tensorflow——cifar10 样例解析

从 tensorflow 提供的 cifar10 分类程序样例学习程序的主要框架

最外层结构：

• 利用 tf.app.flags 解析命令行参数
• 利用 tf.app.run 启动主函数，在主函数中训练

import tensorflow as tf


FLAGS = tf.app.flags.FLAGS
tf.app.flags.DEFINE_string('train_dir', '/tmp/cifar10_train',
                           """Directory where to write event logs """
                           """and checkpoint.""")
tf.app.flags.DEFINE_integer('max_steps', 100000,
                            """Number of batches to run.""")
tf.app.flags.DEFINE_boolean('log_device_placement', False,
                            """Whether to log device placement.""")
tf.app.flags.DEFINE_integer('log_frequency', 10,
                            """How often to log results to the console.""")


def train():
	pass

def main(argv=None):  # pylint: disable=unused-argument
	cifar10.maybe_download_and_extract()
  	if tf.gfile.Exists(FLAGS.train_dir):
    	tf.gfile.DeleteRecursively(FLAGS.train_dir)
 	tf.gfile.MakeDirs(FLAGS.train_dir)
 	train()


if __name__ == '__main__':
  tf.app.run()

下面来看训练过程：搭建计算流图，计算 loss，选择合适的优化器以减小 loss 为目标来更新权重，周期性的记录日志。

def train():
  """Train CIFAR-10 for a number of steps."""
  with tf.Graph().as_default():
    global_step = tf.train.get_or_create_global_step()

    # 强制使用 CPU
    with tf.device('/cpu:0'):
      images, labels = cifar10.distorted_inputs()

    # 这里的 logits 是计算图中的计算 softmax 之前的节点
    logits = cifar10.inference(images)

    # 这里的 loss 是计算 loss 的节点
    loss = cifar10.loss(logits, labels)

    # 这个是利用反向传播更新网络权值的计算节点
    train_op = cifar10.train(loss, global_step)

	# 这里通过重载 SessionRunHook 这个运行时的回调函数，实现周期写日志的功能
    class _LoggerHook(tf.train.SessionRunHook):
      """Logs loss and runtime."""

      def begin(self):
        self._step = -1
        self._start_time = time.time()

      def before_run(self, run_context):
        self._step += 1
        return tf.train.SessionRunArgs(loss)  # Asks for loss value.

      def after_run(self, run_context, run_values):
        if self._step % FLAGS.log_frequency == 0:
          current_time = time.time()
          duration = current_time - self._start_time
          self._start_time = current_time

          loss_value = run_values.results
          examples_per_sec = FLAGS.log_frequency * FLAGS.batch_size / duration
          sec_per_batch = float(duration / FLAGS.log_frequency)

          format_str = ('%s: step %d, loss = %.2f (%.1f examples/sec; %.3f '
                        'sec/batch)')
          print (format_str % (datetime.now(), self._step, loss_value,
                               examples_per_sec, sec_per_batch))

    with tf.train.MonitoredTrainingSession(
        checkpoint_dir=FLAGS.train_dir,
        # 一系列回调函数
        hooks=[tf.train.StopAtStepHook(last_step=FLAGS.max_steps), #达到最大步长限制
               tf.train.NanTensorHook(loss), # loss 发散
               _LoggerHook()], # 周期写日志
        config=tf.ConfigProto(
            log_device_placement=FLAGS.log_device_placement)) as mon_sess:
      while not mon_sess.should_stop():
        mon_sess.run(train_op)

上面的代码中使用回调函数来写日志，看起来比较高级。对于初学者来说，可能更习惯于把 log 的代码直接加在循环里，向下面这样，就不用多写一个内部类_LoggerHook：

def train():
  """Train CIFAR-10 for a number of steps."""
  with tf.Graph().as_default():
    global_step = tf.train.get_or_create_global_step()

