The code here shows idiomatic ways to write TensorFlow 2.0 code. It doubles as an integration test.
Code should minimize dependencies and be self-contained in one file. A user should be able to copy-paste the example code into their project and have it just work.
Code should emphasize simplicity over performance, as long as it performs within a factor of 2-3x of the optimized implementation.
Code should work on CPU and single GPU.
Code should run in Python 3.
Code should conform to the Google Python Style Guide
Code should follow these guidelines:
for
, while
, and if
statements.