Bishõgrad

A Bishõ Autograd engine in python along with a lightweight deep neural-network library! (inspired from mircrograd by Andrej Karpathy)

what's Bishõ?

微小 - Bishõ is Japanese word for 'tiny' since my implementation is very tinyyyyy ^_^ compared to PyTorch/Tensorflow

what's Hako?

箱 - Hako is Japanese word for 'box' , here Hako signify the neurons in our network ;D

Installation

• Install via pip:

  pip install bishograd==0.1.0
  

Example

Wonder how this works? Checkout the examples.

Targets :

• ReLU Activation function
• Add MLP.training() to automate the whole training code
• Add Sigmoid, LeakyReLU & other activation functions
• Add loss functions - categorical loss, mean-square loss etc

Contribution :

This project is open for contribution!

• Clone this repository:

git clone https://github.com/AK3847/Bishograd.git

Primary way to contribute is to either raise an issue or a pull request with proper description and code format. You can contribute to any Targets or suggest new features as well.

Drop a star if this project helped you in anyway! ( ﾉ ﾟｰﾟ)ﾉ