    # 强制使用 CPU
    with tf.device('/cpu:0'):
      images, labels = cifar10.distorted_inputs()

    # 这里的 logits 是计算图中的计算 softmax 之前的节点
    logits = cifar10.inference(images)

    # 这里的 loss 是计算 loss 的节点
    loss = cifar10.loss(logits, labels)

    # 这个是利用反向传播更新网络权值的计算节点
    train_op = cifar10.train(loss, global_step)

    with tf.train.MonitoredTrainingSession(
        checkpoint_dir=FLAGS.train_dir,
        hooks=[tf.train.StopAtStepHook(last_step=FLAGS.max_steps),
               tf.train.NanTensorHook(loss)],  # 这里把写日志的回调函数删除了
        config=tf.ConfigProto(
            log_device_placement=FLAGS.log_device_placement)) as mon_sess:
        step = -1
        start_time = time.time()
        while not mon_sess.should_stop():
            step += 1
            lss,_ = mon_sess.run([loss,train_op])
            # 直接在 while 循环里写日志， 效果和回调函数是一样的
            if step % FLAGS.log_frequency == 0:
                current_time = time.time()
                duration = current_time - start_time
                start_time = current_time

                examples_per_sec = FLAGS.log_frequency * FLAGS.batch_size / duration
                sec_per_batch = float(duration / FLAGS.log_frequency)

                format_str = ('%s: step %d, loss = %.2f (%.1f examples/sec; %.3f '
                              'sec/batch)')
                print(format_str % (datetime.now(), step, lss,
                                    examples_per_sec, sec_per_batch))

记录的日志如下：

2019-03-19 19:04:37.806384: step 0, loss = 4.67 (260.6 examples/sec; 0.491 sec/batch)
2019-03-19 19:04:42.597658: step 10, loss = 4.62 (267.2 examples/sec; 0.479 sec/batch)
2019-03-19 19:04:47.308927: step 20, loss = 4.49 (271.7 examples/sec; 0.471 sec/batch)
...

最后深入到计算流图的细节，看看每一个节点是怎么搭建的（cifar10.py）：
（源代码中关于summary，use_fp16 等细枝末节已被剔除，不影响主干程序）

"""Builds the CIFAR-10 network.

Summary of available functions:

 # Compute input images and labels for training. If you would like to run
 # evaluations, use inputs() instead.
 inputs, labels = distorted_inputs()

 # Compute inference on the model inputs to make a prediction.
 predictions = inference(inputs)

 # Compute the total loss of the prediction with respect to the labels.
 loss = loss(predictions, labels)

 # Create a graph to run one step of training with respect to the loss.
 train_op = train(loss, global_step)
"""

import os
import re
import sys
import tarfile

from six.moves import urllib
import tensorflow as tf

import cifar10_input

FLAGS = tf.app.flags.FLAGS
# Basic model parameters.
tf.app.flags.DEFINE_integer('batch_size', 128,
                            """Number of images to process in a batch.""")
tf.app.flags.DEFINE_string('data_dir', '/tmp/cifar10_data',
                           """Path to the CIFAR-10 data directory.""")


# Global constants describing the CIFAR-10 data set.
IMAGE_SIZE = 24
NUM_CLASSES = 10
NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN = 50000
NUM_EXAMPLES_PER_EPOCH_FOR_EVAL = 10000


# Constants describing the training process.
MOVING_AVERAGE_DECAY = 0.9999     # The decay to use for the moving average.
NUM_EPOCHS_PER_DECAY = 350.0      # Epochs after which learning rate decays.
LEARNING_RATE_DECAY_FACTOR = 0.1  # Learning rate decay factor.
INITIAL_LEARNING_RATE = 0.1       # Initial learning rate.

# If a model is trained with multiple GPUs, prefix all Op names with tower_name
# to differentiate the operations. Note that this prefix is removed from the
# names of the summaries when visualizing a model.
TOWER_NAME = 'tower'

DATA_URL = 'https://www.cs.toronto.edu/~kriz/cifar-10-binary.tar.gz'


def _variable_on_cpu(name, shape, initializer):
  """Helper to create a Variable stored on CPU memory.

  Args:
    name: name of the variable
    shape: list of ints
    initializer: initializer for Variable

  Returns:
    Variable Tensor
  """
  with tf.device('/cpu:0'):
    dtype = tf.float16 if FLAGS.use_fp16 else tf.float32
    var = tf.get_variable(name, shape, initializer=initializer, dtype=dtype)
  return var


def _variable_with_weight_decay(name, shape, stddev, wd):
  """Helper to create an initialized Variable with weight decay.

