Keras: Lambda output layer

[heshameraqi]

I've a sequential model as follows with a linear activation function (Keras default) for the single output neuron:

model = Sequential()
model.add( ...
...
model.add(Dense(100, activation='relu'))
model.add(Dense(1))

I need the final number to be bounded by 100, so I modified the last line of code above to be:
model.add(Lambda(lambda x: x%100, output_shape=(1)))
1- Is it correct? Does x here means the net as I expect ?
2- I get an error: "In Lambda, output_shape must be a list, a tuple, or a function".

[farizrahman4u]

(Notice the comma after 1)

model.add(Lambda(lambda x: x%100, output_shape=(1,))

You can also igonre the output_shape argument in this case.

model.add(Lambda(lambda x: x%100))

For bounding by 100, I think you should use K.clip.

[heshameraqi]

@farizrahman4u Thank you so much. But doesn't this mean that I will have 100 output neurons not a single one ?

[kgrm]

No. Since your last layer, Dense(1) already reduces the dimensionality to 1, and it is conserved through the Lambda layer unless specified otherwise.

[heshameraqi]

@kgrm I say "I modified the last line ...", so it's not there any more. So, I think it will produce 100 output neurons, and the workaround is to add a dense(1) layer before.
Can I do it with single layer directly ?

[kgrm]

You can, if you use a function that takes 100 inputs and produces a single output.

[heshameraqi]

@kgrm can you please refer me to an article or something that tells how to code this please? Thank you.

[kgrm]

For an example of a function that takes an arbitrary-shape tensor and outputs a scalar, see K.sum().

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs, but feel free to re-open it if needed.