Practical AI TensorFlow Dev Summit 2019
May 7, 2019
The discussion kicks off with thrilling announcements from the TensorFlow Dev Summit 2019, including the alpha release of TensorFlow 2.0. The integration of Keras simplifies machine learning development with eager execution. They also highlight the importance of accessible datasets and innovations like TensorFlow Federated and TFX for decentralized data processing. Rapid prototyping tools are showcased to encourage experimentation, while insights into remote work emphasize the need for effective communication and empathy in tech teams.
AI Snips
Chapters
Transcript
Episode notes
TensorFlow 2.0 and Keras
- Use Keras as the primary interface for TensorFlow 2.0.
- It simplifies the user experience and offers eager execution by default.
Eager Execution Improves TensorFlow 2.0
- TensorFlow 1.x's graph execution made debugging and model optimization challenging.
- Eager execution in TensorFlow 2.0 allows immediate command execution, improving interactive development.
TensorFlow Datasets Simplifies Data Handling
- TensorFlow Datasets simplifies data import and preparation.
- It provides easy access to many popular research datasets like MNIST and others, saving time.