Add AdaBelief optimizer

[cryu854]

Description

Fixes #2203 by adding AdaBelief optimizer. Project page, arXiv, GitHub

Although @juntang-zhuang has an official implementation at juntang-zhuang/Adabelief-Optimizer, but it was heavily based on tensorflow_addons/optimizers/rectified_adam.py, and we only modified the core algorithm part.

So we decided to contribute this optimizer to leverage the friendly API from addons, with the hope that it can help anyone who is struggling for training a neural network.

Type of change

• Bug fix
• New Tutorial
• Updated or additional documentation
• Additional Testing
• New Activation and the changes conform to the activation contribution guidelines
• New Callback and the changes conform to the callback contribution guidelines
• New Image addition and the changes conform to the image op contribution guidelines
• New Layer and the changes conform to the layer contribution guidelines
• New Loss and the changes conform to the loss contribution guidelines
• New Metric and the changes conform to the metric contribution guidelines
• New Optimizer and the changes conform to the optimizer contribution guidelines
• New RNN Cell and the changes conform to the rnn contribution guidelines
• New Seq2seq addition and the changes conform to the seq2seq contribution guidelines
• New Text addition and the changes conform to the text op contribution guidelines

Checklist:

• I've properly formatted my code according to the guidelines
  • By running Black + Flake8
  • By running pre-commit hooks
• This PR addresses an already submitted Add AdaBelief optimizer [https://arxiv.org/abs/2010.07468] #2203 issue for TensorFlow Addons
• I have made corresponding changes to the documentation
• I have added tests that prove my fix is effective or that my feature works
• This PR contains modifications to C++ custom-ops

How Has This Been Tested?

Unit tests for AdaBelief, the results are compared with numpy update using the same algorithm.

[bhack]

I think It is a little bit early for this. See #2194

[cryu854]

Thank you for your feedback. I think AdaBelief would be a good choice for training because it uses curvature information like an ideal optimizer, which makes it more intuitively reasonable. Once addons decides to add this feature, I will help maintain it.

[bhack]

Is that currently we have just a single citation #2203 (comment)

[dathudeptrai]

@bhack @cryu854 will it merge ?

[cryu854]

@dathudeptrai I think it depends on the usage rate by community and the number of citations. Currently, please use juntang-zhuang/Adabelief-Optimizer.

[denadai2]

isn't this done?

[bhack]

Yes