WhyTorch - a visual explainer for PyTorch functions

I just built WhyTorch, an open source website I built to explain PyTorch functions. It makes tricky functions like torch.gather and torch.scatter more intuitive by showing element-level relationships between inputs and outputs.

I built this site as a companion to the PyTorch docs, and you can jump straight to WhyTorch by just modifying the URL of a PyTorch doc page.

For each function, you can click elements in the result to see where they came from, or elements in the inputs to see how they contribute to the result to see exactly how it contributes to the result. Visually tracing tensor operations can help clarify how indexing, slicing, and broadcasting work.

Would love feedback on which PyTorch functions people most want visualized next.

1 Like

Would you be so kind posting a link to your website so we could check it out?

That would have been a helpful thing to include!

And here’s a screenshot.

2 Likes

Here’s a direct link to the source: GitHub - Kukanani/whytorch: A visual explainer for PyTorch tensor operations

I also wrote up some of what I learned building it in a blog post: What I learned building WhyTorch | Adam Allevato