Datasets:

JaineLi
/

VL-MIA-image

configs:
  - config_name: img_dalle
    data_files: DALLE/*
  - config_name: img_Flickr
    data_files: flickr_2024/*
  - config_name: img_Flickr_2k
    data_files: flickr_2024_2k/*
  - config_name: img_Flickr_10k
    data_files: flickr_2024_10k/*
license: cc-by-4.0
tags:
  - MIA
  - ' LLM'
  - ' VLLM'

VL-MIA

VL-MIA is elaborated for membership inference attacks on VLLM :

Label 0: Refers to the unseen non-member data. Label 1: Refers to member data.

For the text dataset, please see https://huggingface.co/datasets/JaineLi/VL-MIA-text