Metadata-Version: 2.1
Name: torch-snake
Version: 1.0.0
Summary: Sample PyTorch implementation of the snake activation function
Home-page: https://github.com/EdwardDixon/snake
Author: Edward Dixon
Author-email: dixon.edward@gmail.com
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.6
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: torch

[![PyPI version](https://badge.fury.io/py/torch-snake.svg)](https://badge.fury.io/py/torch-snake)

# Snake
Based on ["Neural Networks Fail to Learn Periodic Functions and How to Fix It"](https://arxiv.org/abs/2006.08195) by Liu Ziyin, Tilman Hartwig, Masahito Ueda

This is a PyTorch implementation of the `snake` activation function from the paper - or at least I _think_ it is, no affiliation with the authors, use at your own risk, etc., etc.

A few variations of the function are discussed in the paper, this package implements:

 <img src="https://render.githubusercontent.com/render/math?math=x%2B\frac{1}{a}sin^{2}(ax)">
 
Snake is **periodic**, but _also_ monotonic. To see how **snake** behaves for a range of x given various choices of `a`, watch this video:

[![snake activation function gets wriggly for higher a](http://img.youtube.com/vi/5g7-8mtywuk/0.jpg)](http://www.youtube.com/watch?v=5g7-8mtywuk "Snake Activation")

## Installation
Two methods:
- Using **pip**, `pip install torch-snake`
- To install from source, first clone this repository.  Then, from the main repo folder, run `python setup.py install`

## Usage
Fairly easy really `from snake.activations import Snake`.  The `Snake` constructor [(code here)](snake/activations.py) has an optional **learnable** parameter alpha which defaults to 1.  The authors of the paper find values between 5 and 50 work quite well for "known-periodic" data, while for better results with non-periodic data, you should choose a small value like 0.2.  The constructor also takes an `alpha_learnable` parameter which defaults to `True`, so that you can disable "learnability" if your experiments so require.  

## Sample code
There's a notebook, still quite rough - [example.ipynb](example.ipynb).  Early indications are that good choices of hyperparameters are quite important for best results (though snake's own parameter trains quite readily).


## Acknowledgements
This code probably wouldn't have gotten written if it hadn't been for [Alexandra Deis](https://towardsdatascience.com/@astakhova.aleksandra) and [her excellent article](https://towardsdatascience.com/extending-pytorch-with-custom-activation-functions-2d8b065ef2fa) .