  Note that the Variable is initialized with a truncated normal distribution.
  A weight decay is added only if one is specified.

  Args:
    name: name of the variable
    shape: list of ints
    stddev: standard deviation of a truncated Gaussian
    wd: add L2Loss weight decay multiplied by this float. If None, weight
        decay is not added for this Variable.

  Returns:
    Variable Tensor
  """
  dtype = tf.float16 if FLAGS.use_fp16 else tf.float32
  var = _variable_on_cpu(
      name,
      shape,
      tf.truncated_normal_initializer(stddev=stddev, dtype=dtype))
  if wd is not None:
    weight_decay = tf.multiply(tf.nn.l2_loss(var), wd, name='weight_loss')
    tf.add_to_collection('losses', weight_decay)
  return var


def distorted_inputs():
  """Construct distorted input for CIFAR training using the Reader ops.

  Returns:
    images: Images. 4D tensor of [batch_size, IMAGE_SIZE, IMAGE_SIZE, 3] size.
    labels: Labels. 1D tensor of [batch_size] size.

  Raises:
    ValueError: If no data_dir
  """
  pass
  return images, labels


def inputs(eval_data):
  """Construct input for CIFAR evaluation using the Reader ops.

  Args:
    eval_data: bool, indicating if one should use the train or eval data set.

  Returns:
    images: Images. 4D tensor of [batch_size, IMAGE_SIZE, IMAGE_SIZE, 3] size.
    labels: Labels. 1D tensor of [batch_size] size.

  Raises:
    ValueError: If no data_dir
  """
  pass
  return images, labels

# CNN 的 核心结构
def inference(images):
  """Build the CIFAR-10 model.

  Args:
    images: Images returned from distorted_inputs() or inputs().

  Returns:
    Logits.
  """
  # We instantiate all variables using tf.get_variable() instead of
  # tf.Variable() in order to share variables across multiple GPU training runs.
  # If we only ran this model on a single GPU, we could simplify this function
  # by replacing all instances of tf.get_variable() with tf.Variable().
  #
  # conv1
  with tf.variable_scope('conv1') as scope:
    kernel = _variable_with_weight_decay('weights',
                                         shape=[5, 5, 3, 64],
                                         stddev=5e-2,
                                         wd=None)
    conv = tf.nn.conv2d(images, kernel, [1, 1, 1, 1], padding='SAME')
    biases = _variable_on_cpu('biases', [64], tf.constant_initializer(0.0))
    pre_activation = tf.nn.bias_add(conv, biases)
    conv1 = tf.nn.relu(pre_activation, name=scope.name)

  # pool1
  pool1 = tf.nn.max_pool(conv1, ksize=[1, 3, 3, 1], strides=[1, 2, 2, 1],
                         padding='SAME', name='pool1')
  # norm1
  norm1 = tf.nn.lrn(pool1, 4, bias=1.0, alpha=0.001 / 9.0, beta=0.75,
                    name='norm1')

  # conv2
  with tf.variable_scope('conv2') as scope:
    kernel = _variable_with_weight_decay('weights',
                                         shape=[5, 5, 64, 64],
                                         stddev=5e-2,
                                         wd=None)
    conv = tf.nn.conv2d(norm1, kernel, [1, 1, 1, 1], padding='SAME')
    biases = _variable_on_cpu('biases', [64], tf.constant_initializer(0.1))
    pre_activation = tf.nn.bias_add(conv, biases)
    conv2 = tf.nn.relu(pre_activation, name=scope.name)

  # norm2
  norm2 = tf.nn.lrn(conv2, 4, bias=1.0, alpha=0.001 / 9.0, beta=0.75,
                    name='norm2')
  # pool2
  pool2 = tf.nn.max_pool(norm2, ksize=[1, 3, 3, 1],
                         strides=[1, 2, 2, 1], padding='SAME', name='pool2')

  # local3
  with tf.variable_scope('local3') as scope:
    # Move everything into depth so we can perform a single matrix multiply.
    reshape = tf.reshape(pool2, [images.get_shape().as_list()[0], -1])
    dim = reshape.get_shape()[1].value
    weights = _variable_with_weight_decay('weights', shape=[dim, 384],
                                          stddev=0.04, wd=0.004)
    biases = _variable_on_cpu('biases', [384], tf.constant_initializer(0.1))
    local3 = tf.nn.relu(tf.matmul(reshape, weights) + biases, name=scope.name)

