TensorFlow 2 quickstart untuk tingkat lanjut

Lihat di TensorFlow.org Jalankan di Google Colab Lihat sumber kode di GitHub Unduh notebook

Ini adalah file notebook Google Colaboratory. Program python akan dijalankan langsung dari browser — cara yang bagus untuk mempelajari dan menggunakan TensorFlow. Untuk mengikuti tutorial ini, jalankan notebook di Google Colab dengan mengklik tombol di bagian atas halaman ini.

  1. Di halaman Colab, sambungkan ke runtime Python: Di menu sebelah kanan atas, pilih * CONNECT *.
  2. Untuk menjalankan semua sel kode pada notebook: Pilih * Runtime *> * Run all *.

Download dan instal TensorFlow 2 dan impor TensorFlow ke dalam program Anda:

Impor TensorFlow ke dalam program Anda:

from __future__ import absolute_import, division, print_function, unicode_literals

import tensorflow as tf

from tensorflow.keras.layers import Dense, Flatten, Conv2D
from tensorflow.keras import Model

Siapkan dataset MNIST.

mnist = tf.keras.datasets.mnist

(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0

# Tambahkan dimensi chanel
x_train = x_train[..., tf.newaxis]
x_test = x_test[..., tf.newaxis]
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz
11493376/11490434 [==============================] - 1s 0us/step

Gunakan tf.data untuk mengelompokkan dan mengatur kembali dataset:

train_ds = tf.data.Dataset.from_tensor_slices(
    (x_train, y_train)).shuffle(10000).batch(32)

test_ds = tf.data.Dataset.from_tensor_slices((x_test, y_test)).batch(32)

Buat model tf.keras menggunakan Keras model subclassing API:

class MyModel(Model):
  def __init__(self):
    super(MyModel, self).__init__()
    self.conv1 = Conv2D(32, 3, activation='relu')
    self.flatten = Flatten()
    self.d1 = Dense(128, activation='relu')
    self.d2 = Dense(10, activation='softmax')

  def call(self, x):
    x = self.conv1(x)
    x = self.flatten(x)
    x = self.d1(x)
    return self.d2(x)

# Buat sebuah contoh dari model
model = MyModel()

Pilih fungsi untuk mengoptimalkan dan fungsi untuk menilai loss dari hasil pelatihan:

loss_object = tf.keras.losses.SparseCategoricalCrossentropy()

optimizer = tf.keras.optimizers.Adam()

Pilih metrik untuk mengukur loss dan keakuratan model. Metrik ini mengakumulasi nilai di atas epochs dan kemudian mencetak hasil secara keseluruhan.

train_loss = tf.keras.metrics.Mean(name='train_loss')
train_accuracy = tf.keras.metrics.SparseCategoricalAccuracy(name='train_accuracy')

test_loss = tf.keras.metrics.Mean(name='test_loss')
test_accuracy = tf.keras.metrics.SparseCategoricalAccuracy(name='test_accuracy')

Gunakan tf.GradientTape untuk melatih model:

@tf.function
def train_step(images, labels):
  with tf.GradientTape() as tape:
    predictions = model(images)
    loss = loss_object(labels, predictions)
  gradients = tape.gradient(loss, model.trainable_variables)
  optimizer.apply_gradients(zip(gradients, model.trainable_variables))

  train_loss(loss)
  train_accuracy(labels, predictions)

Tes modelnya:

@tf.function
def test_step(images, labels):
  predictions = model(images)
  t_loss = loss_object(labels, predictions)

  test_loss(t_loss)
  test_accuracy(labels, predictions)
EPOCHS = 5

for epoch in range(EPOCHS):
  for images, labels in train_ds:
    train_step(images, labels)

  for test_images, test_labels in test_ds:
    test_step(test_images, test_labels)

  template = 'Epoch {}, Loss: {}, Accuracy: {}, Test Loss: {}, Test Accuracy: {}'
  print(template.format(epoch+1,
                        train_loss.result(),
                        train_accuracy.result()*100,
                        test_loss.result(),
                        test_accuracy.result()*100))

  # Menghitung ulang metrik untuk epoch selanjutnya
  train_loss.reset_states()
  train_accuracy.reset_states()
  test_loss.reset_states()
  test_accuracy.reset_states()
WARNING:tensorflow:Layer my_model is casting an input tensor from dtype float64 to the layer's dtype of float32, which is new behavior in TensorFlow 2.  The layer has dtype float32 because it's dtype defaults to floatx.

If you intended to run this layer in float32, you can safely ignore this warning. If in doubt, this warning is likely only an issue if you are porting a TensorFlow 1.X model to TensorFlow 2.

To change all layers to have dtype float64 by default, call `tf.keras.backend.set_floatx('float64')`. To change just this layer, pass dtype='float64' to the layer constructor. If you are the author of this layer, you can disable autocasting by passing autocast=False to the base Layer constructor.

Epoch 1, Loss: 0.13782905042171478, Accuracy: 95.9000015258789, Test Loss: 0.060800302773714066, Test Accuracy: 97.94999694824219
Epoch 2, Loss: 0.04384945333003998, Accuracy: 98.6199951171875, Test Loss: 0.0552654005587101, Test Accuracy: 98.33999633789062
Epoch 3, Loss: 0.022658096626400948, Accuracy: 99.25, Test Loss: 0.055065643042325974, Test Accuracy: 98.29999542236328
Epoch 4, Loss: 0.013757845386862755, Accuracy: 99.54000091552734, Test Loss: 0.05910263583064079, Test Accuracy: 98.38999938964844
Epoch 5, Loss: 0.010206885635852814, Accuracy: 99.64666748046875, Test Loss: 0.05727027729153633, Test Accuracy: 98.45999908447266

Penggolong gambar tersebut, sekarang dilatih untuk akurasi ~ 98% pada dataset ini. Untuk mempelajari lebih lanjut, baca tutorial TensorFlow.