Understanding the RepeatVector layer
You will now explore how the RepeatVector layer works. The RepeatVector layer adds an extra dimension to your dataset. For example if you have an input of shape (batch size, input size) and you want to feed that to a GRU layer, you can use a RepeatVector layer to convert the input to a tensor with shape (batch size, sequence length, input size).
In this exercise, you will define a model that repeats a given input a fixed number of times. You will then feed a numpy array to the model and investigate how the model changes the output.
This exercise is part of the course
Machine Translation with Keras
Exercise instructions
- Define a
RepeatVectorlayer that repeats the input6times. - Define a
Modelthat takes the input layer in and produces the repeat vector output. - Define a
numpyarray object that has data[[0,1], [2,3]]. - Predict the output of the model by feeding
xas an input.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
from tensorflow.keras.layers import Input, RepeatVector
from tensorflow.keras.models import Model
import numpy as np
inp = Input(shape=(2,))
# Define a RepeatVector that repeats the input 6 times
rep = ____(____)(inp)
# Define a model
model = ____(____=____, ____=____)
# Define input x
x = ____.____([____,____])
# Get model prediction y
y = ____.____(____)
print('x.shape = ',x.shape,'\ny.shape = ',y.shape)