原文链接:https://blog.csdn.net/jasminexjf/article/details/82817912
tensorboard 作为一款可视化神器,可以说是学习tensorflow时模型训练以及参数可视化的法宝。
而在训练过程中,主要用到了tf.summary()的各类方法,能够保存训练过程以及参数分布图并在tensorboard显示。
tf.summary有诸多函数:
1、tf.summary.scalar
用来显示标量信息,其格式为:
tf.summary.scalar(tags, values, collections=None, name=None)
例如:tf.summary.scalar('mean', mean)
一般在画loss,accuary时会用到这个函数。
2、tf.summary.histogram
用来显示直方图信息,其格式为:
tf.summary.histogram(tags, values, collections=None, name=None)
例如: tf.summary.histogram('histogram', var)
一般用来显示训练过程中变量的分布情况
3、tf.summary.distribution
分布图,一般用于显示weights分布
4、tf.summary.text
可以将文本类型的数据转换为tensor写入summary中:
例如:
text = """/a/b/c\\_d/f\\_g\\_h\\_2017""" summary_op0 = tf.summary.text('text', tf.convert_to_tensor(text))
5、tf.summary.image
输出带图像的probuf,汇总数据的图像的的形式如下: ' tag /image/0', ' tag /image/1'...,如:input/image/0等。
格式:tf.summary.image(tag, tensor, max_images=3, collections=None, name=Non
6、tf.summary.audio
展示训练过程中记录的音频
7、tf.summary.merge_all
merge_all 可以将所有summary全部保存到磁盘,以便tensorboard显示。如果没有特殊要求,一般用这一句就可一显示训练时的各种信息了。
格式:tf.summaries.merge_all(key='summaries')
8、tf.summary.FileWriter
指定一个文件用来保存图。
格式:tf.summary.FileWritter(path,sess.graph)
可以调用其add_summary()方法将训练过程数据保存在filewriter指定的文件中
Tensorflow Summary 用法示例:
-
tf.summary.scalar(
'accuracy',acc)
#生成准确率标量图
-
merge_summary = tf.summary.merge_all()
-
train_writer = tf.summary.FileWriter(dir,sess.graph)
#定义一个写入summary的目标文件,dir为写入文件地址
-
......(交叉熵、优化器等定义)
-
for step
in xrange(training_step):
#训练循环
-
train_summary = sess.run(merge_summary,feed_dict = {...})
#调用sess.run运行图,生成一步的训练过程数据
-
train_writer.add_summary(train_summary,step)
#调用train_writer的add_summary方法将训练过程以及训练步数保存
此时开启tensorborad:
- tensorboard --logdir=/summary_dir
便能看见accuracy曲线了。
另外,如果我不想保存所有定义的summary信息,也可以用tf.summary.merge方法有选择性地保存信息:
9、tf.summary.merge
格式:tf.summary.merge(inputs, collections=None, name=None)
一般选择要保存的信息还需要用到tf.get_collection()函数
示例:
-
tf.summary.scalar(
'accuracy',acc)
#生成准确率标量图
-
merge_summary = tf.summary.merge([tf.get_collection(tf.GraphKeys.SUMMARIES,
'accuracy'),...(其他要显示的信息)])
-
train_writer = tf.summary.FileWriter(dir,sess.graph)
#定义一个写入summary的目标文件,dir为写入文件地址
-
......(交叉熵、优化器等定义)
-
for step
in xrange(training_step):
#训练循环
-
train_summary = sess.run(merge_summary,feed_dict = {...})
#调用sess.run运行图,生成一步的训练过程数据
-
train_writer.add_summary(train_summary,step)
#调用train_writer的add_summary方法将训练过程以及训练步数保存
使用tf.get_collection函数筛选图中summary信息中的accuracy信息,这里的
tf.GraphKeys.SUMMARIES 是summary在collection中的标志。
当然,也可以直接:
-
acc_summary = tf.summary.scalar(
'accuracy',acc)
#生成准确率标量图
-
merge_summary = tf.summary.merge([acc_summary ,...(其他要显示的信息)])
#这里的[]不可省
如果要在tensorboard中画多个数据图,需定义多个tf.summary.FileWriter并重复上述过程。
这里有一段代码供阅读参考用法:with tf.Graph().as_default():
session_conf = tf.ConfigProto(
allow_soft_placement=FLAGS.allow_soft_placement, #是否允许tensorflow进行运算的软分配
log_device_placement=FLAGS.log_device_placement) ##是否允许tensorflow打印操作步骤
sess = tf.Session(config=session_conf)
with sess.as_default():
cnn = TextCNN(
sequence_length=x_train.shape[1],
num_classes=y_train.shape[1],
vocab_size=len(vocab_processor.vocabulary_),
embedding_size=FLAGS.embedding_dim,
filter_sizes=list(map(int, FLAGS.filter_sizes.split(","))),
num_filters=FLAGS.num_filters,
l2_reg_lambda=FLAGS.l2_reg_lambda) #L2正则化的系数lambda
#另外一种计算梯度下降的方式,之前的都是tf.compat.v1.train.GradientDescentOptimizer(0.2).minimize(loss)
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
optimizer = tf.train.AdamOptimizer(1e-3)
grads_and_vars = optimizer.compute_gradients(cnn.loss) #计算梯度
