ValueError «You are trying to use the old GPU back-end» when importing keras

python machine-learning neural-network keras theano
Tshilidzi Mudau picture Tshilidzi Mudau · Jun 26, 2017 · Viewed 10k times · Source

I'm using Keras with the Theano backend on Ubuntu 16.04. My setup has been working without issues, however, all of a sudden I get the following error when I import Keras (import keras):

ValueError: You are trying to use the old GPU back-end. It was removed from Theano. Use device=cuda* now. See https://github.com/Theano/Theano/wiki/Converting-to-the-new-gpu-back-end%28gpuarray%29 for more information.

How do I resolve this?

Answer

Wilmar van Ommeren picture Wilmar van Ommeren · Jun 27, 2017

You should change (or add) your environmental variable called THEANO_FLAGS. If you set the variable so that it contains device=cuda instead of device=gpu the error will be gone.

Also set the floating point precision to float32 when working on the GPU as that is usually much faster (THEANO_FLAGS='device=cuda,floatX=float32').

More info on this variable can be found here and here.

Related questions

Keras model.summary() result - Understanding the # of Parameters

I have a simple NN model for detecting hand-written digits from a 28x28px image written in python using Keras (Theano backend): model0 = Sequential() #number of epochs to train for nb_epoch = 12 #amount of data each iteration in an epoch …

Read More
How to concatenate two layers in keras?

I have an example of a neural network with two layers. The first layer takes two arguments and has one output. The second should take one argument as result of the first layer and one additional argument. It should looks …

Read More
keras: how to save the training history attribute of the history object

In Keras, we can return the output of model.fit to a history as follows: history = model.fit(X_train, y_train, batch_size=batch_size, nb_epoch=nb_epoch, validation_data=(X_test, y_test)) Now, how to save …

Read More