1、准备mnists数据集,其方法有两种:
第一种使用tensorflow的接口函数--input_data.read_data_sets(),其会警告:提示要自己实现下载函数,但是其是可以实现下载数据的,其需要翻墙。否则会报错退出。
第二种自己手动到LeCun官网上下在数据集,其网站是:mnist数据集。把里面的四个文件都下载下来
注:把下载好的数据放到相应的文件夹下。
然后代码相应的代码如下:
import os
import tensorflow as tf
import urllib.request #python3使用的接口方法。python2使用的是另一种方法
import tensorflow.examples.tutorials.mnist.input_data as input_data
#这个文件路径在开头不能有‘/’,否则会路径出错
LOGDIR = 'tmp/mnist_tutorial/'
GITHUB_URL = 'https://raw.githubusercontent.com/mamcgrath/TensorBoard-TF-Dev-Summit-Tutorial/master/'
LOGDIR = 'tmp/mnist_tutorial/'
### MNIST EMBEDDINGS ###
mnist = input_data.read_data_sets(LOGDIR, one_hot=True)
### Get a sprite and labels file for the embedding projector ###
urllib.request.urlretrieve(GITHUB_URL + 'labels_1024.tsv', LOGDIR + 'labels_1024.tsv')
urllib.request.urlretrieve(GITHUB_URL + 'sprite_1024.png', LOGDIR + 'sprite_1024.png')
def conv_layer(input, size_in, size_out, name="conv"):
# tf.name_scope creates namespace for operators in the default graph , places into group, easier to read
# A graph maintains a stack of name scopes. A `with name_scope(...):`
# statement pushes a new name onto the stack for the lifetime of the context.
# Ops have names, name scopes group ops
# 为下面创建的变量添加一个命名空间,其类似于C++里的命名空间,还有一个是variable_scope命名空间。其跟name_scope有区别,具体看相应博客
with tf.name_scope(name):
# 其中为w、b进行初始化为适合的数值,这样子有助于网络的训练,否则容易出现梯度消失的现象。这里使用tensorflow自带的初始化函数接口
w = tf.Variable(tf.truncated_normal([5, 5, size_in, size_out], stddev=0.1), name="W")
b = tf.Variable(tf.constant(0.1, shape=[size_out]), name="B")
conv = tf.nn.conv2d(input, w, strides=[1, 1, 1, 1], padding="SAME")
act = tf.nn.relu(conv + b)
# collect this data by attaching tf.summary.histogram ops to the gradient outputs and to the variable that holds weights, respectively.
# visualize the the distribution of weights and biases
tf.summary.histogram("weights", w)
tf.summary.histogram("biases", b)
tf.summary.histogram("activations", act)
return tf.nn.max_pool(act, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding="SAME")
def fc_layer(input, size_in, size_out, name="fc"):
with tf.name_scope(name):
w = tf.Variable(tf.truncated_normal([size_in, size_out], stddev=0.1), name="W")
b = tf.Variable(tf.constant(0.1, shape=[size_out]), name="B")
act = tf.nn.relu(tf.matmul(input, w) + b)
tf.summary.histogram("weights", w)
tf.summary.histogram("biases", b)
tf.summary.histogram("activations", act)
return act
def mnist_model(learning_rate, use_two_conv, use_two_fc, hparam):
tf.reset_default_graph()
# 这里时设置模型训练占用gpu的大小。
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.6)
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
# Setup placeholders, and reshape the data
# 为变量申请占位符,相当于申请内存
x = tf.placeholder(tf.float32, shape=[None, 784], name="x")
x_image = tf.reshape(x, [-1, 28, 28, 1])
