I have been able to run code against the full dogs vs. cats dataset on 960M with Theano by limiting GPU memory used to 60% via the “cnmem” parameter in my theanorc file.
Is there a way to do this with TensorFlow?
I have been able to run code against the full dogs vs. cats dataset on 960M with Theano by limiting GPU memory used to 60% via the “cnmem” parameter in my theanorc file.
Is there a way to do this with TensorFlow?