NumPy 데이터 로드

TensorFlow.org에서 보기 Google Colab에서 실행하기 GitHub에서소스 보기 노트북 다운로드하기

이 튜토리얼은 NumPy 배열에서 tf.data.Dataset으로 데이터를 로드하는 예제를 제공합니다.

이 예제에서는 .npz 파일에서 MNIST 데이터세트를 로드합니다. 그러나 NumPy 배열의 소스는 중요하지 않습니다.

설정

import numpy as np
import tensorflow as tf
2022-12-14 21:10:19.807402: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer.so.7'; dlerror: libnvinfer.so.7: cannot open shared object file: No such file or directory
2022-12-14 21:10:19.807509: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer_plugin.so.7'; dlerror: libnvinfer_plugin.so.7: cannot open shared object file: No such file or directory
2022-12-14 21:10:19.807519: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.

.npz 파일에서 로드하기

DATA_URL = 'https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz'

path = tf.keras.utils.get_file('mnist.npz', DATA_URL)
with np.load(path) as data:
  train_examples = data['x_train']
  train_labels = data['y_train']
  test_examples = data['x_test']
  test_labels = data['y_test']

tf.data.Dataset를 사용하여 NumPy 배열 로드하기

예제 배열과 레이블의 해당 배열이 있다고 가정하면, tf.data.Dataset.from_tensor_slices에 튜플로 두 배열을 전달하여 tf.data.Dataset을 만듭니다.

train_dataset = tf.data.Dataset.from_tensor_slices((train_examples, train_labels))
test_dataset = tf.data.Dataset.from_tensor_slices((test_examples, test_labels))

데이터세트 사용하기

데이터세트 셔플 및 배치

BATCH_SIZE = 64
SHUFFLE_BUFFER_SIZE = 100

train_dataset = train_dataset.shuffle(SHUFFLE_BUFFER_SIZE).batch(BATCH_SIZE)
test_dataset = test_dataset.batch(BATCH_SIZE)

모델 빌드 및 훈련

model = tf.keras.Sequential([
    tf.keras.layers.Flatten(input_shape=(28, 28)),
    tf.keras.layers.Dense(128, activation='relu'),
    tf.keras.layers.Dense(10)
])

model.compile(optimizer=tf.keras.optimizers.RMSprop(),
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
              metrics=['sparse_categorical_accuracy'])
model.fit(train_dataset, epochs=10)
Epoch 1/10
938/938 [==============================] - 3s 2ms/step - loss: 3.6223 - sparse_categorical_accuracy: 0.8792
Epoch 2/10
938/938 [==============================] - 2s 2ms/step - loss: 0.5732 - sparse_categorical_accuracy: 0.9266
Epoch 3/10
938/938 [==============================] - 2s 2ms/step - loss: 0.3837 - sparse_categorical_accuracy: 0.9460
Epoch 4/10
938/938 [==============================] - 2s 2ms/step - loss: 0.3161 - sparse_categorical_accuracy: 0.9568
Epoch 5/10
938/938 [==============================] - 2s 2ms/step - loss: 0.2733 - sparse_categorical_accuracy: 0.9628
Epoch 6/10
938/938 [==============================] - 2s 2ms/step - loss: 0.2418 - sparse_categorical_accuracy: 0.9673
Epoch 7/10
938/938 [==============================] - 2s 2ms/step - loss: 0.2175 - sparse_categorical_accuracy: 0.9707
Epoch 8/10
938/938 [==============================] - 2s 2ms/step - loss: 0.2040 - sparse_categorical_accuracy: 0.9726
Epoch 9/10
938/938 [==============================] - 2s 2ms/step - loss: 0.1928 - sparse_categorical_accuracy: 0.9743
Epoch 10/10
938/938 [==============================] - 2s 2ms/step - loss: 0.1807 - sparse_categorical_accuracy: 0.9773
<keras.callbacks.History at 0x7f42daeda730>
model.evaluate(test_dataset)
157/157 [==============================] - 0s 2ms/step - loss: 0.5589 - sparse_categorical_accuracy: 0.9606
[0.5589265823364258, 0.9606000185012817]