👋 Hey Product Hunt, I'm Krrish, founder of Verace. We're a small AI lab based in Jaipur, India, and Verace V1 is the first thing we're putting out into the open. 🚀 Verace V1 replaces attention, the KV cache, mixture-of-experts, and fixed-depth compute with continuous, manifold-constrained alternatives, each one backed by an executable test, not just a derivation on paper. ✅ It's a reference implementation, not a trained model. No pretrained weights, no benchmark claims. Just the architecture, the tests, and an honest write-up of what worked and what didn't . 🔍 📄 Full paper: https://verace.in/research/verac... 💻 Code: https://github.com/Verace-Pvt-Lt... Would love feedback 🙏, especially from anyone who's worked on manifold-constrained optimization or linear attention variants.

Verace V1

A next generation AI architecture, open source

A next generation AI architecture, open source