task(ium_04) Tensorflow improvement

main2.py

@@ -1,59 +1,107 @@
-import pandas as pd
-from sklearn.model_selection import train_test_split
-from tensorflow.keras import layers
-import numpy as np
+import re
+import string
+from silence_tensorflow import silence_tensorflow
+silence_tensorflow()
 import tensorflow as tf
+from tensorflow.keras import layers
+from tensorflow.keras import losses
+from tensorflow.keras.layers.experimental.preprocessing import TextVectorization
 
 
-def normalization(label):
-    return 0 if label == False else 1
+def vectorize_text(text, label):
+    text = tf.expand_dims(text, -1)
+    return vectorize_layer(text), label
 
 
-def main():
-    data = pd.read_csv('Amazon_Consumer_Reviews.csv', header=0, sep=',')
-    column_names = ['reviews.doRecommend', 'reviews.title']
-    data = data[column_names]
-
-    data_train, data_test = train_test_split(data, train_size=0.6, random_state=1)
-    data_test, data_val = train_test_split(data_test, test_size=0.5, random_state=1)
-
-    train_labels = [normalization(x) for x in np.array(data_train['reviews.doRecommend'])]
-    train_examples = np.array(data_train['reviews.title'])
-    test_examples = np.array(data_test['reviews.title'])
-    test_labels = [normalization(x) for x in np.array(data_test['reviews.doRecommend'])]
-    val_labels = [normalization(x) for x in np.array(data_val['reviews.doRecommend'])]
-    val_examples = np.array(data_val['reviews.title'])
-
-    # print("Training entries: {}, test entries: {}".format(len(data_train), len(data_test)))
-    # print(train_examples)
-    # print(train_labels)
-
-    model = tf.keras.Sequential([
-        layers.Input(shape=(12,)),
-        layers.Dense(32),
-        layers.Dense(16),
-        layers.Dense(2, activation='softmax')
-    ])
-
-    model.summary()
-
-    model.compile(
-        loss=tf.losses.BinaryCrossentropy(),
-        optimizer=tf.optimizers.Adam(),
-        metrics=[tf.keras.metrics.BinaryAccuracy()])
-
-    history = model.fit(train_examples, train_labels,
-                        epochs=40,
-                        batch_size=512,
-                        validation_data=(val_examples, val_labels),
-                        verbose=1)
-
-    results = model.evaluate(test_examples, test_labels)
-
-    file = open('results.txt', 'w')
-    file.write('test loss: ' + str(results[0]) + '\n' + 'test accuracy: ' + str(results[1]))
-    file.close()
+def custom_standardization(input_data):
+    lowercase = tf.strings.lower(input_data)
+    stripped_html = tf.strings.regex_replace(lowercase, '<br />', ' ')
+    return tf.strings.regex_replace(stripped_html, '[%s]' % re.escape(string.punctuation), '')
 
 
-if __name__ == '__main__':
-    main()
+batch_size = 32
+seed = 42
+
+raw_train_ds = tf.keras.preprocessing.text_dataset_from_directory(
+    'aclImdb/train',
+    batch_size=batch_size,
+    validation_split=0.2,
+    subset='training',
+    seed=seed)
+
+raw_val_ds = tf.keras.preprocessing.text_dataset_from_directory(
+    'aclImdb/train',
+    batch_size=batch_size,
+    validation_split=0.2,
+    subset='validation',
+    seed=seed)
+
+raw_test_ds = tf.keras.preprocessing.text_dataset_from_directory(
+    'aclImdb/test',
+    batch_size=batch_size)
+
+max_features = 10000
+sequence_length = 250
+
+vectorize_layer = TextVectorization(
+    standardize=custom_standardization,
+    max_tokens=max_features,
+    output_mode='int',
+    output_sequence_length=sequence_length)
+
+train_text = raw_train_ds.map(lambda x, y: x)
+vectorize_layer.adapt(train_text)
+
+train_ds = raw_train_ds.map(vectorize_text)
+val_ds = raw_val_ds.map(vectorize_text)
+test_ds = raw_test_ds.map(vectorize_text)
+
+AUTOTUNE = tf.data.AUTOTUNE
+
+train_ds = train_ds.cache().prefetch(buffer_size=AUTOTUNE)
+val_ds = val_ds.cache().prefetch(buffer_size=AUTOTUNE)
+test_ds = test_ds.cache().prefetch(buffer_size=AUTOTUNE)
+
+embedding_dim = 16
+
+model = tf.keras.Sequential([
+    layers.Embedding(max_features + 1, embedding_dim),
+    layers.Dropout(0.2),
+    layers.GlobalAveragePooling1D(),
+    layers.Dropout(0.2),
+    layers.Dense(1)])
+
+model.summary()
+
+model.compile(loss=losses.BinaryCrossentropy(from_logits=True),
+              optimizer='adam',
+              metrics=tf.metrics.BinaryAccuracy(threshold=0.0))
+
+epochs = 10
+history = model.fit(
+    train_ds,
+    validation_data=val_ds,
+    epochs=epochs)
+
+loss, accuracy = model.evaluate(test_ds)
+print("Loss: ", loss)
+print("Accuracy: ", accuracy)
+
+export_model = tf.keras.Sequential([
+    vectorize_layer,
+    model,
+    layers.Activation('sigmoid')
+])
+
+export_model.compile(
+    loss=losses.BinaryCrossentropy(from_logits=False), optimizer="adam", metrics=['accuracy']
+)
+
+loss, accuracy = export_model.evaluate(raw_test_ds)
+print("Loss: ", loss)
+print("Accuracy: ", accuracy)
+
+file = open('results.txt', 'w')
+file.write('test loss: ' + loss + '\n' + 'test accuracy: ' + accuracy)
+file.close()
+