[TensorFlow] MirroredStrategyを用いて複数GPU計算を行う ref: https://qiita.com/shu-yusa/items/e93e934a14849541de78

file0.txt

distribution = tf.contrib.distribute.MirroredStrategy()
config = tf.estimator.RunConfig(train_distribute=distribution)
classifier = tf.estimator.Estimator(model_fn=model_fn, config=config)

file1.txt

ValueError: dataset_fn() must return a tf.data.Dataset when using a DistributionStrategy.

file2.txt

class InputFnProvider:
  def __init__(self, train_batch_size):
    self.train_batch_size = train_batch_size
    self.__load_data()

  def __load_data(self):
    # Load training and eval data
    mnist = tf.contrib.learn.datasets.load_dataset("mnist")
    self.train_data = mnist.train.images  # Returns np.array
    self.train_labels = np.asarray(mnist.train.labels, dtype=np.int32)
    self.eval_data = mnist.test.images  # Returns np.array
    self.eval_labels = np.asarray(mnist.test.labels, dtype=np.int32)

  def train_input_fn(self):
    """An input function for training"""
    # Shuffle, repeat, and batch the examples.
    dataset = tf.data.Dataset.from_tensor_slices(({"x": self.train_data}, self.train_labels))
    dataset = dataset.shuffle(1000).repeat().batch(self.train_batch_size)
    return dataset

  def eval_input_fn(self):
    """An input function for evaluation or prediction"""
    dataset = tf.data.Dataset.from_tensor_slices(({"x": self.eval_data}, self.eval_labels))
    dataset = dataset.batch(1)
    return dataset

# (中略)

# Train the model
mnist_classifier.train(
    input_fn=input_fn_provider.train_input_fn,
    steps=10000)

# Evaluate the model and print results
eval_results = mnist_classifier.evaluate(input_fn=input_fn_provider.eval_input_fn)

file3.txt

return dataset.make_one_shot_iterator().get_next()