Build an ML pipeline for BERT models with TensorFlow Extended – An end-to-end Tutorial


During the workshop, the audience will gain not only a holistic overview of the TensorFlow ecosystem but will also learn the necessary steps to bring ML projects from experiments to production. With the knowledge, the participants can translate their ML projects into TFX pipelines and simplify their ML model production processes.

Session Outline
In this tutorial, Hannes will explain how to take an experimental BERT model from a Jupyter notebook to production using TensorFlow Extended. During the 90 min workshop, Hannes will guide the audience through the steps required to take a TensorFlow 2.x model and build an ML model pipeline for it. The end-to-end implementation highlights:
* Fine-tuning of a pre-trained BERT model from TensorFlow Hub
* Implementing the BERT preprocessing steps with TensorFlow Transform
* Using TensorFlow Text to manipulate strings with TensorFlow Ops
* Training and analyzing a custom sentiment model using the pre-trained BERT model
* Deploying the trained model with TensorFlow Serving


Hannes Hapke is a senior machine learning for Concur Labs at SAP Concur, where he explores innovative ways to use machine learning to improve the experience of a business traveler. Prior to joining SAP Concur, Hannes solved machine learning and ML infrastructure problems in various industries including healthcare, retail, recruiting, and renewable energies. He was recognized as a Google Developer Expert for ML and has co-authored two machine learning publications: "Building Machine Learning Pipeline" by O'Reilly Media and "NLP in Action" by Manning Publications.