
CapMASModel=Qwen3-32B2026.04817276.20.5
CapMASModel=Qwen3-8B2026.0479.368.873.71.1
Self-Corr.Model=GPT-4.1-nano2026.0479.262.6700.2
Zero-shotModel=Qwen3-32B2026.0479.272.575.7—
ReflectCAPModel=Qwen3-32B2026.0479.273.876.40.7
Zero-shotModel=GPT-4.1-nano2026.0478.962.669.8—
CapMASModel=InternVL-38B2026.0478.657.566.42.1
Self-Corr.Model=Qwen3-32B2026.0478.673.6760.3
ReflectCAPModel=GPT-4.1-nano2026.047867.272.22.4
Few-shotModel=GPT-4.1-nano2026.0477.167.171.82
ReflectCAPModel=Qwen3-8B2026.047771.2741.4
Zero-shotModel=Qwen3-8B2026.0476.369.272.6—
Self-Corr.Model=Qwen3-8B2026.0476.370730.4
Self-Corr.Model=InternVL-38B2026.0473.758.2650.7
ReflectCAPModel=InternVL-38B2026.0473.764.268.64.3
CapMASModel=Qwen2.5-32B2026.0473.563.167.90.2
Few-shotModel=Qwen3-32B2026.0473.474.473.91.8
Zero-shotModel=InternVL-38B2026.0472.557.864.3—
CapMASModel=InternVL-4B2026.0472.253.361.32.2
CapMASModel=Qwen2.5-7B2026.047257.463.91.2
Zero-shotModel=Qwen2.5-32B2026.0471.96467.7—
Few-shotModel=Qwen3-8B2026.0471.372.271.80.8
Self-Corr.Model=Qwen2.5-32B2026.0470.266.268.10.4
ReflectCAPModel=Qwen2.5-32B2026.0469.966.468.10.4
ReflectCAPModel=Qwen2.5-7B2026.0468.862.365.42.7
Zero-shotModel=Qwen2.5-7B2026.0468.357.962.7—
Few-shotModel=Qwen2.5-32B2026.0464.964.564.73
ReflectCAPModel=InternVL-4B2026.0464.861.3633.9
Zero-shotModel=InternVL-4B2026.0464.154.959.1—
Self-Corr.Model=Qwen2.5-7B2026.0463.857.360.42.3
Few-shotModel=InternVL-38B2026.0463.262.162.61.7
Self-Corr.Model=InternVL-4B2026.0462.154.157.81.3
Few-shotModel=Qwen2.5-7B2026.0449.956.653.19.6
Few-shotModel=InternVL-4B2026.0449.452.851.18