Following the warm-up and repeated in March this year announced TensorFlow 2.0 Alpha version (alpha) on the TensorFlow Developer Summit, TensorFlow 2.0 Beta version (public beta) finally released today.
When TensorFlow 2.0 Alpha release, detailing the TF2.0 version of the plan default Keras, use the eager execution, cross-platform support by default, more friendly to researchers, better charts (graph) computing support, simplified API, etc. major update.
Since TF 2.0 Alpha release, Google yourself and try the Alpha version of the users on the improvements made in this version given high praise, the whole ecological TensorFlow also continued to expand. Google also released in Alpha version is also on the line deeplearning.ai and distinctions of Xuecheng introductory courses for TF 2.0 Alpha currently has more than 130,000 applications learning; TF 2.0 Alpha's Github project also has received near to 10 thirty thousand star, more than seventy thousand fork.
In today released Beta version brings the following updates:
Complete the symbol name TF2.0 API updates and deletions. This means that the entire version of the API is finalized version. Meanwhile, the API also with TF 1.14 official version together as a module compatible release of version 2.0. (Where you can see all the updated list of symbols)
Keras 2.0 supports more functions, including sub-classification model to simplify the custom training cycle API, increasing the hardware can support most types of distributed computing strategy, and so on.
Alpha version of the user who submitted many questions on Github, Google currently has repaired more than 100 issues. This work will continue to be, and continue to gather more user feedback.
TensorFlow 2.0 Beta already can be quickly installed by pip, only you need to do this line of code
> pip install tensorflow==2.0.0-beta0
Currently, TensorFlow family of products has been partially supported TF2.0 Beta, including TensorBoard, TensorHub, TensorFlow Lite, TensorFlow.js. Support TensorFlow Extended (TFX) and the end of the calculation process is still in development.
After TensorFlow 2.0 Beta release, before TensorFlow 2.0 official release, it also need to go through the development of RC (release candidate) phase, Google target at this stage is to increase the Google cloud TPU and TPU Cluster (TPU Pods) of Keras model support to further improve operating performance, as well as amendments more problems. RC version released some time in the summer, it is not far off.
END
Bi-mao wonderful classroom courses recommended:
1.Cloudera data analysis courses;
2.Spark Hadoop development and training;
3. Big Data machine learning recommendation systems;
4.Python data analysis and machine learning practical;
For more details, please look at our public number: Bi Mao data - Course Products - Bi-mao classroom
Register now interactive learning was massive currency, a large number of quality courses free delivery!