The TensorFlow* image classification sample codes below describe a step-by-step approach to modify the code in order to scale the deep learning training across multiple nodes of HPC data centers. The detailed descriptions of each step are part of a separate article that is going to be published.
All samples are based on original samples from the TensorFlow* repository.
The samples were tested on a CentOS* 7 installation with an Intel optimized TensorFlow* version 1.10 that makes use of the Intel® Math Kernel Library for Deep Neural Networks (Intel® MKL-DNN) library along with the Horovod* framework.
Product and Performance Information
Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.