How do you get this to work with more than 5 fonts?

[Burkino]

I am trying to train a model that recognizes many fonts, however it appears to not work with more than 5 fonts.
I have replaced the conv_label function with

font_list = {}
def conv_label(label):
    if label in font_list:
        return font_list[label]
    else:
        font_id = len(font_list)
        font_list[label] = font_id
        return font_id

and this allows me to automatically give each font a unique id. However upon generating training data and then trying to train the model I get the error IndexError: index 6 is out of bounds for axis 1 with size 5.
I then change

trainY = to_categorical(trainY, num_classes=5)
testY = to_categorical(testY, num_classes=5)

to

trainY = to_categorical(trainY, num_classes=len(font_list))
testY = to_categorical(testY, num_classes=len(font_list))

and run it again. However, this time I get the error ValueError: Dimensions must be equal, but are 5 and 6 for '{{node mean_squared_error/SquaredDifference}} = SquaredDifference[T=DT_FLOAT](sequential_2/dense_11/Softmax, IteratorGetNext:1)' with input shapes: [?,5], [?,6].

[kelogs]

you have not happen to solve this problem, have you ?

[jerome-labonte-udem]

Looks like you need to change the model creation method (line 103 of train.py). The last layer is hard-coded to 5, so you need to replace that with your desired number of classes. Simplest way is probably to pass num_classes to the model_create method and use that number to determine the size of the last layer output (the one using softmax)