1
$\begingroup$

Does using GPU instead of CPU with tensorflow and keras model reduces the time of .predict() method on a trained network or maybe it only reduces time of training of the network?

$\endgroup$
1
  • $\begingroup$ Please, ask programming questions on Stack Overflow. $\endgroup$ – nbro Mar 12 at 17:44
0
$\begingroup$

Yes, Keras will use the GPU for inference (predictions) as well as training.

TensorFlow code, and tf.keras models will transparently run on a single GPU with no code changes required.

For more information, you can read the TensorFlow GPU guide, but in effect you shouldn't need to do anything to get inference to run on the GPU.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.