实现单元测试
测试代码可以加快原型设计速度，提高调试效率，加快更改速度，并且可以更轻松地共享代码。在 TensorFlow 中有许多简单的方法可以实现单元测试，我们将在本文中介绍它们。
做好准备
在编写 TensorFlow 模型时，有助于进行单元测试以检查程序的功能。这有助于我们，因为当我们想要对程序单元进行更改时，测试将确保这些更改不会以未知方式破坏模型。在这个秘籍中，我们将创建一个依赖于MNIST数据的简单 CNN 网络。有了它，我们将实现三种不同类型的单元测试来说明如何在 TensorFlow 中编写它们。
请注意，Python 有一个很棒的测试库，名为 Nose。 TensorFlow 还具有内置测试功能，我们将在其中查看，这样可以更轻松地测试 Tensor 对象的值，而无需评估会话中的值。
- 首先，我们需要加载必要的库并格式化数据，如下所示：
import sys
import numpy as np
import tensorflow as tf
from tensorflow.python.framework import ops
ops.reset_default_graph()
# Start a graph session
sess = tf.Session()
# Load data
data_dir = 'temp'
mnist = tf.keras.datasets.mnist
(train_xdata, train_labels), (test_xdata, test_labels) = mnist.load_data()
train_xdata = train_xdata / 255.0
test_xdata = test_xdata / 255.0
# Set model parameters
batch_size = 100
learning_rate = 0.005
evaluation_size = 100
image_width = train_xdata[0].shape[0]
image_height = train_xdata[0].shape[1]
target_size = max(train_labels) + 1
num_channels = 1 # greyscale = 1 channel
generations = 100
eval_every = 5
conv1_features = 25
conv2_features = 50
max_pool_size1 = 2 # NxN window for 1st max pool layer
max_pool_size2 = 2 # NxN window for 2nd max pool layer
fully_connected_size1 = 100
dropout_prob = 0.75
- 然后，我们需要声明我们的占位符，变量和模型公式，如下所示：
# Declare model placeholders
x_input_shape = (batch_size, image_width, image_height, num_channels)
x_input = tf.placeholder(tf.float32, shape=x_input_shape)
y_target = tf.placeholder(tf.int32, shape=(batch_size))
eval_input_shape = (evaluation_size, image_width, image_height, num_channels)
eval_input = tf.placeholder(tf.float32, shape=eval_input_shape)
eval_target = tf.placeholder(tf.int32, shape=(evaluation_size))
dropout = tf.placeholder(tf.float32, shape=())
# Declare model parameters
conv1_weight = tf.Variable(tf.truncated_normal([4, 4, num_channels, conv1_features],
stddev=0.1, dtype=tf.float32))
conv1_bias = tf.Variable(tf.zeros([conv1_features], dtype=tf.float32))
conv2_weight = tf.Variable(tf.truncated_normal([4, 4, conv1_features, conv2_features],
stddev=0.1, dtype=tf.float32))
conv2_bias = tf.Variable(tf.zeros([conv2_features], dtype=tf.float32))
# fully connected variables
resulting_width = image_width // (max_pool_size1 * max_pool_size2)
resulting_height = image_height // (max_pool_size1 * max_pool_size2)
full1_input_size = resulting_width * resulting_height * conv2_features
full1_weight = tf.Variable(tf.truncated_normal([full1_input_size, fully_connected_size1],
stddev=0.1, dtype=tf.float32))
full1_bias = tf.Variable(tf.truncated_normal([fully_connected_size1], stddev=0.1, dtype=tf.float32))
full2_weight = tf.Variable(tf.truncated_normal([fully_connected_size1, target_size],
stddev=0.1, dtype=tf.float32))
full2_bias = tf.Variable(tf.truncated_normal([target_size], stddev=0.1, dtype=tf.float32))
# Initialize Model Operations
def my_conv_net(input_data):
# First Conv-ReLU-MaxPool Layer
conv1 = tf.nn.conv2d(input_data, conv1_weight, strides=[1, 1, 1, 1], padding='SAME')
relu1 = tf.nn.relu(tf.nn.bias_add(conv1, conv1_bias))
max_pool1 = tf.nn.max_pool(relu1, ksize=[1, max_pool_size1, max_pool_size1, 1],
strides=[1, max_pool_size1, max_pool_size1, 1], padding='SAME')
# Second Conv-ReLU-MaxPool Layer
conv2 = tf.nn.conv2d(max_pool1, conv2_weight, strides=[1, 1, 1, 1], padding='SAME')
relu2 = tf.nn.relu(tf.nn.bias_add(conv2, conv2_bias))
max_pool2 = tf.nn.max_pool(relu2, ksize=[1, max_pool_size2, max_pool_size2, 1],
strides=[1, max_pool_size2, max_pool_size2, 1], padding='SAME')
# Transform Output into a 1xN layer for next fully connected layer
final_conv_shape = max_pool2.get_shape().as_list()
final_shape = final_conv_shape[1] * final_conv_shape[2] * final_conv_shape[3]
flat_output = tf.reshape(max_pool2, [final_conv_shape[0], final_shape])
# First Fully Connected Layer
fully_connected1 = tf.nn.relu(tf.add(tf.matmul(flat_output, full1_weight), full1_bias))
# Second Fully Connected Layer
final_model_output = tf.add(tf.matmul(fully_connected1, full2_weight), full2_bias)
# Add dropout
final_model_output = tf.nn.dropout(final_model_output, dropout)
return final_model_output
model_output = my_conv_net(x_input)
test_model_output = my_conv_net(eval_input)
- 接下来，我们创建我们的损失函数以及我们的预测和精确操作。然后，我们初始化以下模型变量：
# Declare Loss Function (softmax cross entropy)
loss = tf.reduce_mean(tf.nn.sparse_softmax_cross_entropy_with_logits(model_output, y_target))
# Create a prediction function
prediction = tf.nn.softmax(model_output)
test_prediction = tf.nn.softmax(test_model_output)
# Create accuracy function
def get_accuracy(logits, targets):
batch_predictions = np.argmax(logits, axis=1)
num_correct = np.sum(np.equal(batch_predictions, targets))
return 100\. * num_correct/batch_predictions.shape[0]
# Create an optimizer
my_optimizer = tf.train.MomentumOptimizer(learning_rate, 0.9)
train_step = my_optimizer.minimize(loss)
# Initialize Variables
init = tf.global_variables_initializer()
sess.run(init)
- 对于我们的第一个单元测试，我们使用类
tf.test.TestCase并创建一种方法来测试占位符（或变量）的值。对于此测试用例，我们确保损失概率（用于保持）大于0.25，因此模型不会更改为尝试训练超过 75％的损失，如下所示：
