Fetching the paper…

Efficient and Robust Parallel DNN Training through Model Parallelism on Multi-GPU Platform · Around