Practical AI avatar

Dexa/Practical AI

Learn more
Published May 7, 2019

TensorFlow Dev Summit 2019

Dive into the exciting developments at TensorFlow Dev Summit 2019, featuring insights on TensorFlow Federated for secure machine learning, advances in AI deployment with TensorFlow Lite and TensorFlow.js, and user-friendly updates in TensorFlow 2.0 to streamline deep learning. Plus, learn about the upcoming TensorFlow World Conference, promising to foster innovation and community engagement across the ecosystem.
Episode Highlights
Practical AI logo

Topics covered

Popular Clips

Episode Highlights

  • Ease of Use

    The introduction of TensorFlow 2.0 marks a significant shift towards ease of use, primarily through eager execution and simplified graph management. shares his struggles with the verbosity and complexity of TensorFlow 1.x, highlighting how eager execution now allows for immediate command execution, making debugging and interactive development more intuitive 1. adds that Keras, integrated into TensorFlow, offers a more pythonic approach, aligning with familiar programming paradigms and easing the logical burden for developers 2. This shift is expected to attract developers who previously found TensorFlow cumbersome:

    It's a more intuitive interface, easier debugging, and a more natural control flow.

    ---

    The combination of Keras and eager execution positions TensorFlow 2.0 as a competitive alternative to other frameworks like PyTorch 3.

       

    Keras Integration

    Keras has become the primary interface for TensorFlow 2.0, making deep learning more accessible and user-friendly. explains that Keras serves as a high-level API, simplifying TensorFlow's complex internals and allowing developers to focus on building models rather than managing intricate details 4. This integration supports both object-oriented and functional programming styles, catering to a wide range of developers 5. notes the abundance of resources available for those eager to explore TensorFlow 2.0, including courses from Deep Learning AI and Udacity, which were highlighted at the Dev Summit 6.

    Keras is kind of giving you a very friendly wrapper... a lot of the complexity that you were subjugated to in TensorFlow 1.0 has been taken away.

    ---

    This strategic move by TensorFlow aims to streamline the development process and broaden its appeal among developers.

Related Episodes