Micrograd

A tiny autograd engine. This is a Ruby implementation of karpathy/micrograd. I created this while working through Andrej Karpathy's Neural Networks: Zero To Hero course.

Installation

gem install micrograd

Usage

Here are some of the operatinos available on Value.

include Micrograd

a = Value.new(2.0)
b = Value.new(-3.0)
c = Value.new(10.0)
e = a * b
d = e + c
f = Value.new(-2.0)

l = d * f

# Walk through all the values and calculate gradients for them.
l.start_backward

Example of training a multi level perceptron (MLP):

mlp = MLP.new(input_size: 3, layer_sizes: [4, 4, 1])

# These are the training inputs
inputs = [
  [2.0, 3.0, -1.0],
  [3.0, -1.0, 0.5],
  [0.5, 1.0, 1.0],
  [1.0, 1.0, -1.0],
]

# These are the outputs for each of the inputs above.
desired_outputs = [1.0, -1.0, -1.0, 1.0]

# Training loop
100.times do |n|
  # forward pass
  mlp_outputs = inputs.map { mlp.call(_1).first }
  loss = desired_outputs.zip(mlp_outputs).sum { (_1 - _2) ** 2 }

  # backward pass
  mlp.parameters.each { _1.grad = 0.0 }
  loss.start_backward

  # update the params
  mlp.parameters.each { _1.data -= _1.grad * 0.1 }
end

Development

After checking out the repo, run bin/setup to install dependencies. Then, run rake test to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and the created tag, and push the .gem file to rubygems.org.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/nithinbekal/micrograd. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the code of conduct.

License

The gem is available as open source under the terms of the MIT License.

Code of Conduct

Everyone interacting in the Micrograd project's codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.

nithinbekal / micrograd