Using Keras repeat_elements in Lambda Layer giving incorrect output dimensions

Question

I have a very simple model which takes in a vector and should output a repeated element vector. Eg: I/P : [1,2,3] rep =3 | O/P = [1,1,1,2,2,2,3,3,3]. This is how my model looks like :

def simple_network(dim):

    x1 = Input(shape = (dim))
    repeated = Lambda(lambda x: K.repeat_elements(x, rep=12, axis=0))(x1)
    model = Model(inputs = x1, outputs = repeated)

    return model

However, this is what model_summary shows :

Model: "model_61"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_19 (InputLayer)        [(None, 7)]               0         
_________________________________________________________________
lambda_17 (Lambda)           (None, 7)                 0         
=================================================================
Total params: 0
Trainable params: 0
Non-trainable params: 0

Dim of my input vector is 7. Shouldn't I be getting an 84 dim vector as output? where am I going wrong?

Tensorflow can't know that - you're lambda isn't executed until the model is run. — c z
– c z, Commented Jun 1, 2021 at 12:06
you should use K.repeat_elements(x, rep=12, axis=1) to get an output dimensionality of (None, 84) — Marco Cerliani
– Marco Cerliani, Commented Jun 1, 2021 at 13:02
@MarcoCerliani yes! Just discovered the mistake! Thanks a lot — xerxes01
– xerxes01, Commented Jun 1, 2021 at 17:21

user11530462 · Accepted Answer · 2021-06-25 06:27:39Z

From comments

You should use K.repeat_elements(x, rep=12, axis=1) to get an output dimensionality of (None, 84) (paraphrased from Marco Cerliani)

from tensorflow.keras import Input, Model
from tensorflow.keras import backend as K
from tensorflow.keras.layers import Lambda

def simple_network(dim):

    x1 = Input(shape = (dim))
    repeated = Lambda(lambda x: K.repeat_elements(x, rep=12, axis=1))(x1)
    model = Model(inputs = x1, outputs = repeated)

    return model

simple_network(7).summary()

Output:

Model: "model"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_1 (InputLayer)         [(None, 7)]               0         
_________________________________________________________________
lambda (Lambda)              (None, 84)                0         
=================================================================
Total params: 0
Trainable params: 0
Non-trainable params: 0

Collectives™ on Stack Overflow

Using Keras repeat_elements in Lambda Layer giving incorrect output dimensions

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related