博客

参考

https://machinelearningmastery.com/check-point-deep-learning-models-keras/

https://keras.io/callbacks/

http://keras-cn.readthedocs.io/en/latest/other/callbacks/

语法

keras.callbacks.ModelCheckpoint(filepath,monitor=’val_loss’,verbose=0,save_best_only=False, save_weights_only=False, mode=’auto’, period=1)

参数说明：

• filename：字符串，保存模型的路径
• monitor：需要监视的值
• verbose：信息展示模式，0或1(checkpoint的保存信息，类似Epoch 00001: saving model to …)
• save_best_only：当设置为True时，监测值有改进时才会保存当前的模型（
  
  the latest best model according to the quantity monitored will not be overwritten
  
  ）
• mode：‘auto’，‘min’，‘max’之一，在save_best_only=True时决定性能最佳模型的评判准则，例如，当监测值为val_acc时，模式应为max，当监测值为val_loss时，模式应为min。在auto模式下，评价准则由被监测值的名字自动推断。
• save_weights_only：若设置为True，则只保存模型权重，否则将保存整个模型（包括模型结构，配置信息等）
• period：CheckPoint之间的间隔的epoch数

例子

代码

from __future__ import print_function
import keras
from keras.datasets import cifar10
from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation, Flatten
from keras.layers import Conv2D, MaxPooling2D
from keras.callbacks import ModelCheckpoint
import os

batch_size = 128
num_classes = 10
epochs = 30
num_predictions = 20
save_dir = os.path.join(os.getcwd(), 'saved_models')
model_name = 'keras_cifar10_trained_model.h5'

# The data, split between train and test sets:
(x_train, y_train), (x_test, y_test) = cifar10.load_data()
print('x_train shape:', x_train.shape)
print(x_train.shape[0], 'train samples')
print(x_test.shape[0], 'test samples')

# Convert class vectors to binary class matrices.
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)

model = Sequential()
model.add(Conv2D(32, (3, 3), padding='same',
                 input_shape=x_train.shape[1:]))
model.add(Activation('relu'))
model.add(Conv2D(32, (3, 3)))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))

model.add(Conv2D(64, (3, 3), padding='same'))
model.add(Activation('relu'))
model.add(Conv2D(64, (3, 3)))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))

model.add(Flatten())
model.add(Dense(512))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(num_classes))
model.add(Activation('softmax'))

# initiate RMSprop optimizer
opt = keras.optimizers.rmsprop(lr=0.0001, decay=1e-6)

# Let's train the model using RMSprop
model.compile(loss='categorical_crossentropy',
              optimizer=opt,
              metrics=['accuracy'])

x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255

filepath="model_{epoch:02d}-{val_acc:.2f}.hdf5"
checkpoint = ModelCheckpoint(os.path.join(save_dir, filepath), monitor='val_acc',verbose=1, 
                            save_best_only=True)

history = model.fit(x_train, y_train,
          batch_size=batch_size,
          epochs=epochs,
          validation_data=(x_test, y_test),
          shuffle=True,
          callbacks=[checkpoint])
print(history.history.keys())
# Save model and weights
if not os.path.isdir(save_dir):
    os.makedirs(save_dir)
model_path = os.path.join(save_dir, model_name)
model.save(model_path)
print('Saved trained model at %s ' % model_path)

# Score trained model.
scores = model.evaluate(x_test, y_test, verbose=1)
print('Test loss:', scores[0])
print('Test accuracy:', scores[1])

输出

x_train shape: (50000, 32, 32, 3)
50000 train samples
10000 test samples
Train on 50000 samples, validate on 10000 samples
Epoch 1/30
50000/50000 [==============================] - 5s 106us/step - loss: 1.9831 - acc: 0.2734 - val_loss: 1.7195 - val_acc: 0.3934


Epoch 00001: val_acc improved from -inf to 0.39340, saving model to /home/weiliu/PG/keras/saved_models/model_01-0.39.hdf5
Epoch 2/30
50000/50000 [==============================] - 4s 84us/step - loss: 1.6651 - acc: 0.3963 - val_loss: 1.5001 - val_acc: 0.4617


Epoch 00002: val_acc improved from 0.39340 to 0.46170, saving model to /home/weiliu/PG/keras/saved_models/model_02-0.46.hdf5
Epoch 3/30
50000/50000 [==============================] - 4s 88us/step - loss: 1.5249 - acc: 0.4482 - val_loss: 1.4158 - val_acc: 0.4923


Epoch 00003: val_acc improved from 0.46170 to 0.49230, saving model to /home/weiliu/PG/keras/saved_models/model_03-0.49.hdf5
Epoch 4/30
50000/50000 [==============================] - 4s 87us/step - loss: 1.4479 - acc: 0.4775 - val_loss: 1.3469 - val_acc: 0.5199


Epoch 00004: val_acc improved from 0.49230 to 0.51990, saving model to /home/weiliu/PG/keras/saved_models/model_04-0.52.hdf5
Epoch 5/30
50000/50000 [==============================] - 4s 89us/step - loss: 1.3838 - acc: 0.5048 - val_loss: 1.3058 - val_acc: 0.5332


Epoch 00005: val_acc improved from 0.51990 to 0.53320, saving model to /home/weiliu/PG/keras/saved_models/model_05-0.53.hdf5
Epoch 6/30
50000/50000 [==============================] - 4s 87us/step - loss: 1.3359 - acc: 0.5226 - val_loss: 1.3543 - val_acc: 0.5226


Epoch 00006: val_acc did not improve from 0.53320
Epoch 7/30
50000/50000 [==============================] - 4s 88us/step - loss: 1.2934 - acc: 0.5399 - val_loss: 1.1970 - val_acc: 0.5769


Epoch 00007: val_acc improved from 0.53320 to 0.57690, saving model to /home/weiliu/PG/keras/saved_models/model_07-0.58.hdf5
Epoch 8/30
50000/50000 [==============================] - 4s 87us/step - loss: 1.2522 - acc: 0.5536 - val_loss: 1.1682 - val_acc: 0.5865

由以上输出不难发现，有部分epoch未保存checkpoint，如第6个epoch。