General
MARINE
MARINE object-hallucination evaluation: per-item outputs for LLaVA-v1.5-7B with and without the MARINE image-grounded guidance intervention, on POPE (binary object probing) and CHAIR (caption hallucination) over MSCOCO images.
3,500items
2subjects
unknownlicense
safetydomain
textmodality
imagemodality
item-level responses released
Saturation status: Unknown
Response matrix
Fit to width. Hover for subject & item; click a cell for details.
Correct (1)Incorrect (0)Unobserved
Scale: 1 = correct · 0 = incorrect
Subjects
- 1LLaVA-v1.5-7B + MARINE0.8643
- 2LLaVA-v1.5-7B0.8111