修复生成tfrecord文件时的bug

tensnsorflow/es/feature.py

@@ -95,6 +95,7 @@ def get_data():
         train[i] = train[i].map(value_map)
         test[i] = test[i].map(value_map)
 
+    print(train[["clevel2_id", "channel","app_list"]].head(6))
     print("train shape")
     print(train.shape)
     print("test shape")
@@ -148,8 +149,8 @@ def get_predict(date,value_map,app_list_map,level2_map):
     df["clevel2_id"] = df["clevel2_id"].fillna("lost_na")
     df["clevel2_id"] = df["clevel2_id"].apply(app_list_func, args=(level2_map,))
 
-    print("predict shape")
-    print(df.shape)
+    # print("predict shape")
+    # print(df.shape)
     df["uid"] = df["device_id"]
     df["city"] = df["ucity_id"]
     features = ["ucity_id", "ccity_name", "device_type", "manufacturer",
@@ -177,13 +178,13 @@ def get_predict(date,value_map,app_list_map,level2_map):
 
     print("native")
     print(native_pre.shape)
-    print(native_pre.head())
+
     native_pre[["uid","city","cid_id"]].to_csv(path+"native.csv",index=False)
     write_csv(native_pre, "native",200000)
 
     print("nearby")
     print(nearby_pre.shape)
-    print(nearby_pre.head())
+
     nearby_pre[["uid","city","cid_id"]].to_csv(path+"nearby.csv",index=False)
     write_csv(nearby_pre, "nearby", 160000)
 

tensnsorflow/es/train.py

@@ -53,7 +53,7 @@ def input_fn(filenames, batch_size=32, num_epochs=1, perform_shuffle=False):
         features = {
             "y": tf.FixedLenFeature([], tf.float32),
             "z": tf.FixedLenFeature([], tf.float32),
-            "ids": tf.FixedLenFeature([8], tf.int64),
+            "ids": tf.FixedLenFeature([FLAGS.field_size], tf.int64),
             "app_list": tf.VarLenFeature(tf.int64),
             "level2_list": tf.VarLenFeature(tf.int64)
         }