Keras array shape error

Question

How come I am getting a shape ValueError for my neural network when what I am passing in is an array of shape (8,1) but the error I am getting is that the neural network is complaining about getting a (1,)?

Neural Network:

>>> observation_dimension
(8,)
>>> q_network = Sequential([
    Dense(40, input_dim=observation_dimension, activation='relu'),
    Dense(40, activation='relu'),
    Dense(number_of_actions, activation='linear')
])
>>> obs
array([-0.00371828,  0.93953934, -0.37663383, -0.07161933,  0.00431531,
        0.08531308,  0.        ,  0.        ])
>>> obs.shape
(8,)

Error:

>>> q_network.predict(obs)
Traceback (most recent call last):
...
...
ValueError: Error when checking input: expected dense_27_input to have shape (8,) but got array with shape (1,)

Dr. Snoopy · Accepted Answer · 2018-06-28 09:18:28Z

2

model.predict takes a batch of samples, if you give it a single sample with the wrong shape, it will interpret the first dimension as the batch one.

A simple solution is to add a dimension with a value of one:

q_network.predict(obs.reshape(1, 8))

answered Jun 28, 2018 at 9:18

Dr. Snoopy

56.5k7 gold badges148 silver badges161 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

mmong · Accepted Answer · 2018-06-28 09:00:00Z

1

The predict method expects a 2d array so just reshape your obs:

obs = np.reshape(obs,(-1,len(obs)))

answered Jun 28, 2018 at 9:00

mmong

642 bronze badges

Collectives™ on Stack Overflow

Keras array shape error

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related