tf_repos icon indicating copy to clipboard operation
tf_repos copied to clipboard

Published 20 hours ago • verified publisher icon lambdaji

连续特征和离散特征 embedding会有重合第问题？

Open chenzk1993 opened this issue 4 years ago • 1 comments

tf_repos/deep_ctr/Feature_pipeline/get_criteo_feature.py 第50行中对离散特征编号从0， 但同时连续特征是1-13，会导致连续特征对应的embedding会和离散特征embedding相同的问题，想问下？

chenzk1993 avatar Mar 04 '21 06:03 chenzk1993

取100条数据跑跑看

lambdaji avatar Mar 24 '21 05:03 lambdaji