I have Keras installed with the Tensorflow backend and CUDA. I’d like to sometimes on demand force Keras to use CPU. Can this be done without say installing a separate CPU-only Tensorflow in a virtual environment? If so how? If the backend were Theano, the flags could be set, but I have not heard of Tensorflow flags accessible via Keras.
If you want to force Keras to use CPU
import os os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID" # see issue #152 os.environ["CUDA_VISIBLE_DEVICES"] = ""
before Keras / Tensorflow is imported.
Run your script as
$ CUDA_VISIBLE_DEVICES="" ./your_keras_code.py
Answered By – Martin Thoma