List your company
ResearchFormerly WizWand

Generative Hallucination Mitigation on Object HalBench (test)

49.6CHAIR_SAVISC
8.5219.18529.8540.515Apr 20, 2026
Updated 6mo ago

Evaluation results

MethodLinks
AVISCStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=ACL’252026.0449.612.9
VCDStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=CVPR’242026.0448.824.3
M3IDStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=CVPR’242026.0447.812.5
ICDStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=ACL’242026.0447.413.9
LLaVA-1.5-7B (Base)Strategy=Standard LVLM2026.0446.322.6
OPERAStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=CVPR’242026.0445.122.3
MoDStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=ACL’252026.0442.612.4
EOSStrategy=Fine-tune LVLMs with external annotation data, Venue=ACL’242026.0440.317.8
EAZYStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=ICCV’252026.0438.811.4
LLaVA-RLHFStrategy=Fine-tune LVLMs with external annotation data, Venue=ACL’242026.0438.118.9
DeCoStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=ICLR’252026.0437.811.1
HALCStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=ICML’242026.0421.77.1
OctopusStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=CVPR’252026.0420.86.6
ONLYStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B, Venue=ICCV’252026.04206.2
HDPOStrategy=Fine-tune LVLMs with external annotation data, Venue=ACL’252026.0416.65.1
GPT-4VStrategy=Standard LVLM2026.0413.67.3
RLAIF-VStrategy=Fine-tune LVLMs with external annotation data, Venue=CVPR’252026.0410.55.2
PSRDStrategy=Post-hoc hallucination mitigation, Base Model=LLaVA-1.5-7B2026.0410.14.1