train_op = optimizer.apply_gradients(grads_and_vars, global_step=global_step) #将compute_gradients()返回的值作为输入参数对variable进行更新
# Keep track of gradient values and sparsity (optional)
grad_summaries = []
for g, v in grads_and_vars:
if g is not None:
grad_hist_summary = tf.summary.histogram("{}/grad/hist".format(v.name), g) #tf.summary显示梯度的直方图
sparsity_summary = tf.summary.scalar("{}/grad/sparsity".format(v.name), tf.nn.zero_fraction(g)) #tf.nn.zero_fraction计算所0在tensor类型的占比
grad_summaries.append(grad_hist_summary)
grad_summaries.append(sparsity_summary)
grad_summaries_merged = tf.summary.merge(grad_summaries) #相当于把信息合并
# Output directory for models and summaries
timestamp = str(int(time.time())) #得到当前的时间
out_dir = os.path.abspath(os.path.join(os.path.curdir, "runs", timestamp)) #根据时间定义了一个路径
print("Writing to {}\n".format(out_dir))
# Summaries for loss and accuracy
loss_summary = tf.summary.scalar("loss", cnn.loss)
acc_summary = tf.summary.scalar("accuracy", cnn.accuracy)
# Train Summaries
train_summary_op = tf.summary.merge([loss_summary, acc_summary, grad_summaries_merged]) #合并信息
train_summary_dir = os.path.join(out_dir, "summaries", "train") #定义训练可视化信息保存的路径
train_summary_writer = tf.summary.FileWriter(train_summary_dir, sess.graph)
# Dev summaries
dev_summary_op = tf.summary.merge([loss_summary, acc_summary])
dev_summary_dir = os.path.join(out_dir, "summaries", "dev")
dev_summary_writer = tf.summary.FileWriter(dev_summary_dir, sess.graph)
# Checkpoint directory. Tensorflow assumes this directory already exists so we need to create it
checkpoint_dir = os.path.abspath(os.path.join(out_dir, "checkpoints")) #返回绝对路径
checkpoint_prefix = os.path.join(checkpoint_dir, "model")
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
saver = tf.train.Saver(tf.global_variables(), max_to_keep=FLAGS.num_checkpoints) #保存模型
# Write vocabulary
vocab_processor.save(os.path.join(out_dir, "vocab")) #保存生成的词库
# Initialize all variables
sess.run(tf.global_variables_initializer())
def train_step(x_batch, y_batch):
"""
A single training step
"""
feed_dict = {
cnn.input_x: x_batch,
cnn.input_y: y_batch,
cnn.dropout_keep_prob: FLAGS.dropout_keep_prob
}
_, step, summaries, loss, accuracy = sess.run(
[train_op, global_step, train_summary_op, cnn.loss, cnn.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat() #获取当前的时间
print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
train_summary_writer.add_summary(summaries, step)
def dev_step(x_batch, y_batch, writer=None):
"""
Evaluates model on a dev set
"""
feed_dict = {
cnn.input_x: x_batch,
cnn.input_y: y_batch,
cnn.dropout_keep_prob: 1.0
}
step, summaries, loss, accuracy = sess.run(
[global_step, dev_summary_op, cnn.loss, cnn.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat()
print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
if writer:
writer.add_summary(summaries, step) #这里Writer还没有给他赋值,后面后面调用函数的时候有赋值tf.summary.FileWrite
# Generate batches
batches = data_helpers.batch_iter(
list(zip(x_train, y_train)), FLAGS.batch_size, FLAGS.num_epochs)
# Training loop. For each batch...
for batch in batches:
x_batch, y_batch = zip(*batch)
train_step(x_batch, y_batch)
current_step = tf.train.global_step(sess, global_step)
if current_step % FLAGS.evaluate_every == 0: #每隔多少步的训练后,评估一次,并保存模型
print("\nEvaluation:")
dev_step(x_dev, y_dev, writer=dev_summary_writer)
print("")
if current_step % FLAGS.checkpoint_every == 0:
path = saver.save(sess, checkpoint_prefix, global_step=current_step)
print("Saved model checkpoint to {}\n".format(path))