Red Chainer and Cumo: Practical Deep Learning in Ruby
Naotoshi Seo and Yusaku Hatanaka • April 18, 2019 • Fukuoka, Japan

Naotoshi Seo @sonots, Yusaku Hatanaka @hatappi

We will introduce Red Chainer which is a deep learning framework purely written in Ruby and Cumo which is GPU aware fast numerical library for Ruby, and talk about their evolutions.

In this talk, we describe:

Current status about Deep Learning and Scientific Computing in Ruby.
Introduction about Red Chainer project.
Introduction of the ONNX data format, and automatic generation of Ruby codes by supporting ONNX with Red Chainer.
Introduction about Cumo project, and its support in Red Chainer.
Recent updates about supporting Fast Convolutional Neural Networks in Red Chainer.
Also, @sonots talks about ChainerX and a plan to make a further faster deep learning with its ruby bindings.

RubyKaigi 2019 https://rubykaigi.org/2019/presentations/sonots.html#apr20

RubyKaigi 2019