# Check values of tensors!
class DropOutTest(tf.test.TestCase):
# Make sure that we don't drop too much
def dropout_greaterthan(self):
with self.test_session():
self.assertGreater(dropout.eval(), 0.25)
- 接下来，我们需要测试我们的
accuracy函数是否按预期运行。为此，我们创建一个概率样本数组和我们期望的样本，然后确保测试精度返回 100％，如下所示：
# Test accuracy function
class AccuracyTest(tf.test.TestCase):
# Make sure accuracy function behaves correctly
def accuracy_exact_test(self):
with self.test_session():
test_preds = [[0.9, 0.1],[0.01, 0.99]]
test_targets = [0, 1]
test_acc = get_accuracy(test_preds, test_targets)
self.assertEqual(test_acc.eval(), 100.)
- 我们还可以确保 Tensor 对象是我们期望的形状。要通过
target_size测试模型输出是batch_size的预期形状，请输入以下代码：
# Test tensorshape
class ShapeTest(tf.test.TestCase):
# Make sure our model output is size [batch_size, num_classes]
def output_shape_test(self):
with self.test_session():
numpy_array = np.ones([batch_size, target_size])
self.assertShapeEqual(numpy_array, model_output)
- 现在我们需要在脚本中使用
main()函数告诉 TensorFlow 我们正在运行哪个应用。脚本如下：
def main(argv):
# Start training loop
train_loss = []
train_acc = []
test_acc = []
for i in range(generations):
rand_index = np.random.choice(len(train_xdata), size=batch_size)
rand_x = train_xdata[rand_index]
rand_x = np.expand_dims(rand_x, 3)
rand_y = train_labels[rand_index]
train_dict = {x_input: rand_x, y_target: rand_y, dropout: dropout_prob}
sess.run(train_step, feed_dict=train_dict)
temp_train_loss, temp_train_preds = sess.run([loss, prediction], feed_dict=train_dict)
temp_train_acc = get_accuracy(temp_train_preds, rand_y)
if (i + 1) % eval_every == 0:
eval_index = np.random.choice(len(test_xdata), size=evaluation_size)
eval_x = test_xdata[eval_index]
eval_x = np.expand_dims(eval_x, 3)
eval_y = test_labels[eval_index]
test_dict = {eval_input: eval_x, eval_target: eval_y, dropout: 1.0}
test_preds = sess.run(test_prediction, feed_dict=test_dict)
temp_test_acc = get_accuracy(test_preds, eval_y)
# Record and print results
train_loss.append(temp_train_loss)
train_acc.append(temp_train_acc)
test_acc.append(temp_test_acc)
acc_and_loss = [(i + 1), temp_train_loss, temp_train_acc, temp_test_acc]
acc_and_loss = [np.round(x, 2) for x in acc_and_loss]
print('Generation # {}. Train Loss: {:.2f}. Train Acc (Test Acc): {:.2f}
({:.2f})'.format(*acc_and_loss))
- 要让我们的脚本执行测试或训练，我们需要以不同的方式从命令行调用它。以下代码段是主程序代码。如果程序收到参数
test，它将执行测试;否则，它将运行训练：
if __name__ == '__main__':
cmd_args = sys.argv
if len(cmd_args) > 1 and cmd_args[1] == 'test':
# Perform unit-tests
tf.test.main(argv=cmd_args[1:])
else:
# Run the TensorFlow app
tf.app.run(main=None, argv=cmd_args)
- 如果我们在命令行上运行程序，我们应该得到以下输出：
$ python3 implementing_unit_tests.py test
...
----------------------------------------------------------------------
Ran 3 tests in 0.001s
OK
前面步骤中描述的完整程序可以在 h ttps：//github.com/nfmcclure/tensorflow_cookbook/ 的书籍 GitHub 仓库和 Packt 仓库中找到： https://github.com/PacktPublishing/TensorFlow-Machine-Learning-Cookbook-Second-Edition 。
工作原理
在本节中，我们实现了三种类型的单元测试：张量值，操作输出和张量形状。 TensorFlow 有更多类型的单元测试函数，可在此处找到： https://www.tensorflow.org/versions/master/api_docs/python/test.html 。
请记住，单元测试有助于确保代码能够按预期运行，为共享代码提供信心，并使再现性更易于访问。