I was able to launch the GPU instance and install everything I need. I tested it with Python codes given by MXNet install instructions on AWS DL AMI and it printed out the expected result. However, after I opened the Jupyter notebook and ran the same test. It gave me the error “Operator _ones is not implemented for GPU.” Any thoughts or ideas? Thank you very much for your help!
I was wondering how you resolve this? I have the same issue.
Are you running DLAMI or on a local installation? If it’s local, then your Cuda versions may be mismatched with the MXNet you’ve installed. What version of Cuda/cuDNN are you running? What version of MXNet?
Take a look at this issue that’s related: https://github.com/apache/incubator-mxnet/issues/7740