提前停止不会在 Keras 停止

如何解决提前停止不会在 Keras 停止

我正在尝试使用 Keras。

我构建了一个包含 3 个 Dense Layers 和 1 个 Masking 的模型。所以我确实放置了 EarlyStopping 来监控 'val-loss' 但是正如您在下面看到的那样，经过 val_loss 的这么多时代是一样的，模型不断训练。我不知道我是否宣布 EarlyStopping 是错误的。

3/3 - 0s - loss: 0.3445 - accuracy: 0.3743 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 460/500
3/3 - 0s - loss: 0.3448 - accuracy: 0.3630 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 461/500
3/3 - 0s - loss: 0.3445 - accuracy: 0.3681 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 462/500
3/3 - 0s - loss: 0.3447 - accuracy: 0.3592 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 463/500
3/3 - 0s - loss: 0.3446 - accuracy: 0.3660 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 464/500
3/3 - 0s - loss: 0.3445 - accuracy: 0.3653 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 465/500
3/3 - 0s - loss: 0.3446 - accuracy: 0.3678 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 466/500
3/3 - 0s - loss: 0.3445 - accuracy: 0.3633 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 467/500
3/3 - 0s - loss: 0.3446 - accuracy: 0.3614 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 468/500
3/3 - 0s - loss: 0.3445 - accuracy: 0.3669 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 469/500
3/3 - 0s - loss: 0.3446 - accuracy: 0.3677 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 470/500
3/3 - 0s - loss: 0.3446 - accuracy: 0.3609 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 471/500
3/3 - 0s - loss: 0.3444 - accuracy: 0.3743 - val_loss: 0.4220 - val_accuracy: 0.3510
Epoch 472/500
3/3 - 0s - loss: 0.3446 - accuracy: 0.3641 - val_loss: 0.4220 - val_accuracy: 0.3510

这是我的模型代码：

model = keras.Sequential()
model.add(layers.Masking(mask_value=0.,input_shape=(498,20)))
model.add(layers.Dense(16,activation ='relu')) 
model.add(layers.Dropout(0.5))
model.add(layers.Dense(8,activation ='relu')) 
model.add(layers.Dropout(0.5))
model.add(layers.Dense(3,activation='softmax'))
model.add(layers.Activation('softmax'))
model.summary()

opt = keras.optimizers.SGD(learning_rate=0.001)
model.compile(optimizer=opt,loss='categorical_crossentropy',metrics=['accuracy'])

model.summary()

es = EarlyStopping(monitor='val_loss',patience=2,verbose=2)
history = model.fit(
    x_train,y_train,epochs=500,batch_size=32,validation_data=(x_test,y_test),verbose = 2,callbacks=[es]
    )
scores = model.evaluate(x_test,y_test,verbose=2)