keras回调函数callback

• earlystopping，保存最佳模型，加载最佳模型。

import tensorflow as tf
print(tf.__version__)
from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint
from tensorflow.keras.models import load_model

# 定义callback
filepath = 'my_model.h5'
callbacks_list = [
  EarlyStopping(
  monitor='val_acc',
  patience=3,
  ),
  ModelCheckpoint(
  filepath=filepath,
  monitor='val_acc',
  save_best_only=True,
  save_weights_only=False
  )
]

# 加载数据及预处理
mnist = tf.keras.datasets.fashion_mnist
(training_images, training_labels), (test_images, test_labels) = mnist.load_data()
training_images=training_images/255.0
test_images=test_images/255.0

# 定义模型
model = tf.keras.models.Sequential([
  tf.keras.layers.Flatten(),
  tf.keras.layers.Dense(512, activation=tf.nn.relu),
  tf.keras.layers.Dense(10, activation=tf.nn.softmax)
])
model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])
model.fit(training_images, training_labels, epochs=30, callbacks=callbacks_list, validation_data=(test_images, test_labels))

# 加载最优模型
model = load_model(filepath)
model.evaluate(test_images, test_labels)

# 使用模型预测
classifications = model.predict(test_images)
print(classifications[0])
print(test_labels[0])

• callbacks

import tensorflow as tf
from tensorflow.keras.layers import Dense, Flatten
mnist = tf.keras.datasets.mnist

(x_train, y_train),(x_test, y_test) = mnist.load_data()

model = tf.keras.models.Sequential([
    Flatten(input_shape = (28, 28)),
    Dense(512, activation='relu'),
    Dense(10, activation='softmax'),
])

class myCallback(tf.keras.callbacks.Callback):
  def on_epoch_end(self, epoch, logs={}):
    if(logs.get('acc')>0.99):
      print("\nReached 99% accuracy so cancelling training!")
      self.model.stop_training = True
callback = myCallback()
model.compile(optimizer='adam',
              loss='sparse_categorical_crossentropy',
              metrics=['accuracy'])
model.fit(x_train, y_train, epochs=10, callbacks=[callback])

model.evaluate(x_test, y_test)
pred = model.predict(x_test)