User 1726 | 4/7/2015, 9:23:35 AM
Right now i am using the deeplearning toolkit to train a neural network. The machine i use has two separate gpu,i.e. gpu 0, gpu 1. I found that the graphlab.neuralnet_classifier.create() only used the gpu 0. My question is can i assign the gpu id so that i can train two different models on two different gpu synchronously?