Could anyone provide a code example fro Wide and Deep model (wdl module)?

[Haizhuolaojisite]

Hi all,

I read the Quick-Start of the document, but I am confused about the "step 3: Generate feature columns", it seems like dnn_feature_columns is same as linear_feature_columns. Is it true for all models?

I tried to train the Wide & Deep model using the wdl module (link). But how should I prepare the dnn_feature_columns and linear_feature_columns? I learned that the wide part takes linear_feature_columns, which should be crossed columns and composite column for those crossed columns. And dnn_feature_columns basically take all columns to deep part.

dnn_feature_columns = sparse_feature_columns + dense_feature_columns
linear_feature_columns = sparse_feature_columns + dense_feature_columns
feature_names = get_feature_names(linear_feature_columns + dnn_feature_columns)

Could anyone guide me how to prepare dnn_feature_columns and linear_feature_columns for wide & deep model please? Thank you so much!!