cottand / ivann

Visual tool for building neural networks with PyTorch, and generating code in Python in order to train and test them.

Ivann

Ivann is a visual tool for building neural networks with PyTorch, and generating code in Python in order to train and test them.

It is aimed at researchers who wish to see the bigger picture by visualising the flow behind their models. You can build through the UI by hand (Ivann aims to have the same feature set as PyTorch) and then modify the generated Python code, or build with a mixture of handwritten and generated code.

What can I make with ivann?

Ivann is engineered to build neural networks. You will find we put extra effort in building your typical CNN.

But we also put a lot of effort in basically supporting all of PyTorch, and we allow for 'inlining' python code in an otherwise fully UI-oriented developing experience. So it is up to you: you can either stick to using PyTorch's layers, or make your own basic building blocks and use them as bricks in your model. We have found that a match and mix of both works best.

Building ivann locally

Setup dependencies

npm install

Produce production artifacts

npm run build

Test

npm run test:unit