List your company

Multi-image hallucination evaluation on MUIRBench

62AccuracyQwen2.5-VL + CAPL
28.56437.244545.92554.6055Mar 7, 2026

Evaluation results

MethodLinks
Qwen2.5-VL + CAPLParams=7B2026.0362
GLM4.1VBase + CAPLParams=9B2026.0360.57
Qwen2.5-VLParams=7B2026.0358.42
GLM4.1VBaseParams=9B2026.0357.84
InternVL2.5 + CAPLParams=8B2026.0352.12
InternVL2.5Params=8B2026.0348.54
InternVL2Params=7B2026.0345.61
Qwen2VLParams=7B2026.0339.57
Idefics3Params=8B2026.0330.96
LLaVA-OVParams=7B2026.0330.85
LLaVA-NextParams=7B2026.0330.5
Idefics2Params=8B2026.0329.85