keras回调函数的使用

# 这里有两个callback函数：早停和模型检查点
callbacks_list=[
    keras.callbacks.EarlyStopping(
        monitor="val_accuracy",#监控指标
        patience=2 #两轮内不再改善中断训练
    ),
    keras.callbacks.ModelCheckpoint(
        filepath="checkpoint_path",
        monitor="val_loss",
        save_best_only=True
    )
]
#模型获取
model=get_minist_model()
model.compile(optimizer="rmsprop",
             loss="sparse_cateGorical_crossentropy",
             metrics=["accuracy"])

model.fit(train_images,train_labels,
         epochs=10,callbacks=callbacks_list, #该参数使用回调函数
         validation_data=(val_images,val_labels))

test_metrics=model.evaluate(test_images,test_labels)#计算模型在新数据上的损失和指标
predictions=model.predict(test_images)#计算模型在新数据上的分类概率

#也可以在训练完成后手动保存模型，只需调用model.save('my_checkpoint_path')。
#重新加载模型
model_new=keras.models.load_model("checkpoint_path.keras")

from matplotlib import pyplot as plt
# 实现记录每一轮中每个batch训练后的损失，并为每个epoch绘制一个图
class Losshistory(keras.callbacks.Callback):
    def on_train_begin(self, logs):
        self.per_batch_losses = []

    def on_batch_end(self, batch, logs):
        self.per_batch_losses.append(logs.get("loss"))

    def on_epoch_end(self, epoch, logs):
        plt.clf()
        plt.plot(range(len(self.per_batch_losses)), self.per_batch_losses,
                 label="Training loss for each batch")
        plt.xlabel(f"Batch (epoch {epoch})")
        plt.ylabel("Loss")
        plt.legend()
        plt.savefig(f"plot_at_epoch_{epoch}")
        self.per_batch_losses = [] #清空，方便下一轮的技术

model = get_mnist_model()
model.compile(optimizer="rmsprop",
              loss="sparse_categorical_crossentropy",
              metrics=["accuracy"])
model.fit(train_images, train_labels,
          epochs=10,
          callbacks=[LossHistory()],
          validation_data=(val_images, val_labels))

def get_minist_model():
    inputs=keras.Input(shape=(28*28,))
    features=layers.Dense(512,activation="relu")(inputs)
    features=layers.Dropout(0.5)(features)
    outputs=layers.Dense(10,activation="softmax")(features)
    model=keras.Model(inputs,outputs)
    return model
    
#datset
from Tensorflow.keras.datasets import mnist
(train_images,train_labels),(test_images,test_labels)=mnist.load_data()
train_images=train_images.reshape((60000,28*28)).astype("float32")/255
test_images=test_images.reshape((10000,28*28)).astype("float32")/255
train_images,val_images=train_images[10000:],train_images[:10000]
train_labels,val_labels=train_labels[10000:],train_labels[:10000]