Safety
DIS-CO (MovieTection)
VLMs identify the source movie of a frame (or its caption) in free-form generation; response is per-frame correct title identification, used to infer training-data membership.
13,999items
4subjects
CC-BY-NC-SA-4.0license
safetydomain
knowledgedomain
imagemodality
textmodality
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
- 1GPT-4o (2024-08-06)0.1966
- 2Gemini 1.5 Pro0.1232
- 3Llama 3.2 90B Vision Instruct0.1135
- 4Qwen2-VL 72B Instruct0.1076