# Outputs a Summary protocol buffer with 3 images.
tf.summary.image('input', x_image, 3)
y = tf.placeholder(tf.float32, shape=[None, 10], name="labels")
if use_two_conv:
conv1 = conv_layer(x_image, 1, 32, "conv1")
conv_out = conv_layer(conv1, 32, 64, "conv2")
else:
conv1 = conv_layer(x_image, 1, 64, "conv")
conv_out = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding="SAME")
# 这里是为了是卷积输出的特征图转换为一个两维的数据,从而变成适合后面全连层的输入数据
flattened = tf.reshape(conv_out, [-1, 7 * 7 * 64])
if use_two_fc:
fc1 = fc_layer(flattened, 7 * 7 * 64, 1024, "fc1")
# we want these embeeddings to visualize them later
#把倒数第二层全连层的输出特征添加到tensorboard的embedding里。到时候可以对这些特征进行一些可视化操作,
#一般这一层已经有样本分类功能
embedding_input = fc1
embedding_size = 1024
logits = fc_layer(fc1, 1024, 10, "fc2")
else:
embedding_input = flattened
embedding_size = 7 * 7 * 64
logits = fc_layer(flattened, 7 * 7 * 64, 10, "fc")
with tf.name_scope("xent"):
xent = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(
logits=logits, labels=y), name="xent")
# save that single number
tf.summary.scalar("xent", xent)
with tf.name_scope("train"):
#这个是使用比较多的网络权重优化器
train_step = tf.train.AdamOptimizer(learning_rate).minimize(xent)
with tf.name_scope("accuracy"):
correct_prediction = tf.equal(tf.argmax(logits, 1), tf.argmax(y, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
tf.summary.scalar("accuracy", accuracy)
# Merges all summaries collected in the default graph
summ = tf.summary.merge_all()
# intiialize embedding matrix as 0s
embedding = tf.Variable(tf.zeros([1024, embedding_size]), name="test_embedding")
# give it calculated embedding
assignment = embedding.assign(embedding_input)
# initialize the saver
# Add ops to save and restore all the variables.
saver = tf.train.Saver()
sess.run(tf.global_variables_initializer())
# filewriter is how we write the summary protocol buffers to disk
# writer = tf.summary.FileWriter(<some-directory>, sess.graph)
writer = tf.summary.FileWriter(LOGDIR + hparam)
writer.add_graph(sess.graph)
## Format: tensorflow/contrib/tensorboard/plugins/projector/projector_config.proto
config = tf.contrib.tensorboard.plugins.projector.ProjectorConfig()
## You can add multiple embeddings. Here we add only one.
embedding_config = config.embeddings.add()
embedding_config.tensor_name = embedding.name
embedding_config.sprite.image_path = LOGDIR + 'sprite_1024.png'
embedding_config.metadata_path = LOGDIR + 'labels_1024.tsv'
# Specify the width and height of a single thumbnail.
embedding_config.sprite.single_image_dim.extend([28, 28])
tf.contrib.tensorboard.plugins.projector.visualize_embeddings(writer, config)
for i in range(2001):
batch = mnist.train.next_batch(100)
if i % 5 == 0:
[train_accuracy, s] = sess.run([accuracy, summ], feed_dict={x: batch[0], y: batch[1]})
# This method wraps the provided summary in an Event protocol buffer and adds it to the event file.
writer.add_summary(s, i)
if i % 500 == 0:
sess.run(assignment, feed_dict={x: mnist.test.images[:1024], y: mnist.test.labels[:1024]})
saver.save(sess, os.path.join(LOGDIR, "model.ckpt"), i)
sess.run(train_step, feed_dict={x: batch[0], y: batch[1]})
def make_hparam_string(learning_rate, use_two_fc, use_two_conv):
conv_param = "conv=2" if use_two_conv else "conv=1"
fc_param = "fc=2" if use_two_fc else "fc=1"
return "lr_{}_{}_{}".format(learning_rate, conv_param, fc_param)
def main():
# You can try adding some more learning rates
for learning_rate in [1E-3, 1E-4]:
# Include "False" as a value to try different model architectures
for use_two_fc in [True, False]:
for use_two_conv in [True, False]:
# Construct a hyperparameter string for each one (example: "lr_1E-3,fc=2,conv=2)
hparam = make_hparam_string(learning_rate, use_two_fc, use_two_conv)
print('Starting run for %s' % hparam)
# Actually run with the new settings
mnist_model(learning_rate, use_two_fc, use_two_conv, hparam)
if __name__ == '__main__':
main()