List your company

Hallucination Evaluation on ObjHal

47.3CRs AccuracyMuffin-13B
27.95632.9783843.022Jul 29, 2025

Evaluation results

MethodLinks
Muffin-13BBackbone=Muffin-13B2025.0747.3——15.2
RLHFBackbone=Muffin-13B2025.0745.5——12.7
DPOBackbone=Muffin-13B2025.0743.8——13.9
CHiP-DPOBackbone=Muffin-13B2025.0735.2——11.5
TARS (Replace)Backbone=Muffin-13B2025.0729.3——8.8
TARS (Mask)Backbone=Muffin-13B2025.0728.7——8.2
LLaVA2025.04—6329.5—
POPEN2025.04—9.24.9—
RLHF-V2025.04—12.27.5—