Tensorflow MeanSquaredError doens't work on one number

Asked Jan 02 '20 at 13:36

Active Jan 02 '20 at 13:36

Viewed 173 times

I am trying calculate the loss of a network with tensorflow mean squared error, but for some reason it doesn't work if the input tensors only have one number. How should do this instead.

Here is some code:

import tensorflow as tf

loss = tf.keras.losses.MeanSquaredError()

a = loss(y_true=tf.constant([1.0, 2.0, 3.0]), y_pred=tf.constant([2.0, 2.0, 4.0]))
print(a)

a = loss(y_true=tf.constant(1.0, dtype=tf.float32), y_pred=tf.constant(2.0, dtype=tf.float32)) #this is where the error occurs.
print(a)

error

tensorflow.python.framework.errors_impl.InvalidArgumentError: Invalid reduction dimension (-1 for input with 0 dimension(s) [Op:Mean]

asked Jan 02 '20 at 13:36

2

you need to pass an array: `a = loss(y_true=tf.constant([1.0], dtype=tf.float32), y_pred=tf.constant([2.0], dtype=tf.float32))` – vital_dml Jan 02 '20 at 13:48
You're right that worked. – Jan 02 '20 at 14:13

Tensorflow MeanSquaredError doens't work on one number

0 Answers0