Mnist Neural network visualization in Odin and Raylib. All cubes now have an opacity based on their activation/weight. I think the codebase is almost ready to be made public

This is awesome - how much work would be needed to make this library compatible with any torch compatible neural net?

I'd say for small models it should be simple, but if you have millions of nodes the viz wouldn't be able to handle it, it would require a lot more optimizations

these sorts of tools would’ve been great in my ml classes, consider continuing building educational visualizations like this / collaborating with professors if that interests you. you’re a great dev bones

Thanks :) One of my older projects was discovered by a professor somehow, and they’re actually using it. But I’m still not sure how to make my work more visible to the academic community though

absolutely love this, great stuff mr bones

Startup idea, "ViziGrad"... dynamic model visualization (or modeling in general) as a service... MaaS. Go to the landing page, upload your weights, view the mapping... like an MRI machine, but for NNets.

@yacineMTB Wow, what a beauty!

I need a total hour tally for the time you spent on this

i don't have an exact hour count, i've been working on this in my free hours for about a month now

impressive, as always