"This is not supposed to happen!" Exception with VGG16 and LRP

I was trying to use innvestigate (1.0.2) in combination with a finetuned vgg16 model which I created in keras.
When trying to analyze a given image, I am running into an Exception that only states that it is not supposed to happen.
Maybe I am using the analyzer in a wrong way or it can not deal with this model (did not have time to dig deeply into the LRP theory, but only saw that sigmoid activations can not be used..but they aren't used in VGG16 afaik)

Here is my code:
```python
from keras.applications import VGG16
from keras import Model
from keras.layers import Input, Flatten, Dense
import innvestigate

vgg = VGG16(weights='imagenet', include_top=False, input_shape=(128, 128, 3))
lname = "block5_conv2"
last_vgg_layer = vgg.get_layer(lname)
vgg_partial = Model(inputs=vgg.input, outputs=last_vgg_layer.output, name="Partial_VGG16_" + lname)
vgg_partial.trainable = False

inputs = Input(shape=(128, 128, 3))
x = vgg_partial(inputs)
x = Flatten()(x)
x = Dense(16, activation='relu')(x)
x = Dense(1, activation=None)(x)

model = Model(inputs=inputs, outputs=x)
print(model.summary())

def analyze_img(img, model):
analyzer = innvestigate.create_analyzer("lrp.z", model)

a = analyzer.analyze(img)
# somehow this seems necesarry since importing once and calling the
# function in a loop crashes after the first iteration (v.1.0.0)
from innvestigate.utils.visualizations import heatmap
return heatmap(a[0, :, :, :])

# Analyze.
img = np.ones((128, 128, 3))
print("img.shape", img.shape)

heat_train = analyze_img(img, model)
print(heat_train.shape)
```

which gives the following stacktrace:

I have installed the following packages:
```bash
Keras 2.2.0
Keras-Applications 1.0.2
tensorflow 1.10.0
innvestigate 1.0.2
````