增加新的特征值

data_fields_v3.config

@@ -715,3 +715,11 @@
  profession STRING
  user_vid_share_tags_1d STRING
  user_vid_share_tags_14d STRING
+ user_vid_return_cate1_14d STRING
+ user_vid_return_cate2_14d STRING
+ user_vid_share_cate1_14d STRING
+ user_vid_share_cate2_14d STRING
+ creative_type STRING
+ creative_hook_embedding STRING
+ creative_why_embedding STRING
+ creative_action_embedding STRING

widedeep_v12_1.py

@@ -21,15 +21,20 @@ sparse_features = [
     "region", "city", "brand",
     "vid", "cate1", "cate2",
     "apptype", "hour", "hour_quarter", "root_source_scene", "root_source_channel", "is_first_layer", "title_split",
-    "profession"
+    "profession", "creative_type"
 ]
 tag_features = [
     "user_vid_return_tags_2h", "user_vid_return_tags_1d", "user_vid_return_tags_3d",
-    "user_vid_return_tags_7d", "user_vid_return_tags_14d", "user_vid_share_tags_1d", "user_vid_share_tags_14d"
+    "user_vid_return_tags_7d", "user_vid_return_tags_14d", "user_vid_share_tags_1d", "user_vid_share_tags_14d",
+    "user_vid_share_tags_1d", "user_vid_share_tags_14d", "user_vid_return_cate1_14d", "user_vid_return_cate2_14d",
+    "user_vid_share_cate1_14d", "user_vid_share_cate2_14d"
 ]
 seq_features = [
     "user_cid_click_list", "user_cid_conver_list"
 ]
+nlp_features = [
+    "creative_hook_embedding", "creative_why_embedding", "creative_action_embedding"
+]
 
 input_type_map = {
     'BIGINT': 'INT64',
@@ -129,6 +134,15 @@ for name in tag_features + seq_features:
   separator: ','
 }}""")
 
+for name in nlp_features:
+    print(f"""feature_configs {{
+  input_names: "{name}"
+  feature_type: TagFeature
+  hash_bucket_size: 1000000
+  embedding_dim: 6
+  separator: '|'
+}}""")
+
 
 def wide_and_deep():
     print("""
@@ -145,7 +159,7 @@ model_config {
   feature_groups: {
     group_name: 'deep'""")
 
-    for name in dense_features + sparse_features + tag_features + seq_features:
+    for name in dense_features + sparse_features + tag_features + seq_features + nlp_features:
         print(f"""    feature_names: '{name}'""")
 
     print("""    wide_deep: DEEP
@@ -180,7 +194,7 @@ model_config {
   feature_groups: {
     group_name: 'deep'""")
 
-    for name in top_dense_features + sparse_features + tag_features + seq_features:
+    for name in top_dense_features + sparse_features + tag_features + seq_features + nlp_features:
         print(f"""    feature_names: '{name}'""")
 
     print("""    wide_deep: DEEP