List your company

Image-to-Text Generation on MS-COCO 5% 2014 (test val)

25.5CLIP Similarity (CLIP)GIT-large
23.21223.80624.424.994May 29, 2026

Evaluation results

MethodLinks
GIT-largeAlpha level (α)=0.1, Model scale=large, Image size=224 × 224, Maximum text length=502026.0525.5-0.0043-0.259386.4964.59-0.219
GIT-baseAlpha level (α)=0.1, Model scale=base, Image size=224 × 224, Maximum text length=502026.0525.11-0.0021-0.253286.264.74-0.2146
BLIP-largeAlpha level (α)=0.1, Model scale=large, Image size=224 × 224, Maximum text length=502026.0524.53-0.0074-0.252781.0656.31-0.2475
BLIP-baseAlpha level (α)=0.1, Model scale=base, Image size=224 × 224, Maximum text length=502026.0523.30.007-0.22683.4963.35-0.2014