1  """Trains and evaluate a simple MLP
 2  on the Reuters newswire topic classification task.
 3  """
 4  
 5  import numpy as np
 6  from tensorflow import keras
 7  from tensorflow.keras.datasets import reuters
 8  from tensorflow.keras.layers import Activation, Dense, Dropout
 9  from tensorflow.keras.models import Sequential
10  from tensorflow.keras.preprocessing.text import Tokenizer
11  
12  # The following import and function call are the only additions to code required
13  # to automatically log metrics and parameters to MLflow.
14  import mlflow
15  
16  mlflow.tensorflow.autolog()
17  
18  max_words = 1000
19  batch_size = 32
20  epochs = 5
21  
22  print("Loading data...")
23  (x_train, y_train), (x_test, y_test) = reuters.load_data(num_words=max_words, test_split=0.2)
24  
25  print(len(x_train), "train sequences")
26  print(len(x_test), "test sequences")
27  
28  num_classes = np.max(y_train) + 1
29  print(num_classes, "classes")
30  
31  print("Vectorizing sequence data...")
32  tokenizer = Tokenizer(num_words=max_words)
33  x_train = tokenizer.sequences_to_matrix(x_train, mode="binary")
34  x_test = tokenizer.sequences_to_matrix(x_test, mode="binary")
35  print("x_train shape:", x_train.shape)
36  print("x_test shape:", x_test.shape)
37  
38  print("Convert class vector to binary class matrix (for use with categorical_crossentropy)")
39  y_train = keras.utils.to_categorical(y_train, num_classes)
40  y_test = keras.utils.to_categorical(y_test, num_classes)
41  print("y_train shape:", y_train.shape)
42  print("y_test shape:", y_test.shape)
43  
44  print("Building model...")
45  model = Sequential()
46  model.add(Dense(512, input_shape=(max_words,)))
47  model.add(Activation("relu"))
48  model.add(Dropout(0.5))
49  model.add(Dense(num_classes))
50  model.add(Activation("softmax"))
51  
52  model.compile(loss="categorical_crossentropy", optimizer="adam", metrics=["accuracy"])
53  
54  history = model.fit(
55      x_train, y_train, batch_size=batch_size, epochs=epochs, verbose=1, validation_split=0.1
56  )
57  score = model.evaluate(x_test, y_test, batch_size=batch_size, verbose=1)
58  print("Test score:", score[0])
59  print("Test accuracy:", score[1])