SATrans
SATrans copied to clipboard
The source code for our paper "Scenario-Adaptive Feature Interaction for Click-Through Rate Prediction" (accepted by KDD2023 Applied Science Track), which proposes a model for Multi-Scenario/Multi-Dom...
``` def norm_df(path,out_path): df = pd.read_csv(path,dtype=np.int32) print(df.shape) df -= (min_v-1) df[df
1.这个meta_dnn_hidden_units变量大小是[32,64,32]  也就是说meta网络只有两层变换,32->64,64->32, 我看论文里提到meta网络在四层变换的时候效果最好,具体参数是什么呢? 