  # local4
  with tf.variable_scope('local4') as scope:
    weights = _variable_with_weight_decay('weights', shape=[384, 192],
                                          stddev=0.04, wd=0.004)
    biases = _variable_on_cpu('biases', [192], tf.constant_initializer(0.1))
    local4 = tf.nn.relu(tf.matmul(local3, weights) + biases, name=scope.name)

  # linear layer(WX + b),
  # We don't apply softmax here because
  # tf.nn.sparse_softmax_cross_entropy_with_logits accepts the unscaled logits
  # and performs the softmax internally for efficiency.
  with tf.variable_scope('softmax_linear') as scope:
    weights = _variable_with_weight_decay('weights', [192, NUM_CLASSES],
                                          stddev=1/192.0, wd=None)
    biases = _variable_on_cpu('biases', [NUM_CLASSES],
                              tf.constant_initializer(0.0))
    softmax_linear = tf.add(tf.matmul(local4, weights), biases, name=scope.name)

  return softmax_linear


def loss(logits, labels):
  """Add L2Loss to all the trainable variables.

  Add summary for "Loss" and "Loss/avg".
  Args:
    logits: Logits from inference().
    labels: Labels from distorted_inputs or inputs(). 1-D tensor
            of shape [batch_size]

  Returns:
    Loss tensor of type float.
  """
  # Calculate the average cross entropy loss across the batch.
  labels = tf.cast(labels, tf.int64)
  cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(
      labels=labels, logits=logits, name='cross_entropy_per_example')
  cross_entropy_mean = tf.reduce_mean(cross_entropy, name='cross_entropy')
  tf.add_to_collection('losses', cross_entropy_mean)

  # The total loss is defined as the cross entropy loss plus all of the weight
  # decay terms (L2 loss).
  return tf.add_n(tf.get_collection('losses'), name='total_loss')


def _add_loss_summaries(total_loss):
  """Add summaries for losses in CIFAR-10 model.

  Generates moving average for all losses and associated summaries for
  visualizing the performance of the network.

  Args:
    total_loss: Total loss from loss().
  Returns:
    loss_averages_op: op for generating moving averages of losses.
  """
  # Compute the moving average of all individual losses and the total loss.
  loss_averages = tf.train.ExponentialMovingAverage(0.9, name='avg')
  losses = tf.get_collection('losses')
  loss_averages_op = loss_averages.apply(losses + [total_loss])

  return loss_averages_op


def train(total_loss, global_step):
  """Train CIFAR-10 model.

  Create an optimizer and apply to all trainable variables. Add moving
  average for all trainable variables.

  Args:
    total_loss: Total loss from loss().
    global_step: Integer Variable counting the number of training steps
      processed.
  Returns:
    train_op: op for training.
  """
  # Variables that affect learning rate.
  num_batches_per_epoch = NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN / FLAGS.batch_size
  decay_steps = int(num_batches_per_epoch * NUM_EPOCHS_PER_DECAY)

  # Decay the learning rate exponentially based on the number of steps.
  lr = tf.train.exponential_decay(INITIAL_LEARNING_RATE,
                                  global_step,
                                  decay_steps,
                                  LEARNING_RATE_DECAY_FACTOR,
                                  staircase=True)

  # Generate moving averages of all losses and associated summaries.
  loss_averages_op = _add_loss_summaries(total_loss)

  # Compute gradients.
  with tf.control_dependencies([loss_averages_op]):
    opt = tf.train.GradientDescentOptimizer(lr)
    grads = opt.compute_gradients(total_loss)

  # Apply gradients.
  apply_gradient_op = opt.apply_gradients(grads, global_step=global_step)


  # Track the moving averages of all trainable variables.
  variable_averages = tf.train.ExponentialMovingAverage(
      MOVING_AVERAGE_DECAY, global_step)
  with tf.control_dependencies([apply_gradient_op]):
    variables_averages_op = variable_averages.apply(tf.trainable_variables())

  return variables_averages_op


def maybe_download_and_extract():
  """Download and extract the tarball from Alex's website."""
 pass