SampleRNN
SampleRNN copied to clipboard
gpu is not really used
hi,
I am trying to use SampleRNN on a g3 aws instance, with the tensorflow_p27 env. When I launch run.sh, the gpu memory is used only partially ( 670MiB out of 7618MiB, not even 10%) and the GPU % is at 0%.
How can I improve this situation?