List your company
ResearchFormerly WizWand

Action-relation hallucination evaluation on R-Bench Image

83.5AccuracyRVE
72.309675.214878.1281.0252May 12, 2026
Updated 5mo ago

Evaluation results

MethodLinks
RVELVLM Model=LLaVA-NeXT-7B2026.0583.588.13
RVELVLM Model=LLaVA-1.5-13B2026.0583.2687.76
VanillaLVLM Model=LLaVA-NeXT-7B2026.0583.0587.83
ICDLVLM Model=LLaVA-NeXT-7B2026.0582.9887.76
VAFLVLM Model=LLaVA-NeXT-7B2026.0582.8987.76
VAFLVLM Model=LLaVA-1.5-13B2026.0582.8987.53
VanillaLVLM Model=LLaVA-1.5-13B2026.0582.7687.38
ICDLVLM Model=LLaVA-1.5-13B2026.0582.5587.2
VCDLVLM Model=LLaVA-NeXT-7B2026.0582.587.49
RVELVLM Model=InstructBLIP-7B2026.0582.4887.59
RVELVLM Model=ShareGPT4V-7B2026.0581.8487.45
VanillaLVLM Model=InstructBLIP-7B2026.0581.7487.31
ICDLVLM Model=InstructBLIP-7B2026.0581.6687.2
VCDLVLM Model=ShareGPT4V-7B2026.0581.486.98
VAFLVLM Model=InstructBLIP-7B2026.0581.3887.24
VCDLVLM Model=LLaVA-1.5-13B2026.0581.3686.23
VCDLVLM Model=InstructBLIP-7B2026.0581.3586.87
VanillaLVLM Model=ShareGPT4V-7B2026.0580.4786.73
VAFLVLM Model=ShareGPT4V-7B2026.0579.8186.38
RVELVLM Model=LLaVA-1.5-7B2026.0579.185.64
VCDLVLM Model=LLaVA-1.5-7B2026.0578.6285.54
VanillaLVLM Model=LLaVA-1.5-7B2026.0578.0585.34
VAFLVLM Model=LLaVA-1.5-7B2026.0577.6985.23
ICDLVLM Model=ShareGPT4V-7B2026.0574.883.22
ICDLVLM Model=LLaVA-1.5-7B2026.0572.7482.26