ahxt / fair_fairness_benchmark Goto Github PK

Hi, can I ask where I can download the "celebs.csv" dataset included in the line below? Also, it would be really appreciated if dataset setup process can be included in the documentations.

"df = pd.read_csv( os.path.join(path, "celeba.csv"), na_values="NA", index_col=None, sep=",", header=0)"

Thanks!

Thank you so much for updating the datasets and the running logs! I noticed that the CelebA dataset has been pre-processed into a .csv file. Would you mind share the details about how the pre-processing was done for CelebA? Thank you!

Thanks for the great work. When downloading the datasets from dropbox I noticed that jigsaw dataset was missing. Can you share the files mentioned in the below code? Also it would be great if you could provide information on how the protected attributes were binarized.

def load_jigsaw_text_data(path="/data/han/data/fairness/jigsaw", sensitive_attribute="sex"):
    # | 48842 instances, mix of continuous and discrete(train=32561, test=16281)
    # | 45222 if instances with unknown values are removed(train=30162, test=15060)

    with open( os.path.join(path, "processed_text_features.pkl"), 'rb') as f:

Thanks.