微调模型（mobilenetv2）后，精度降低，张量流

Question

def create_model(n_classes,train_data,val_data,target_dims,batch_size,epochs,optimizer="adam"):
    conv_base=MobileNetV2(weights="imagenet",include_top=False,input_shape=target_dims)
    model=Sequential()
    model.add(conv_base)
    model.add(Flatten())
    model.add(Dense(256,activation="relu"))
    
    model.add(Dense(64,activation="relu"))
 
    model.add(Dense(n_classes,activation="softmax"))
    for layer in conv_base.layers:
        layer.trainable=False
    
    model.compile(loss="categorical_crossentropy",optimizer=optimizer,metrics=['acc'])
    print("training the model without fine tuning.... \n ")
    es=EarlyStopping(monitor='val_loss')
    model.fit_generator(train_data,epochs=epochs,
                        validation_data=val_data,
                        callbacks=[es])
    for layers in conv_base.layers[:150]:
        layers.trainable=True
    model.compile(loss="categorical_crossentropy",optimizer=optimizer,metrics=['acc'])
    print("training the model after fine tuning.... \n ")
    model.fit_generator(train_data,epochs=epochs,
                        validation_data=val_data,
                        callbacks=[es])
    

**我的精度在微调后正在降低** enter image description here