Our new X account is live! Follow @wizwand_team for updates

Vision-Language Understanding and Reasoning on LLaVA Multimodal Evaluation Suite 1.5 (test/val)

0.619GQA

Vanilla

Updated 3d ago

Evaluation Results

Method	Links
Vanilla 2026.01		0.619	0.647	1,862	0.859	0.695	0.785	0.582	0.586	0.311	0.501	1
VisionTrim (Ours) 2026.01		0.61	0.644	1,798	0.868	0.708	0.784	0.584	0.583	0.332	0.513	1.006
VScan 2026.01		0.606	0.639	1,806	0.862	0.686	0.778	0.577	-	-	0.504	0.989
VisionTrim (Ours) 2026.01		0.603	0.64	1,788	0.866	0.697	0.782	0.582	0.575	0.327	0.514	0.999
VScan 2026.01		0.598	0.63	1,792	0.861	0.689	0.771	0.573	-	-	0.517	0.986
VisionZip 2026.01		0.593	0.63	1,783	0.853	0.689	0.768	0.573	0.564	0.317	0.505	0.983
VisionTrim (Ours) 2026.01		0.588	0.63	1,780	0.862	0.71	0.768	0.568	0.548	0.322	0.523	0.988
VScan 2026.01		0.583	0.621	1,698	0.85	0.691	0.754	0.556	-	-	0.518	0.968
SparseVLM 2026.01		0.576	0.625	1,721	0.836	0.691	0.756	0.561	0.558	0.302	0.5	0.964
VisionZip 2026.01		0.576	0.62	1,762	0.832	0.689	0.756	0.568	0.549	0.326	0.5	0.972
PDrop 2026.01		0.573	0.633	1,797	0.848	0.692	0.764	0.565	0.572	0.308	0.5	0.976
PDrop 2026.01		0.571	0.616	1,761	0.826	0.684	0.76	0.566	0.562	0.31	0.496	0.965
SparseVLM 2026.01		0.56	0.6	1,696	0.805	0.671	0.738	0.549	0.534	0.3	0.51	0.942
VisionZip 2026.01		0.551	0.601	1,690	0.77	0.69	0.724	0.555	0.522	0.317	0.519	0.944
SparseVLM 2026.01		0.527	0.562	1,505	0.751	0.622	0.682	0.518	0.511	0.233	0.496	0.867
PDrop 2026.01		0.475	0.588	1,561	0.762	0.69	0.733	0.506	0.53	0.305	0.502	0.908