Ton is a senior software developer in the IBM Cognitive OpenTech Group at the IBM Silicon Valley Lab. He has been active in the open-source community in the past 4 years, currently working on TensorFlow and Deep Learning. He has taught TensorFlow programming classes in San Francisco and Seattle. He was a core contributor in OpenStack, focusing on the networking and storage support for container orchestrator such as Kubernetes. Previously he was with the IBM Research Lab and has published papers on a wide range of subjects.
Developers building model for Deep Learning now have some 15 major/minor frameworks to choose from. Such a wide set of choices reflects the level of interest in Deep Learning across academia and enterprises, but choosing an appropriate framework can be confusing given the trade-off among them. In this talk, we will take a close look at the capabilities of TensorFlow, the open source framework from Google that currently has the highest rate of adoption. We will cover the extensive language API's, high level API's, scaling the training across CPU and GPU with distributed mode, and the recently announced eager execution mode. We will also demonstrate the tools for debugging, visualization and model serving.