I used I-JEPA to generate SVG's and here is my code!

Imagine trying to turn complex images into SVGs with AI — sounds simple, but it’s tricky. Haghiri75, a Reddit user, took Yann LeCunn’s JEPA idea and decided to see if it could generate vector graphics. Instead of relying on traditional models like Stable Diffusion or VTracer, he jumped into I-JEPA, a newer approach that’s all about understanding images more deeply. Now, here’s where it gets interesting — he managed to produce SVGs, but they weren’t perfect yet. Still, he shared his code, licensed under MIT, hoping others can improve it. According to Haghiri75, the available JEPA weights are CC licensed, which makes experimentation easier. Plus, he’s curious — could JEPA not just reshape image generation but also revolutionize text or other AI tasks by being more energy-efficient? This work is a step toward smarter, cheaper AI. And honestly, the potential is exciting — especially if it can make complex visual tasks faster and more accessible, as Haghiri75 points out in his Reddit post.