godl

• 

  VGG16 example detecting a mug

• 

  VGG16 example detecting a zebra

Inspiration

this project is mainly inspired by pytorch

What it does

Easy to use deep learning framework fully written in Go

How we built it

At its core it uses gorgonia, a project to which I also contributed a lot of code to make this possible

Challenges we ran into

Probably the main challenge is that gorgonia lacked operations to make this possible. I enjoyed contributing to it as well

Accomplishments that we're proud of

I'm proud that it includes the first full pure-go implementation for VGG16 (https://arxiv.org/abs/1409.1556) and TabNet (https://arxiv.org/abs/1908.07442) VGGFace2 is also in progress

What we learned

I learned a lot about low level Machine Learning, implementing operations and I also learned that it's possible to have a fully featured deep learning framework for Go

What's next for godl

I like to complete VGGFace2 and implement a new network based on transformers, probably EfficientNet It also need to support a standard format like ONNX Full roadmap is in the README

Built With

• go
• gorgonia