简单的CNN训练准确度太低的问题。

Question

开始我以为是我的函数优化和配置的问题，后来我找到了《Python 深度学习》上的代码直接运行，结果发现准确度还是只有百分之十几，和书上说的
97.8%相差太远，甚至远低于只使用Dense层的神经网络，问题到底出在哪里？
平台：Windows 虚拟环境 Python3.7 Keras 2.3.1 Tensorflow 2.1.0 CUDA 10.1
硬件：RTX3070 Laptop 115 W

from keras import  layers, models
from keras.datasets import mnist
from keras.utils import to_categorical

(train_images, train_labels), (test_images, test_labels) = mnist.load_data()

train_images = train_images.reshape((60000, 28, 28, 1))
train_images = train_images.astype('float32') / 255

test_images = test_images.reshape((10000, 28, 28, 1))
test_images = test_images.astype('float32') / 255

train_labels = to_categorical(train_labels)
test_labels = to_categorical(test_labels)

model = models.Sequential()
model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1)))
model.add(layers.MaxPool2D(2, 2))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.MaxPool2D(2, 2))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.Flatten())
model.add(layers.Dense(64, activation='relu'))
model.add(layers.Dense(10, activation='softmax'))
model.compile(optimizer='rmsprop',
              loss='categorical_crossentropy',
              metrics=['accuracy'])
history = model.fit(train_images, train_labels, epochs=2, batch_size=64)


test_loss, test_acc = model.evaluate(test_images, test_labels)

运行结果：

52992/60000 [=========================>....] - ETA: 0s - loss: 2.3026 - accuracy: 0.1119
54592/60000 [==========================>...] - ETA: 0s - loss: 2.3026 - accuracy: 0.1118
56128/60000 [===========================>..] - ETA: 0s - loss: 2.3025 - accuracy: 0.1121
57728/60000 [===========================>..] - ETA: 0s - loss: 2.3025 - accuracy: 0.1123
59200/60000 [============================>.] - ETA: 0s - loss: 2.3026 - accuracy: 0.1123
60000/60000 [==============================] - 2s 41us/step - loss: 2.3026 - accuracy: 0.1122

进程已结束，退出代码为 0

Answer 1

迭代次数是不是有点少了