Update 'train.py'

train.py

@@ -19,38 +19,35 @@ exint.observers.append(FileStorageObserver('z-s487174-training/master'))
 @exint.config
 def my_config():
     EPOCHS = int(os.environ['EPOCHS'])
+    
 
 @exint.main
-def main(EPOCHS, _run):
-    if EPOCHS == 0:
-        EPOCHS = 500
-    train_data_x = pd.read_csv('./train_data.csv')
-    train_data_x_tmp = train_data_x.copy()
-
-    price_train = train_data_x.copy()
-    price_predict = train_data_x_tmp.pop('Index')
+def my_main(EPOCHS, _run):
+    _run.info["epochs"] = EPOCHS,
     normalize = layers.Normalization()
-    normalize.adapt(price_train)
-
-    price_model = tensorflow.keras.Sequential([
+    train_data_x = pandas.read_csv('./data_train.csv')
+    _run.info["dataset"] = train_data_x
+    dat_all = train_data_x.copy()
+    dat_predict = train_data_x.pop('Price')
+    normalize.adapt(dat_all)
+    norm_model = tensorflow.keras.Sequential([
         normalize,
         layers.Dense(64),
         layers.Dense(1)
     ])
-
-    price_model.compile(
+    norm_model.compile(
         loss=tensorflow.keras.losses.MeanSquaredError(),
         optimizer=tensorflow.keras.optimizers.Adam())
 
-    price_model.fit(price_train, price_predict, epochs=EPOCHS)
+    norm_model.fit(dat_all, dat_predict, epochs=EPOCHS)
+    counter = 0
+    while counter < 20:
+        counter+=1
+        value = counter * random.randint(5, 5000)
+        _run.log_scalar("training.accuracy", value * 2)
+    norm_model.save('test')
 
-    price_model.save('model')
-
-
-if __name__ == "__main__":
-    EPOCHS = int(os.environ['EPOCHS'])
-    main(EPOCHS)
 
 exint.run()
-exint.add_artifact('saved_model.pb')
+exint.add_artifact('test/saved_model.pb')
 exint.add_source_file('./ium_z487174/train.py')