Keras Dense input_shape as a list or a tuple

user_843150 0
Viewed 526

I've joined the ranks of software developers learning ML through TensorFlow and Keras, and one detail that keeps bothering me as I work through tutorials is the two flavors of specifying the input_shape to the first layer of a Sequential model:

# input_shape as a list
layer0 = tensorflow.keras.layers.Dense(units=1, input_shape=[1])

vs.

# input_shape as a tuple
layer0 = tensorflow.keras.layers.Dense(units=1, input_shape=(1,))

This appears to be functionally equivalent, but I'm not sure why, nor do I understand why both forms are used but the latter form (tuple) seems to be preferred.

keras tensorflow neural-network python
1 Answers

It actually does not matter to use which one of them. Both are accepted and the result is the same.

As you said, tuple is preferred due to immutability. The major difference between tuples and lists is that a list is mutable, whereas a tuple is immutable. This means that a list can be changed, but a tuple cannot.

user_24232780
Related