无法更改现有Keras模型中的激活

Question

我有一个正常的VGG16模型,有relu激活,即

def VGG_16(weights_path=None):
    model = Sequential()
    model.add(ZeroPadding2D((1, 1),input_shape=(3, 224, 224)))
    model.add(Convolution2D(64, 3, 3, activation='relu'))
    model.add(ZeroPadding2D((1, 1)))
    model.add(Convolution2D(64, 3, 3, activation='relu'))
    model.add(MaxPooling2D((2, 2), strides=(2, 2)))
[...]
    model.add(Flatten())
    model.add(Dense(4096, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(4096, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(1000, activation='softmax'))

    if weights_path:
        model.load_weights(weights_path)

    return model

我用现有的权重实例化它,现在想要将所有relu激活更改为softmax(没有用,我知道)

model = VGG_16('vgg16_weights.h5')
sgd = SGD(lr=0.1, decay=1e-6, momentum=0.9, nesterov=True)

softmax_act = keras.activations.softmax
for (n, layer) in enumerate(model.layers):
    if 'activation' in layer.get_config() and layer.get_config()['activation'] == 'relu':
        print('replacing #{}: {}, {}'.format(n, layer, layer.activation))
        layer.activation = softmax_act
        print('-> {}'.format(layer.activation))

model.compile(optimizer=sgd, loss='categorical_crossentropy')

注意:在更改后model.compile调用,因此模型应该仍然可以修改.

但是,即使调试打印正确说

replacing #1: <keras.layers.convolutional.Convolution2D object at 0x7f7d7c497f50>, <function relu at 0x7f7dbe699a28>
-> <function softmax at 0x7f7d7c4972d0>
[...]

实际结果与relu激活模型相同.
为什么Keras不使用改变的激活功能？

Answer 1

您可能要使用apply_modifications

idx_of_layer_to_change = -1
model.layers[idx_of_layer_to_change].activation = activations.softmax
model = utils.apply_modifications(model)