While the name could be better, I would instead focus on (1) making mxnet much more extensible (e.g. support ability to dynamically load operators from external shared libraries), (2) feature parity with tensorflow, (3) support for non-NVIDIA GPUs, (4) clearly demonstrating and publicizing better performance for large models.
On 4/11/18, 1:18 PM, "Chris Olivier" <[email protected]> wrote: Should we consider renaming MXNet to something more "friendly"? IMHO, I think this may be related to adoption problems. MXNet, CMTK -- both seem sort of sterile and hard to use, don't they? Tensorflow, PyTorch, Caffe -- sound cool.
