4 ms·
When it comes to scalability, Apache MXNet (http://mxnet.io/ http://mxnet.io/) is actually the best choice. Multi-GPU support and distributed training on multip
by julsimon 9y ago
When it comes to scalability, Apache MXNet (http://mxnet.io/ http://mxnet.io/) is actually the best choice. Multi-GPU support and distributed training on multiple hosts are extremely easy to set up. It's also supported by Keras (still in beta, though).