image image | label int64 | parent_dataset string |
|---|---|---|
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
0 | coyo-700m | |
0 | coyo-700m | |
1 | laion-2b | |
0 | coyo-700m | |
1 | laion-2b | |
1 | laion-2b | |
1 | laion-2b |
End of preview. Expand in Data Studio
OpenLVLM-MIA
OpenLVLM-MIA: A Controlled Benchmark Revealing the Limits of Membership Inference Attacks on Large Vision-Language Models
Overview
- OpenLVLM-MIA offers a controlled benchmark to reassess membership inference attacks (MIA) on large vision-language models beyond dataset-induced biases.
- The benchmark consists of a 6,000-image dataset with controlled member/non-member distributions and ground-truth membership at three training stages.
- On this setup, state-of-the-art MIA approaches perform at chance level, clarifying the true difficulty of the problem and motivating more robust privacy defenses.
Resources
- Codes: yamanalab/openlvlm-mia
- Model: paper-2229/openclip-llava
- Downloads last month
- 34