TensorFlow 2.0.0-beta0 released. 2.0 focused on simplicity and ease of use, with the following update:
- Easy to use to perform modeling and eager Keras
- Be robust model deployed in production on any platform
- Strong research experiments
- API to simplify and reduce duplication by deleting deprecated endpoints
This version brings some improvements and bug fixes, highlights include:
- Distributed Strategy: TF 2.0 users will be able to use the new tf.distribute.Strategy API with minimal code changes for distributed training, resulting in a good performance out of the box. Beta support more policies, as well as custom training to improve circulation and Keras subclasses model support.
- API freeze: Symbol rename / deprecated and 2.0 API changes have been completed. 2.0 API is the final, also be provided as part of the module TensorFlow 1.14 compat.v2 version.
Details View Update:
https://github.com/tensorflow/tensorflow/releases/tag/v2.0.0-beta0