List your company

Spatial Reasoning on WhatsUp-VLM COCOQA (test)

98.4Accuracy(IA)3
54.61665.98377.3588.717Jun 28, 2026

Evaluation results

MethodLinks
(IA)3Backbone=Qwen2.5-VL-3B, Trainable Parameters=524.K2026.0698.4
(IA)3Backbone=LLaVA-7B, Trainable Parameters=664.K2026.0696.6
ScAle (attn)Backbone=Qwen2.5-VL-3B, Trainable Parameters=5762026.0695
ScAle (attn)Backbone=LLaVA-7B, Trainable Parameters=1.02K2026.0691.5
FrozenBackbone=Qwen2.5-VL-3B2026.0683.9
FrozenBackbone=LLaVA-7B2026.0656.3