Distributed Tensorflow on a multi-node cluster of CPUs

This is a proposal for tutorial on O'reilly conference.

To train large Deep Learning models efficiently across multiple CPUs we would need a highly scalable framework. Tensorflow provides several strategies to train on a multi node CPU cluster. In this video we will present an evaluation of Tensorflow’s different Distribution Strategies and provide the best practices to get over 90% scaling across a cluster of Intel(R) Xeon((R) Scalable Processors.